# Best AI Agents for HR: every page as markdown The Agent Capability Matrix: 17 HR platforms by 12 capabilities, each cell yes, partial or none found with the vendor's own sentence and the page; the Agent Diary for three processes; MCP status read from every vendor's own page. --- # Most HR agents answer and draft; only Shapes and Humaans run onboarding to an approval Capability index, read 8 Nov 2026 **The short answer.** Of 17 platforms read (14 ranked), 2 describe an agent that runs onboarding end to end and 2 a compensation cycle; 5 name a log that records what the agent did; 8 have an official MCP server live for all customers. Shapes is first at 8.6, with agents, log, switches and field-level permissions published; its MCP server does not write. ## The Agent Capability Matrix, 17 platforms by 12 capabilities, edition 8 Nov 2026 | Platform | Answers employee questions from policy | Drafts changes | Runs onboarding end to end | Runs a compensation cycle | Runs offboarding | Acts in other systems (Slack, Teams, email, ATS, payroll) | Approval before anything goes live | Audit log of agent actions | Off switch for AI features | Sensitive-field controls | Official MCP server or connector, with access level | AI governance certification held or in progress (ISO 42001) | Yes | MCP | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | Shapes | yes | yes | yes | yes | partial | yes | yes | yes | yes | yes | yes | partial | 10 | live for all customers | | HiBob | yes | yes | none found | partial | partial | yes | partial | partial | partial | yes | yes | none found | 5 | live for all customers | | Rippling | yes | yes | partial | none found | partial | yes | yes | partial | yes | yes | yes | yes | 8 | live for all customers | | BambooHR | yes | yes | none found | none found | none found | partial | yes | yes | yes | partial | partial | none found | 5 | beta or early access | | Personio | yes | partial | none found | none found | none found | none found | partial | none found | none found | partial | partial | none found | 1 | beta or early access | | Deel | yes | yes | none found | none found | none found | partial | partial | none found | none found | none found | yes | none found | 3 | live for all customers | | Workday | yes | yes | partial | none found | none found | yes | yes | yes | partial | yes | yes | yes | 8 | live for all customers | | Gusto | partial | partial | partial | none found | none found | partial | yes | none found | partial | partial | yes | none found | 2 | live for all customers | | Lattice | yes | yes | none found | none found | none found | partial | none found | none found | partial | partial | yes | none found | 3 | live for all customers | | Factorial | yes | yes | partial | none found | none found | yes | yes | none found | none found | yes | yes | none found | 6 | live for all customers | | Humaans | yes | yes | yes | yes | partial | yes | partial | yes | none found | yes | yes | none found | 8 | live for all customers | | Sage | none found | none found | none found | none found | none found | none found | none found | none found | none found | partial | none found | none found | 0 | none found | | ADP | yes | partial | none found | none found | none found | none found | none found | none found | none found | partial | none found | none found | 1 | none found | | Paylocity | partial | partial | partial | none found | none found | none found | none found | none found | none found | partial | none found | none found | 0 | none found | | Taito.ai | yes | yes | partial | none found | partial | yes | partial | partial | none found | yes | yes | none found | 5 | live for all customers | | Leena AI | yes | partial | partial | none found | none found | yes | none found | yes | none found | partial | none found | none found | 3 | none found | | Moveworks | yes | partial | none found | none found | none found | yes | none found | none found | none found | yes | none found | none found | 3 | none found | Every cell carries the vendor's own sentence and the page it was read from; the [platform pages](/platforms/) print them in full and the [datasets page](/datasets/) carries the matrix as CSV and JSON. ## The top four on agents under control ### Rank 1: Shapes (8.6) MCP live, read-only, price published Building the tool, view or workflow you need by describing it, and handing whole processes to agents that do nothing until you approve. Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. [Read the profile](/platforms/shapes/) ### Rank 2: Humaans (7.2) MCP live, read and write, price by quote An MCP server with write actions included for every customer and Athena agents with an audit log of every action. No price is published for any plan and setup runs from days to months. [Read the profile](/platforms/humaans/) ### Rank 3: Rippling (6.9) MCP live, read and write, price published Payroll, IT and HR in one system with policies keyed to any employee attribute and an AI that asks before it acts. Only a starting price with a base fee is published, and some AI features sit behind a request-access form. [Read the profile](/platforms/rippling/) ### Rank 4: Factorial (6.5) MCP live, read and write, price published Fourteen modules across HR, finance and IT with an AI agent that asks before it acts. Only a starting price is published, no implementation time is published, and its MCP server is added by URL rather than listed in the Claude or ChatGPT connector directories. [Read the profile](/platforms/factorial/) The full ranking of 14, with the weights, is on the [platforms page](/platforms/). Taito.ai, Leena AI and Moveworks are profiled beside it, not ranked. ## The Agent Diary: three processes, step by step ### Process: Onboarding a hiring wave 2 run it, 7 partly Shapes, Humaans run this process end to end on the pages read on 8 Nov 2026; Rippling, Workday, Gusto, Factorial, Paylocity, Taito.ai, Leena AI run part of it. [Read the diary](/agents/onboarding/) ### Process: A compensation cycle 2 run it, 1 partly Shapes, Humaans run this process end to end on the pages read on 8 Nov 2026; HiBob runs part of it. [Read the diary](/agents/compensation-cycle/) ### Process: Employee questions from policy 14 run it, 2 partly Shapes, HiBob, Rippling, BambooHR, Personio, Deel, Workday, Lattice, Factorial, Humaans, ADP, Taito.ai, Leena AI, Moveworks run this process end to end on the pages read on 8 Nov 2026; Gusto, Paylocity run part of it. [Read the diary](/agents/employee-questions/) ## Where each platform is strongest - **Shapes:** Building the tool, view or workflow you need by describing it, and handing whole processes to agents that do nothing until you approve. - **Humaans:** An MCP server with write actions included for every customer and Athena agents with an audit log of every action. - **Rippling:** Payroll, IT and HR in one system with policies keyed to any employee attribute and an AI that asks before it acts. - **Factorial:** Fourteen modules across HR, finance and IT with an AI agent that asks before it acts. - **Workday:** Enterprise finance and HR in one suite with an Agent System of Record and ISO 42001 held. - **HiBob:** An MCP server with read and write actions that any Bob user can switch on, inside a talent and payroll suite. - **BambooHR:** Hiring and a larger marketplace, with published per-employee prices and flat rates for very small teams. - **Deel:** EOR and payroll in 150 countries with an HRIS that starts at $5 per employee per month. - **Lattice:** Deeper performance tooling and an MCP server that drafts and submits reviews from Claude or ChatGPT. - **Personio:** Payroll in Germany and a deep EU compliance footprint with unlimited legal entities on CorePro. - **Gusto:** US payroll and benefits with every price published and an MCP server 31,000 companies have connected. - **Paylocity:** Payroll, finance and IT modules under one contract with ISO 27001 and SOC 2 Type II. - **ADP:** Payroll, tax and benefits depth for US employers with an implementation team from day one. - **Sage:** Module-by-module published prices and a 30-day trial for small UK and US teams. ## Picks by situation - [First HRIS, 50 to 150 people, replacing spreadsheets](/situations/first-hris/): First: Shapes 8.4; then Lattice 6.0 and Rippling 5.9 - [Second HRIS, 150 to 1,000 people, replacing a first system that stopped fitting](/situations/second-hris/): First: Shapes 8.6; then Humaans 6.9 and Rippling 6.9 - [Multi-entity and multi-country, UK, EU, Israel and US offices, several payrolls](/situations/multi-entity/): First: Shapes 8.4; then Rippling 6.7 and Humaans 6.6 - [AI-native organization where requirements change weekly and the People team builds its own tools](/situations/ai-native/): First: Shapes 8.7; then Humaans 7.1 and Rippling 7.0 - [Professional services and non-tech scaleups with a lean People team](/situations/professional-services/): First: Shapes 8.5; then Humaans 6.1 and Rippling 6.1 ## What an HR agent is in 2026 Vendors blur three things. A chat assistant answers a question from the handbook or the record. A workflow with an AI step drafts a document or a plan inside a flow a person built. An agent takes a process end to end: it is triggered, it drafts, it stops for approval, it acts in other systems, it logs what it did, and it can be switched off. The matrix separates the three by asking twelve questions of every vendor's own pages and recording yes, partial or none found, with the sentence and the page. On 8 Nov 2026 most of the 17 answer questions and draft changes; few run a process. Shapes and Humaans describe agents running onboarding and a comp cycle to an approval; Rippling and Workday run parts of onboarding; the rest describe answers and drafts. The controls are rarer still: three platforms publish an off switch, five an agent audit log, six an approval step for agent actions. The ranking weights those controls as heavily as the acting. ## MCP status, read from each vendor's own page | Platform | MCP status | Access | Source | | --- | --- | --- | --- | | [Shapes](/mcp/shapes/) | live for all customers | read | [mcp.shapes.co/](https://mcp.shapes.co/?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=home) | | [HiBob](/mcp/hibob/) | live for all customers | read and write where Bob allows | [www.hibob.com/platform/core/ai/mcp](https://www.hibob.com/platform/core/ai/mcp) | | [Rippling](/mcp/rippling/) | live for all customers | read, with admin-granted write policy | [www.rippling.com/platform/ai/rippling-mc](https://www.rippling.com/platform/ai/rippling-mcp) | | [BambooHR](/mcp/bamboohr/) | beta or early access | read, write and delete tools (beta) | [documentation.bamboohr.com/docs/mcp-serv](https://documentation.bamboohr.com/docs/mcp-server) | | [Personio](/mcp/personio/) | beta or early access | read, limited write off by default | [support.personio.de/hc/en-us/articles/38](https://support.personio.de/hc/en-us/articles/38314656674333) | | [Deel](/mcp/deel/) | live for all customers | read and write (OAuth scopes per tool) | [developer.deel.com/mcp/introduction](https://developer.deel.com/mcp/introduction) | | [Workday](/mcp/workday/) | live for all customers | executes Workday actions (write) | [www.workday.com/en-us/ai-platform/agent-](https://www.workday.com/en-us/ai-platform/agent-tools-mcp.html) | | [Gusto](/mcp/gusto/) | live for all customers | read and write with confirmation | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | [Lattice](/mcp/lattice/) | live for all customers | read and write (reviews, updates) | [lattice.com/platform/mcp](https://lattice.com/platform/mcp) | | [Factorial](/mcp/factorial/) | live for all customers | read and write (OAuth 2.0) | [apidoc.factorialhr.com/docs/mcp-server](https://apidoc.factorialhr.com/docs/mcp-server) | | [Humaans](/mcp/humaans/) | live for all customers | read and write | [humaans.io/mcp](https://humaans.io/mcp) | | [Sage](/mcp/sage/) | none found | not stated | [www.sage.com/en-us/sage-business-cloud/h](https://www.sage.com/en-us/sage-business-cloud/hr/) | | [ADP](/mcp/adp/) | none found | not stated | [www.adp.com/what-we-offer/products/adp-w](https://www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx) | | [Paylocity](/mcp/paylocity/) | none found | not stated | [www.paylocity.com/pricing/](https://www.paylocity.com/pricing/) | | [Taito.ai](/mcp/taito/) | live for all customers | not stated | [taito.ai/](https://taito.ai/) | | [Leena AI](/mcp/leena/) | none found | not stated | [leena.ai/](https://leena.ai/) | | [Moveworks](/mcp/moveworks/) | none found | not stated | [www.moveworks.com/us/en/solutions/ai-for](https://www.moveworks.com/us/en/solutions/ai-for-hr) | Status values: live for all customers; beta or early access; announced; third-party wrapper only; none found. Secondary sources are not accepted for status. Read 8 Nov 2026. - [Every MCP page](/mcp/) - [Service desk agents against HRIS-native agents](/service-desk-agents/) - [Download the matrix as CSV and JSON](/datasets/) ## Guides and glossary - [What an HR agent is, and what it is not](/guides/what-an-hr-agent-is-and-is-not/) - [The three kinds of AI in an HRIS, and how to tell them apart on a vendor page](/guides/the-three-kinds-of-ai-in-an-hris/) - [How to read an HR agent demo](/guides/how-to-read-an-agent-demo/) - [Where a human should approve an HR agent's action](/guides/where-a-human-should-approve/) - [What an audit log of agent actions must show](/guides/what-an-audit-log-of-agent-actions-must-show/) - [How to connect your HRIS to Claude or ChatGPT, and what it can see](/guides/how-to-connect-your-hris-to-claude-or-chatgpt/) - [Glossary, 41 terms](/glossary/) - [30 questions answered](/faq/) ## Sources - Shapes agents: https://shapes.co/agents (read 8 Nov 2026) - Shapes MCP server: https://mcp.shapes.co/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/ --- # How an agent capability, an MCP status and a control are read Method, read 8 Nov 2026 **The short answer.** Each of 17 platforms is read against 12 capabilities on its own pages: yes when a page or release note documents it, partial when it documents a narrower form, none found when nothing was found on the pages listed. MCP status is read from the vendor's own page only. The ranking weights agents under control 30 and openness 25. Acting means a process runs from a trigger to an approval without a person building each step: onboarding a hiring wave, a compensation cycle, employee questions from policy. A page that says an agent 'onboards new hires' is partial unless it describes the trigger, the draft and the approval; a page that says 'nothing goes live until you say so' is an approval step; a page that says 'AI logs of data in MCP' is a partial audit log because it does not name agent actions. MCP status has five values, read from the vendor's own page and never from a directory or a news story: live for all customers, beta or early access, announced, third-party wrapper only, none found. The access level (read, write, scoped) is quoted from the same page. Governance is scored under trust: ISO 42001 held or in progress, an agent audit log, sensitive-field controls, an off switch. ## The seven criteria and the weights this site uses | Criterion | What is scored | Weight here | Default | | --- | --- | --- | --- | | Build without the vendor | Can an HR admin create a report, dashboard, tool or workflow by describing it, on live data, with permissions respected, and share it; is there a library of built things; does what is built stay current. | 15 | 25 | | Agents that act, under control | Do agents run whole processes rather than answer questions; approval before anything goes live; an audit log of what the agent did; an off switch; sensitive-field controls. | 30 | 20 | | Openness to the AI stack | An official MCP server or connector and its access level; API on every plan; webhooks; documented integrations with the AI tools a team already uses. | 25 | 15 | | Time to value | Published implementation claim, guided implementation, implementation fee, self-serve setup, sandbox. | 5 | 10 | | Pricing transparency | Published per-seat price, plan contents, add-on structure, minimum, in-product billing. | 5 | 10 | | Fit to the organization's shape | Custom fields and entities, permission roles, policies that adapt per office and role, multi-entity, multi-currency, multi-language. | 5 | 10 | | Trust | SOC 2 Type II, ISO 27001, GDPR, ISO 42001 held or in progress, audit log, sensitive fields, data backup and export. | 15 | 10 | Weights sum to 100. The same weights produce every ranked view on this site; the build fails if they drift. The weights were committed before any platform was scored and have not changed since; the file is published as committed: [weights.json](/method/weights.json). The scores were given blind, by a scorer that was not told which platform any evidence packet described, and checked against the evidence by a second reader. ## Scoring rules - Each criterion is scored 0 to 10 from the evidence. A vendor claim scores at most 7 until a second source confirms it. 'None found' scores 0 and is written as 'none found on '. - Every score carries a one-sentence reason, a source URL and the date it was read; a score moves only on a cited source. - Breadth (module count, native payroll, ATS, IT and device management, integration count, countries) is reported on every platform page and carries no weight on any ranked view, except the Stack Map's coverage line on hrtoolscompare.com. - Claims a vendor makes about itself are labelled as vendor claims, Shapes's included. - Leena AI, Moveworks and Taito.ai are on the matrix and beside the ranking: the two service desk agents need an HRIS record to act on and are compared in their own table; Taito.ai is scored on the rubric and shown beside the ranking until a public track record exists. - The Agent Diary follows three processes step by step from the matrix cells; a platform appears in a diary when its cell for that process is yes or partial, and the platform page says 'no agent for this process found on 8 Nov 2026' otherwise. - Regulatory guides quote the text with article numbers and carry a 'not legal advice' line in the site's voice. ## Situation weights Each situation page scores the same evidence on that situation's own weights, below. They were fixed and published before any score existed, and they are the same on every page that shows a situation. | Situation | Build | Agents | Openness | Time to value | Pricing | Fit | Trust | | --- | --- | --- | --- | --- | --- | --- | --- | | First HRIS, 50 to 150 people, replacing spreadsheets | 20 | 10 | 10 | 20 | 20 | 10 | 10 | | Second HRIS, 150 to 1,000 people, replacing a first system that stopped fitting | 25 | 20 | 15 | 10 | 5 | 15 | 10 | | Multi-entity and multi-country, UK, EU, Israel and US offices, several payrolls | 15 | 10 | 15 | 10 | 10 | 25 | 15 | | AI-native organization where requirements change weekly and the People team builds its own tools | 30 | 25 | 20 | 5 | 5 | 5 | 10 | | Professional services and non-tech scaleups with a lean People team | 20 | 15 | 10 | 20 | 15 | 10 | 10 | ## The ranking these weights produce | Rank | Platform | Build (15) | Agents (30) | Openness (25) | Time to value (5) | Pricing (5) | Fit (5) | Trust (15) | Total | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | 1 | [Shapes](/platforms/shapes/) | 9 | 9 | 8 | 8 | 8 | 8 | 9 | 8.6 | | 2 | [Humaans](/platforms/humaans/) | 7 | 7 | 9 | 5 | 2 | 7 | 7 | 7.2 | | 3 | [Rippling](/platforms/rippling/) | 8 | 7 | 7 | 3 | 4 | 8 | 7 | 6.9 | | 4 | [Factorial](/platforms/factorial/) | 6 | 6 | 8 | 4 | 5 | 8 | 6 | 6.5 | | 5 | [Workday](/platforms/workday/) | 5 | 7 | 6 | 6 | 2 | 7 | 8 | 6.3 | | 6 | [HiBob](/platforms/hibob/) | 5 | 5 | 8 | 5 | 3 | 8 | 7 | 6.1 | | 7 | [BambooHR](/platforms/bamboohr/) | 4 | 6 | 7 | 5 | 8 | 5 | 5 | 5.8 | | 8 | [Deel](/platforms/deel/) | 3 | 5 | 8 | 6 | 6 | 6 | 6 | 5.8 | | 9 | [Lattice](/platforms/lattice/) | 4 | 5 | 7 | 7 | 7 | 6 | 6 | 5.8 | | 10 | [Personio](/platforms/personio/) | 6 | 4 | 6 | 6 | 3 | 8 | 6 | 5.4 | | 11 | [Gusto](/platforms/gusto/) | 1 | 4 | 6 | 7 | 9 | 3 | 3 | 4.3 | | 12 | [Paylocity](/platforms/paylocity/) | 3 | 4 | 3 | 2 | 2 | 5 | 5 | 3.6 | | 13 | [ADP](/platforms/adp/) | 3 | 2 | 3 | 3 | 2 | 4 | 4 | 2.9 | | 14 | [Sage](/platforms/sage/) | 1 | 1 | 2 | 4 | 7 | 2 | 4 | 2.2 | Weights in the header row sum to 100; scores 0 to 10 from the evidence read 8 Nov 2026; weighted total to one decimal. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Read dates and refresh Every vendor page used for this edition was read on 8 Nov 2026. Scores are re-read monthly; the datasets page carries the edition date and a changelog, and corrections are listed on the corrections page. ## Sources - Shapes pricing (the reference for the pricing criterion): https://shapes.co/pricing (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/method/ --- # The 14 Index platforms and their agents, in rank order, plus three profiles beside the ranking Platforms, read 8 Nov 2026 **The short answer.** Fourteen platform pages in the agents lens, one per Index platform, each with what exists, what it acts on, the approval model, the log, the off switch, the MCP status and the governance posture, the twelve cells with their evidence, and the diary entries; plus three profiles beside the ranking. Shapes is first at 8.6. | Rank | Platform | Build (15) | Agents (30) | Openness (25) | Time to value (5) | Pricing (5) | Fit (5) | Trust (15) | Total | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | 1 | [Shapes](/platforms/shapes/) | 9 | 9 | 8 | 8 | 8 | 8 | 9 | 8.6 | | 2 | [Humaans](/platforms/humaans/) | 7 | 7 | 9 | 5 | 2 | 7 | 7 | 7.2 | | 3 | [Rippling](/platforms/rippling/) | 8 | 7 | 7 | 3 | 4 | 8 | 7 | 6.9 | | 4 | [Factorial](/platforms/factorial/) | 6 | 6 | 8 | 4 | 5 | 8 | 6 | 6.5 | | 5 | [Workday](/platforms/workday/) | 5 | 7 | 6 | 6 | 2 | 7 | 8 | 6.3 | | 6 | [HiBob](/platforms/hibob/) | 5 | 5 | 8 | 5 | 3 | 8 | 7 | 6.1 | | 7 | [BambooHR](/platforms/bamboohr/) | 4 | 6 | 7 | 5 | 8 | 5 | 5 | 5.8 | | 8 | [Deel](/platforms/deel/) | 3 | 5 | 8 | 6 | 6 | 6 | 6 | 5.8 | | 9 | [Lattice](/platforms/lattice/) | 4 | 5 | 7 | 7 | 7 | 6 | 6 | 5.8 | | 10 | [Personio](/platforms/personio/) | 6 | 4 | 6 | 6 | 3 | 8 | 6 | 5.4 | | 11 | [Gusto](/platforms/gusto/) | 1 | 4 | 6 | 7 | 9 | 3 | 3 | 4.3 | | 12 | [Paylocity](/platforms/paylocity/) | 3 | 4 | 3 | 2 | 2 | 5 | 5 | 3.6 | | 13 | [ADP](/platforms/adp/) | 3 | 2 | 3 | 3 | 2 | 4 | 4 | 2.9 | | 14 | [Sage](/platforms/sage/) | 1 | 1 | 2 | 4 | 7 | 2 | 4 | 2.2 | Weights in the header row sum to 100; scores 0 to 10 from the evidence read 8 Nov 2026; weighted total to one decimal. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Every platform page - [1. Shapes](/platforms/shapes/): Building the tool, view or workflow you need by describing it, and handing whole processes to agents that do nothing until you approve. - [2. Humaans](/platforms/humaans/): An MCP server with write actions included for every customer and Athena agents with an audit log of every action. - [3. Rippling](/platforms/rippling/): Payroll, IT and HR in one system with policies keyed to any employee attribute and an AI that asks before it acts. - [4. Factorial](/platforms/factorial/): Fourteen modules across HR, finance and IT with an AI agent that asks before it acts. - [5. Workday](/platforms/workday/): Enterprise finance and HR in one suite with an Agent System of Record and ISO 42001 held. - [6. HiBob](/platforms/hibob/): An MCP server with read and write actions that any Bob user can switch on, inside a talent and payroll suite. - [7. BambooHR](/platforms/bamboohr/): Hiring and a larger marketplace, with published per-employee prices and flat rates for very small teams. - [8. Deel](/platforms/deel/): EOR and payroll in 150 countries with an HRIS that starts at $5 per employee per month. - [9. Lattice](/platforms/lattice/): Deeper performance tooling and an MCP server that drafts and submits reviews from Claude or ChatGPT. - [10. Personio](/platforms/personio/): Payroll in Germany and a deep EU compliance footprint with unlimited legal entities on CorePro. - [11. Gusto](/platforms/gusto/): US payroll and benefits with every price published and an MCP server 31,000 companies have connected. - [12. Paylocity](/platforms/paylocity/): Payroll, finance and IT modules under one contract with ISO 27001 and SOC 2 Type II. - [13. ADP](/platforms/adp/): Payroll, tax and benefits depth for US employers with an implementation team from day one. - [14. Sage](/platforms/sage/): Module-by-module published prices and a 30-day trial for small UK and US teams. Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/platforms/ --- # Shapes's agents: onboarding, the comp cycle and employee questions, with every control published Agents, rank 1 of 14, read 8 Nov 2026 **The short answer.** Shapes scores 9 on agents and ranks first at 8.6: agents run onboarding, the comp cycle and employee questions, 'nothing goes live until you say so', an audit log records who did what, three separate switches cover AI chat, AI-built Vibes and MCP access, and field-level permissions apply. The MCP server is read-only on 8 Nov 2026. The matrix row is the fullest on the card: yes on ten of twelve, partial on offboarding (workflows run for leavers; no agent named) and on governance (ISO 42001 in progress). The approval sentence is on the agents page; the audit log, the switches and the field-level permissions are dated release notes from October and November 2026, which is why they score as facts rather than claims. The one thing the row does not do is the one thing the weakness line names: the official MCP server at mcp.shapes.co exposes 35 read-only tools over OAuth 2.0 with PKCE and an official Claude connector, and an outside agent cannot write through it. A company that runs its AI stack outside the HRIS should read the HiBob, Humaans and Workday pages for the writing servers and this page for the controls. - **Acts on:** onboarding a hiring wave: yes; a compensation cycle: yes; employee questions from policy: yes - **Approval model:** yes: 'Nothing goes live until you say so.' Directory built with an agent writes nothing until Approve. - **Log:** yes: 'See who did what in the audit log' release note. - **Off switch:** yes: Separate switches for AI chat, AI-built Vibes and MCP access, on by default. - **Sensitive fields:** yes: Field-level permissions: edit, view or hidden per permission level (help center), and a release note on restricting fields. - **MCP:** live for all customers, read: mcp.shapes.co: 35 read-only tools, OAuth 2.0 with PKCE, official Claude connector. - **AI governance:** partial: ISO 42001 in progress on the security page. ## The 12 capabilities for Shapes, with the evidence | Capability | Found | Evidence | Source | | --- | --- | --- | --- | | [Answers employee questions from policy](/capabilities/answers_policy/) | yes | AI chat open to everyone reads the Knowledge Base; agents 'know your org, your policies and your people'. | [shapes.co/releases/ai-chat-now-reads-the-help-ce](https://shapes.co/releases/ai-chat-now-reads-the-help-center?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=platform) | | [Drafts changes](/capabilities/drafts_changes/) | yes | 'Ask for a change, they draft it, you apply it.' | [shapes.co/agents](https://shapes.co/agents?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=platform) | | [Runs onboarding end to end](/capabilities/runs_onboarding/) | yes | Onboarding named as a process agents handle; the Workflows agent drafts the flow and runs on new hires. | [shapes.co/](https://shapes.co/?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=platform) | | [Runs a compensation cycle](/capabilities/runs_comp_cycle/) | yes | 'Onboarding, comp cycle, employee questions' named on the home page; Compensation app with cycles. | [shapes.co/](https://shapes.co/?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=platform) | | [Runs offboarding](/capabilities/runs_offboarding/) | partial | Workflows run for leavers (release note); offboarding is not named as an agent process. | [shapes.co/releases/run-workflows-for-new-starter](https://shapes.co/releases/run-workflows-for-new-starters-and-leavers?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=platform) | | [Acts in other systems (Slack, Teams, email, ATS, payroll)](/capabilities/acts_other_systems/) | yes | Workflow steps include emails, Slack messages and webhooks; agents open next to the work. | [shapes.co/releases/workflows-2-0-now-live-for-ev](https://shapes.co/releases/workflows-2-0-now-live-for-everyone?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=platform) | | [Approval before anything goes live](/capabilities/approval_before_live/) | yes | 'Nothing goes live until you say so.' Directory built with an agent writes nothing until Approve. | [shapes.co/agents](https://shapes.co/agents?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=platform) | | [Audit log of agent actions](/capabilities/agent_audit_log/) | yes | 'See who did what in the audit log' release note. | [shapes.co/releases/see-who-did-what-in-the-audit](https://shapes.co/releases/see-who-did-what-in-the-audit-log?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=platform) | | [Off switch for AI features](/capabilities/off_switch/) | yes | Separate switches for AI chat, AI-built Vibes and MCP access, on by default. | [shapes.co/releases/turn-ai-features-on-or-off](https://shapes.co/releases/turn-ai-features-on-or-off?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=platform) | | [Sensitive-field controls](/capabilities/sensitive_controls/) | yes | Field-level permissions: edit, view or hidden per permission level (help center), and a release note on restricting fields. | [help.shapes.co/en/articles/9569339-permission-le](https://help.shapes.co/en/articles/9569339-permission-levels?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=platform) | | [Official MCP server or connector, with access level](/capabilities/mcp_official/) | yes | mcp.shapes.co: 35 read-only tools, OAuth 2.0 with PKCE, official Claude connector. | [mcp.shapes.co/](https://mcp.shapes.co/?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=platform) | | [AI governance certification held or in progress (ISO 42001)](/capabilities/ai_governance_cert/) | partial | ISO 42001 in progress on the security page. | [shapes.co/security-and-compliance](https://shapes.co/security-and-compliance?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=platform) | Read 8 Nov 2026. A cell says "none found" about the pages read on that date, never about the product. ## In the Agent Diary - [Shapes: onboarding a hiring wave](/agents/onboarding/shapes/): runs the process - [Shapes: a compensation cycle](/agents/compensation-cycle/shapes/): runs the process - [Shapes: employee questions from policy](/agents/employee-questions/shapes/): runs the process - [Shapes MCP: status, access and what it does](/mcp/shapes/) ## The seven criteria for Shapes | Criterion | Weight | Score | Why | | --- | --- | --- | --- | | Build without the vendor | 15 | 9 | Dashboards, tools and workflows described in plain language on live permission-aware data, a community hub of named built views, and documented Workflows agent, any-field and date triggers, conditional tasks and task views, with only a template gallery missing. | | Agents that act, under control | 30 | 9 | Agents handle onboarding and the comp cycle, nothing goes live until approved, an audit log, separate AI off switches and field-level permissions are documented across several pages, with offboarding only partial. | | Openness to the AI stack | 25 | 8 | An official read-only MCP server and Claude connector with 35 tools, OAuth 2.0, admin off switch, API on Core, webhooks on Plus and a long integrations list are documented, but MCP cannot write. | | Time to value | 5 | 8 | Days-not-weeks claim, step-by-step guide, spreadsheet or agent directory build, guided implementation and sandbox on Plus are documented across two sources, but no fee amount. | | Pricing transparency | 5 | 8 | Per-seat prices ($8 and $12 billed annually), plan contents, seat counting, cancellation terms and in-product billing are documented, but add-ons are talk-to-sales and the minimum amount is not stated. | | Fit to the organization's shape | 5 | 8 | Custom fields and tables, permission roles with filters, unlimited entities on Plus, workflows assigning policies per office and role, and per-policy approvers are documented with two sources, with plan limits and no currency or language detail. | | Trust | 15 | 9 | SOC 2 Type II, ISO 27001, GDPR, ISO 42001 in progress, audit log, field-level permissions, backups, export and 60 day deletion or return, across several pages; minor gap on some cert detail. | Weighted total 8.6, rank 1 of 14 on Best AI Agents for HR. ## Breadth, reported and not weighted - **Modules:** 9: People Directory, Time Off, Workflows, Attendance, Surveys, Docs and eSign, Performance and Growth, People Analytics, Compensation - **Native payroll:** none; integrates with ADP, Moorepay, PayFit, Sage, Staffology, Papaya Global, Deel, Remote, Bright - **ATS:** none; integrates with Greenhouse, Lever, Ashby, Workable, Teamtailor, JazzHR, Spark Hire - **IT and devices:** none; identity via Okta, OneLogin, JumpCloud, Microsoft Entra - **Integrations:** 35 named on the home page wall - **Countries:** one country or ten (About page); UK, Israel, EU and US payroll integrations ## Where it is strongest, and where it is weak Building the tool, view or workflow you need by describing it, and handing whole processes to agents that do nothing until you approve. Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Best for A People team that wants agents to run onboarding and the comp cycle under an approval it can see, with the switches and the log published, and its outside AI tools reading rather than writing. ## Sources - Shapes: shapes.co/vibe-views: https://shapes.co/vibe-views (read 10 Nov 2026) - Shapes: shapes.co/releases/workflows-2-0-now-live-for-everyone: https://shapes.co/releases/workflows-2-0-now-live-for-everyone (read 10 Nov 2026) - Shapes: shapes.co/agents: https://shapes.co/agents (read 10 Nov 2026) - Shapes: shapes.co/releases/turn-ai-features-on-or-off: https://shapes.co/releases/turn-ai-features-on-or-off (read 10 Nov 2026) - Shapes: mcp.shapes.co/: https://mcp.shapes.co/ (read 10 Nov 2026) - Shapes: shapes.co/pricing: https://shapes.co/pricing (read 10 Nov 2026) - Shapes: shapes.co/: https://shapes.co/ (read 10 Nov 2026) - Shapes: shapes.co/releases/build-your-directory-with-an-agent: https://shapes.co/releases/build-your-directory-with-an-agent (read 10 Nov 2026) - Shapes: help.shapes.co/en/articles/9569350-shapes-pricing: https://help.shapes.co/en/articles/9569350-shapes-pricing (read 10 Nov 2026) - Shapes: shapes.co/releases/workflows-assign-time-and-payroll-policies: https://shapes.co/releases/workflows-assign-time-and-payroll-policies (read 10 Nov 2026) - Shapes: shapes.co/security-and-compliance: https://shapes.co/security-and-compliance (read 10 Nov 2026) - Shapes: shapes.co/legal/dpa: https://shapes.co/legal/dpa (read 10 Nov 2026) - Shapes: shapes.co/releases/ai-chat-now-reads-the-help-center: https://shapes.co/releases/ai-chat-now-reads-the-help-center (read 8 Nov 2026) - Shapes: shapes.co/agents: https://shapes.co/agents (read 8 Nov 2026) - Shapes: shapes.co/: https://shapes.co/ (read 8 Nov 2026) - Shapes: shapes.co/releases/run-workflows-for-new-starters-and-leavers: https://shapes.co/releases/run-workflows-for-new-starters-and-leavers (read 8 Nov 2026) - Shapes: shapes.co/releases/workflows-2-0-now-live-for-everyone: https://shapes.co/releases/workflows-2-0-now-live-for-everyone (read 8 Nov 2026) - Shapes: shapes.co/releases/see-who-did-what-in-the-audit-log: https://shapes.co/releases/see-who-did-what-in-the-audit-log (read 8 Nov 2026) - Shapes: shapes.co/releases/turn-ai-features-on-or-off: https://shapes.co/releases/turn-ai-features-on-or-off (read 8 Nov 2026) - Shapes: help.shapes.co/en/articles/9569339-permission-levels: https://help.shapes.co/en/articles/9569339-permission-levels (read 8 Nov 2026) - Shapes: mcp.shapes.co/: https://mcp.shapes.co/ (read 8 Nov 2026) - Shapes: shapes.co/security-and-compliance: https://shapes.co/security-and-compliance (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/platforms/shapes/ --- # Humaans's agents: Athena runs processes with an audit log of every action, and MCP writes for everyone Agents, rank 2 of 14, read 8 Nov 2026 **The short answer.** Humaans scores 7 on agents and ranks second at 7.2. Athena runs end-to-end processes with pre-built agents, a natural-language agent builder, request triaging, approvals and an audit log of all actions; the MCP server is included for all customers with read and write actions over OAuth 2.0. No off switch was found on 8 Nov 2026. Humaans is the other platform besides Shapes that describes an agent running onboarding and a comp cycle end to end, and the only one with a writing MCP server included for every customer. The approval column is partial because approvals and notifications exist and letters are 'ready for approval' without a blanket rule. The missing control is the switch: nothing on the AI page says how to turn Athena off for a team. - **Acts on:** onboarding a hiring wave: yes; a compensation cycle: yes; employee questions from policy: yes - **Approval model:** partial: Approvals and notifications; letters 'ready for approval'; no blanket 'nothing goes live' statement found. - **Log:** yes: 'Audit log of all actions' on Athena; Audit Log feature. - **Off switch:** none found: Not stated. - **Sensitive fields:** yes: Role-based permissions and explicit data scopes when connecting. - **MCP:** live for all customers, read and write: Included at no extra cost for all customers, OAuth 2.0 with DCR, write actions. - **AI governance:** none found: SOC 2 Type II, ISO 27001; ISO 42001 not named. ## The 12 capabilities for Humaans, with the evidence | Capability | Found | Evidence | Source | | --- | --- | --- | --- | | [Answers employee questions from policy](/capabilities/answers_policy/) | yes | Knowledge base answers, time away policy answers, employee request resolution. | [humaans.io/product-releases-2025](https://humaans.io/product-releases-2025) | | [Drafts changes](/capabilities/drafts_changes/) | yes | Stage comp increases and generate letters ready for approval. | [humaans.io/mcp](https://humaans.io/mcp) | | [Runs onboarding end to end](/capabilities/runs_onboarding/) | yes | 'Onboarding that runs itself'; end-to-end process automation. | [humaans.io/careers](https://humaans.io/careers) | | [Runs a compensation cycle](/capabilities/runs_comp_cycle/) | yes | Run the April comp cycle example: review in scope, stage increases, generate letters for approval. | [humaans.io/mcp](https://humaans.io/mcp) | | [Runs offboarding](/capabilities/runs_offboarding/) | partial | Onboarding and offboarding automation on Growth; Athena end-to-end processes. | [humaans.io/pricing](https://humaans.io/pricing) | | [Acts in other systems (Slack, Teams, email, ATS, payroll)](/capabilities/acts_other_systems/) | yes | Slack-native Athena; MCP sends e-sign documents and triggers workflows. | [humaans.io/mcp](https://humaans.io/mcp) | | [Approval before anything goes live](/capabilities/approval_before_live/) | partial | Approvals and notifications; letters 'ready for approval'; no blanket 'nothing goes live' statement found. | [humaans.io/pricing](https://humaans.io/pricing) | | [Audit log of agent actions](/capabilities/agent_audit_log/) | yes | 'Audit log of all actions' on Athena; Audit Log feature. | [humaans.io/pricing](https://humaans.io/pricing) | | [Off switch for AI features](/capabilities/off_switch/) | none found | Not stated. | [humaans.io/ai](https://humaans.io/ai) | | [Sensitive-field controls](/capabilities/sensitive_controls/) | yes | Role-based permissions and explicit data scopes when connecting. | [humaans.io/mcp](https://humaans.io/mcp) | | [Official MCP server or connector, with access level](/capabilities/mcp_official/) | yes | Included at no extra cost for all customers, OAuth 2.0 with DCR, write actions. | [humaans.io/mcp](https://humaans.io/mcp) | | [AI governance certification held or in progress (ISO 42001)](/capabilities/ai_governance_cert/) | none found | SOC 2 Type II, ISO 27001; ISO 42001 not named. | [humaans.io/mcp](https://humaans.io/mcp) | Read 8 Nov 2026. A cell says "none found" about the pages read on that date, never about the product. ## In the Agent Diary - [Humaans: onboarding a hiring wave](/agents/onboarding/humaans/): runs the process - [Humaans: a compensation cycle](/agents/compensation-cycle/humaans/): runs the process - [Humaans: employee questions from policy](/agents/employee-questions/humaans/): runs the process - [Humaans MCP: status, access and what it does](/mcp/humaans/) ## The seven criteria for Humaans | Criterion | Weight | Score | Why | | --- | --- | --- | --- | | Build without the vendor | 15 | 7 | Tables with live data and agents created in natural language, with workflow templates, date triggers, conditional actions and task views, but event triggers are only partly documented and sharing and permissions are not detailed. | | Agents that act, under control | 30 | 7 | Agents run onboarding and a comp cycle end to end with approvals, an audit log of all actions and scoped permissions, but there is no off switch and no blanket approval-before-live statement. | | Openness to the AI stack | 25 | 9 | MCP is included for all customers with read and write actions, OAuth 2.0 with dynamic client registration, many named clients, audited actions, API on Growth and webhooks on Enterprise, with the gap that webhooks need a higher plan. | | Time to value | 5 | 5 | A few days to a few months and a three week claim are published and CSV import and an Enterprise sandbox are listed, but only by marketing pages, with no fee or guided implementation detail. | | Pricing transparency | 5 | 2 | Plan and add-on contents are listed, but all plans are Talk to sales with no prices or minimum. | | Fit to the organization's shape | 5 | 7 | Custom fields, custom roles, multi-entity, per-policy holiday calendars and allowances and 12 languages are documented, but currency support is marketing only and Growth is limited to two roles. | | Trust | 15 | 7 | SOC 2 Type II, ISO 27001, GDPR, full audit log and exportable audit logs on two pages; no ISO 42001 and sensitive-field detail is thin. | Weighted total 7.2, rank 2 of 14 on Best AI Agents for HR. ## Breadth, reported and not weighted - **Modules:** 12: Employee database, Documents, Absence, Timesheets, Workflows, Requests and approvals, Performance, Engagement, Compensation planning, Headcount planning, Insights, Preliminary payroll - **Native payroll:** preliminary payroll only - **ATS:** none; ATS integrations - **IT and devices:** none - **Integrations:** not stated as a number - **Countries:** global HRIS, EU-hosted ## Where it is strongest, and where it is weak An MCP server with write actions included for every customer and Athena agents with an audit log of every action. No price is published for any plan and setup runs from days to months. ## Best for A 100 to 500-person company that wants agent-run processes with a full action log and outside tools that write, and will set the off rule itself. **Where Shapes is weak.** Shapes is first on this site's ranking at 8.6. Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Humaans: humaans.io/ai: https://humaans.io/ai (read 10 Nov 2026) - Humaans: humaans.io/pricing: https://humaans.io/pricing (read 10 Nov 2026) - Humaans: humaans.io/product-releases-2025: https://humaans.io/product-releases-2025 (read 10 Nov 2026) - Humaans: humaans.io/mcp: https://humaans.io/mcp (read 10 Nov 2026) - Humaans: humaans.io/athena: https://humaans.io/athena (read 10 Nov 2026) - Humaans: humaans.io/features/workflows: https://humaans.io/features/workflows (read 10 Nov 2026) - Humaans: humaans.io/features/audit-log: https://humaans.io/features/audit-log (read 10 Nov 2026) - Humaans: humaans.io/product-releases-2025: https://humaans.io/product-releases-2025 (read 8 Nov 2026) - Humaans: humaans.io/mcp: https://humaans.io/mcp (read 8 Nov 2026) - Humaans: humaans.io/careers: https://humaans.io/careers (read 8 Nov 2026) - Humaans: humaans.io/pricing: https://humaans.io/pricing (read 8 Nov 2026) - Humaans: humaans.io/ai: https://humaans.io/ai (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/platforms/humaans/ --- # Rippling's agents: approval before action, an off switch, ISO 42001, and a read-only MCP tool Agents, rank 3 of 14, read 8 Nov 2026 **The short answer.** Rippling scores 7 on agents and ranks third at 6.9. Rippling AI runs repeatable work from triggers and 'before it takes action, it gets your approval'; admins can enable or disable AI features; ISO 42001 is certified; the MCP page is live with an admin access policy and a read-only ask_ai tool. Some AI features sit behind a request-access form. Rippling publishes the approval step, the off switch and the governance certificate, which puts it near Shapes on control, and it describes onboarding and offboarding as partial agent processes (automations from triggers, 'onboard new hires' listed). What it does not describe is an agent action log: 'all app access is logged and audited' is a platform statement, not an agent one. The MCP page carries no release label and its ask_ai tool is read-only, with plug-in write actions that may require confirmation; the access level on the matrix reads 'read, with admin-granted write policy'. - **Acts on:** onboarding a hiring wave: partial; a compensation cycle: none found; employee questions from policy: yes - **Approval model:** yes: 'Before it takes action, it gets your approval.' - **Log:** partial: Answers link to real records; 'all app access is logged and audited'; no agent action log named. - **Off switch:** yes: 'Admins can enable or disable AI features.' - **Sensitive fields:** yes: Data permissions configured at the platform level. - **MCP:** live for all customers, read, with admin-granted write policy: Rippling MCP page with ask_ai (read-only) and an admin access policy; no release label. - **AI governance:** yes: ISO 42001 certified. ## The 12 capabilities for Rippling, with the evidence | Capability | Found | Evidence | Source | | --- | --- | --- | --- | | [Answers employee questions from policy](/capabilities/answers_policy/) | yes | Employees get personalized pay, benefits and policy explanations. | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | | [Drafts changes](/capabilities/drafts_changes/) | yes | Plan reorgs, draft hires; AI maps bonus data and flags issues. | [www.rippling.com/platform/ai/rippling-mcp](https://www.rippling.com/platform/ai/rippling-mcp) | | [Runs onboarding end to end](/capabilities/runs_onboarding/) | partial | 'Onboard new hires' listed; automations run repeatable work from triggers. | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | | [Runs a compensation cycle](/capabilities/runs_comp_cycle/) | none found | Compensation cycles are not mentioned on the AI pages read. | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | | [Runs offboarding](/capabilities/runs_offboarding/) | partial | Offboarding disables devices, apps and access through IAM and MDM (automation). | [www.rippling.com/trust/security](https://www.rippling.com/trust/security) | | [Acts in other systems (Slack, Teams, email, ATS, payroll)](/capabilities/acts_other_systems/) | yes | Acts across HR, IT and finance and connected tools via plug-ins, limited to actions the integration supports. | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | | [Approval before anything goes live](/capabilities/approval_before_live/) | yes | 'Before it takes action, it gets your approval.' | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | | [Audit log of agent actions](/capabilities/agent_audit_log/) | partial | Answers link to real records; 'all app access is logged and audited'; no agent action log named. | [www.rippling.com/trust/security](https://www.rippling.com/trust/security) | | [Off switch for AI features](/capabilities/off_switch/) | yes | 'Admins can enable or disable AI features.' | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | | [Sensitive-field controls](/capabilities/sensitive_controls/) | yes | Data permissions configured at the platform level. | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | | [Official MCP server or connector, with access level](/capabilities/mcp_official/) | yes | Rippling MCP page with ask_ai (read-only) and an admin access policy; no release label. | [www.rippling.com/platform/ai/rippling-mcp](https://www.rippling.com/platform/ai/rippling-mcp) | | [AI governance certification held or in progress (ISO 42001)](/capabilities/ai_governance_cert/) | yes | ISO 42001 certified. | [www.rippling.com/trust/security](https://www.rippling.com/trust/security) | Read 8 Nov 2026. A cell says "none found" about the pages read on that date, never about the product. ## In the Agent Diary - [Rippling: onboarding a hiring wave](/agents/onboarding/rippling/): runs part of it - [Rippling: employee questions from policy](/agents/employee-questions/rippling/): runs the process - [Rippling MCP: status, access and what it does](/mcp/rippling/) ## The seven criteria for Rippling | Criterion | Weight | Score | Why | | --- | --- | --- | --- | | Build without the vendor | 15 | 8 | Reports and dashboards from a prompt that are editable and shareable, automations described in plain language with schedule, data and webhook triggers, branching, templates, App Studio and an inbox, with the builder claim largely vendor-stated and a library not clearly documented. | | Agents that act, under control | 30 | 7 | Human confirmation before actions, AI off switches, draft state for high-impact MCP actions, hidden sensitive fields and an activity log are documented, but the automations are waitlisted and agent action logging is partial. | | Openness to the AI stack | 25 | 7 | A first-party MCP with an admin gateway that writes within permissions (high-impact actions become drafts), named clients, plug-ins and webhooks are documented, but MCP calls need a Rippling AI subscription and carry no release label. | | Time to value | 5 | 3 | An implementation team and DIY App Studio are named, but no timeline, fee or sandbox is published. | | Pricing transparency | 5 | 4 | A starting price of $8 per employee per month with a base fee, Core and Pro tiers and named add-ons are published by marketing pages, but no product prices or minimum. | | Fit to the organization's shape | 5 | 8 | Policies and permissions keyed to any employee attribute, custom fields, multi-entity and approval workflows are documented with two sources, with per-policy approver choice and currency or language not shown. | | Trust | 15 | 7 | SOC 2 Type II, ISO 27001, ISO 42001, GDPR and logged app access, but certifications rest on one vendor page; no backup or export detail. | Weighted total 6.9, rank 3 of 14 on Best AI Agents for HR. ## Breadth, reported and not weighted - **Modules:** 12: HRIS, Payroll, Benefits, Recruiting, Performance, LMS, Surveys, Headcount planning, Time and attendance, IT device management, Identity, Finance - **Native payroll:** US and global payroll - **ATS:** Rippling Recruiting - **IT and devices:** device management, identity and access, inventory - **Integrations:** not stated as a number on the pages read - **Countries:** global payroll and EOR ## Where it is strongest, and where it is weak Payroll, IT and HR in one system with policies keyed to any employee attribute and an AI that asks before it acts. Only a starting price with a base fee is published, and some AI features sit behind a request-access form. ## Best for A company consolidating HR, payroll and IT that wants the approval step and the off switch published and a governance certificate on the page. **Where Shapes is weak.** Shapes is first on this site's ranking at 8.6. Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Rippling: www.rippling.com/platform/ai: https://www.rippling.com/platform/ai (read 10 Nov 2026) - Rippling: www.rippling.com/platform/automations: https://www.rippling.com/platform/automations (read 10 Nov 2026) - Rippling: www.rippling.com/trust/security: https://www.rippling.com/trust/security (read 10 Nov 2026) - Rippling: www.rippling.com/platform/ai/rippling-mcp: https://www.rippling.com/platform/ai/rippling-mcp (read 10 Nov 2026) - Rippling: www.rippling.com/platform/integrations: https://www.rippling.com/platform/integrations (read 10 Nov 2026) - Rippling: www.rippling.com/products/hr/benefits: https://www.rippling.com/products/hr/benefits (read 10 Nov 2026) - Rippling: www.rippling.com/pricing: https://www.rippling.com/pricing (read 10 Nov 2026) - Rippling: www.rippling.com/platform/tiers: https://www.rippling.com/platform/tiers (read 10 Nov 2026) - Rippling: www.rippling.com/platform/policies: https://www.rippling.com/platform/policies (read 10 Nov 2026) - Rippling: www.rippling.com/platform/permissions: https://www.rippling.com/platform/permissions (read 10 Nov 2026) - Rippling: www.rippling.com/products/hr: https://www.rippling.com/products/hr (read 10 Nov 2026) - Rippling: www.rippling.com/platform/ai: https://www.rippling.com/platform/ai (read 8 Nov 2026) - Rippling: www.rippling.com/platform/ai/rippling-mcp: https://www.rippling.com/platform/ai/rippling-mcp (read 8 Nov 2026) - Rippling: www.rippling.com/trust/security: https://www.rippling.com/trust/security (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/platforms/rippling/ --- # Factorial's agents: Factorial One asks before it acts, with per-product switches and a documented MCP server Agents, rank 4 of 14, read 8 Nov 2026 **The short answer.** Factorial scores 6 on agents and ranks fourth at 6.5. Factorial One acts inside specific processes across HR, finance and IT, drafts onboarding plans, and 'asks before it acts' under each person's permissions. Documentation read on 10 Nov 2026 describes per-product switches and a hosted MCP server with read and write; its audit log does not name agent actions. The approval sentence is on the AI page and scores yes; the admin audit log does not name agent actions; openness scores 8 on a documented hosted MCP server with OAuth 2.0, read and write, a public API and retried webhooks, though it is not in connector directories. The matrix MCP cell, read on 8 Nov 2026, is unchanged until the next edition. The diary records Factorial as partial on onboarding: plans drafted and tickets resolved on approval, not a wave run end to end. - **Acts on:** onboarding a hiring wave: partial; a compensation cycle: none found; employee questions from policy: yes - **Approval model:** yes: 'It asks before it acts.' - **Log:** none found: Audit logging is a trust-center heading; no agent log named. - **Off switch:** none found: Not stated. - **Sensitive fields:** yes: 'ONE enforces each person's permissions on every request. No permission, no action.' - **MCP:** live for all customers, read and write (OAuth 2.0): Hosted MCP server at mcp.factorialhr.com with OAuth 2.0, added to AI clients by URL; not yet in the Claude or ChatGPT connector directories (API docs, read 2026-11-10). - **AI governance:** none found: Not named. ## The 12 capabilities for Factorial, with the evidence | Capability | Found | Evidence | Source | | --- | --- | --- | --- | | [Answers employee questions from policy](/capabilities/answers_policy/) | yes | 'Ask One what the rule is and it quotes your handbook.' | [factorial.com/ai](https://factorial.com/ai) | | [Drafts changes](/capabilities/drafts_changes/) | yes | Drafts job offers, policies, onboarding plans, dashboards. | [factorial.com/ai](https://factorial.com/ai) | | [Runs onboarding end to end](/capabilities/runs_onboarding/) | partial | Drafts onboarding plans; tickets routed and resolved on approval. | [factorial.com/ai](https://factorial.com/ai) | | [Runs a compensation cycle](/capabilities/runs_comp_cycle/) | none found | Not described. | [factorial.com/ai](https://factorial.com/ai) | | [Runs offboarding](/capabilities/runs_offboarding/) | none found | Not described. | [factorial.com/ai](https://factorial.com/ai) | | [Acts in other systems (Slack, Teams, email, ATS, payroll)](/capabilities/acts_other_systems/) | yes | Meets you in Slack, HubSpot and email; acts, calls, confirms and notifies. | [factorial.com/ai](https://factorial.com/ai) | | [Approval before anything goes live](/capabilities/approval_before_live/) | yes | 'It asks before it acts.' | [factorial.com/ai](https://factorial.com/ai) | | [Audit log of agent actions](/capabilities/agent_audit_log/) | none found | Audit logging is a trust-center heading; no agent log named. | [trust.factorial.co/](https://trust.factorial.co/) | | [Off switch for AI features](/capabilities/off_switch/) | none found | Not stated. | [factorial.com/ai](https://factorial.com/ai) | | [Sensitive-field controls](/capabilities/sensitive_controls/) | yes | 'ONE enforces each person's permissions on every request. No permission, no action.' | [factorial.com/ai](https://factorial.com/ai) | | [Official MCP server or connector, with access level](/capabilities/mcp_official/) | yes | Hosted MCP server at mcp.factorialhr.com with OAuth 2.0, added to AI clients by URL; not yet in the Claude or ChatGPT connector directories (API docs, read 2026-11-10). | [apidoc.factorialhr.com/docs/mcp-server](https://apidoc.factorialhr.com/docs/mcp-server) | | [AI governance certification held or in progress (ISO 42001)](/capabilities/ai_governance_cert/) | none found | Not named. | [trust.factorial.co/](https://trust.factorial.co/) | Read 8 Nov 2026. A cell says "none found" about the pages read on that date, never about the product. ## In the Agent Diary - [Factorial: onboarding a hiring wave](/agents/onboarding/factorial/): runs part of it - [Factorial: employee questions from policy](/agents/employee-questions/factorial/): runs the process - [Factorial MCP: status, access and what it does](/mcp/factorial/) ## The seven criteria for Factorial | Criterion | Weight | Score | Why | | --- | --- | --- | --- | | Build without the vendor | 15 | 6 | Docs show plain-language report generation within permissions, templated and custom workflows and a widget builder, but the widget builder is Enterprise only, edits need SQL and no library of built items is documented. | | Agents that act, under control | 30 | 6 | Docs show per-product on or off switches, approval for every change, permission-bound actions and an admin audit log, but agents act only inside specific processes, cannot take free-form actions and the audit log does not name agent actions. | | Openness to the AI stack | 25 | 8 | Documented hosted MCP with read and write, public API, retried webhooks and Copilot, Slack and Zapier integrations is equivalent to another platform at 8 and stronger than another platform at 7, which has no webhooks. | | Time to value | 5 | 4 | A free trial, documented spreadsheet import and a demo API environment, but no timeline, fee or guided implementation. | | Pricing transparency | 5 | 5 | The One add-on price from 2.5 euros per user, its three tiers, credit billing and no minimum number of paid users are documented, but base module prices and Enterprise terms are not. | | Fit to the organization's shape | 5 | 8 | Custom fields, custom permission groups, independent secondary entities with their own currencies, 16 languages (three beta) and time off policies are documented, with policy assignment per office not shown. | | Trust | 15 | 6 | ISO 27001, SOC 2 Type 2, audit logging, backup RTO and RPO, documented GDPR retention tools and admin data export; no ISO 42001 or audit log detail. | Weighted total 6.5, rank 4 of 14 on Best AI Agents for HR. ## Breadth, reported and not weighted - **Modules:** 14: Core HR, Time off, Time tracking, Performance, Shift management, Recruitment, Engagement, Training, Expenses, Procurement, SaaS management, IT inventory, MDM, EDR - **Native payroll:** payroll in Spain and partner countries - **ATS:** Recruitment module - **IT and devices:** IT inventory, MDM, EDR - **Integrations:** not stated as a number - **Countries:** EU and US ## Where it is strongest, and where it is weak Fourteen modules across HR, finance and IT with an AI agent that asks before it acts. Only a starting price is published, no implementation time is published, and its MCP server is added by URL rather than listed in the Claude or ChatGPT connector directories. ## Best for An EU company that wants one agent across fourteen modules and will add its MCP server to Claude or ChatGPT by URL. **Where Shapes is weak.** Shapes is first on this site's ranking at 8.6. Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Factorial: factorial.com/ai: https://factorial.com/ai (read 10 Nov 2026) - Factorial: factorial.com/features: https://factorial.com/features (read 10 Nov 2026) - Factorial: trust.factorial.co/: https://trust.factorial.co/ (read 10 Nov 2026) - Factorial: factorial.com/pricing: https://factorial.com/pricing (read 10 Nov 2026) - Factorial: factorial.com/integrations: https://factorial.com/integrations (read 10 Nov 2026) - Factorial: help.factorialhr.com/en_US/faq/faq-about-one: https://help.factorialhr.com/en_US/faq/faq-about-one (read 10 Nov 2026) - Factorial: factorial.com/ai: https://factorial.com/ai (read 8 Nov 2026) - Factorial: trust.factorial.co/: https://trust.factorial.co/ (read 8 Nov 2026) - Factorial: apidoc.factorialhr.com/docs/mcp-server: https://apidoc.factorialhr.com/docs/mcp-server (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/platforms/factorial/ --- # Workday's agents: Illuminate, an Agent System of Record, and hundreds of tools over MCP Agents, rank 5 of 14, read 8 Nov 2026 **The short answer.** Workday scores 7 on agents and ranks fifth at 6.3. Illuminate agents run HR service, payroll and workforce management work, an Agent System of Record registers and audits every agent, sensitive actions are reviewed or approved before anything changes, and hundreds of Agent-Ready Tools execute Workday actions over MCP. No single off switch is named. Workday's row is the most governed on the card: approval, audit log, sensitive controls and ISO 42001 all yes, with the Agent Passport verifying agents and the Agent Gateway admitting Claude, Gemini, OpenAI and Copilot. The off switch is partial because agents are governed through registration rather than a setting. Onboarding is partial (workflows in HCM; agents for HR service and admin) and the comp cycle none found as an agent process; the builders are aimed at developers, which is why build scores 5; time to value scores 6. - **Acts on:** onboarding a hiring wave: partial; a compensation cycle: none found; employee questions from policy: yes - **Approval model:** yes: 'bring people in to review or approve sensitive actions before anything changes'; sandboxed what-ifs. - **Log:** yes: Agent System of Record with full auditability; Agent Analytics. - **Off switch:** partial: Agents registered and governed in the Agent System of Record; no single off switch named. - **Sensitive fields:** yes: Actions scoped to the user's identity with safeguards for sensitive enterprise data. - **MCP:** live for all customers, executes Workday actions (write): Hundreds of Agent-Ready Tools over MCP; no release label. - **AI governance:** yes: ISO 42001 certificate; NIST AI RMF attestation. ## The 12 capabilities for Workday, with the evidence | Capability | Found | Evidence | Source | | --- | --- | --- | --- | | [Answers employee questions from policy](/capabilities/answers_policy/) | yes | HR service and self-service agents. | [www.workday.com/en-us/artificial-intelligence/ai](https://www.workday.com/en-us/artificial-intelligence/ai-agents/hr-service.html) | | [Drafts changes](/capabilities/drafts_changes/) | yes | Agents draft and execute within guardrails. | [www.workday.com/en-us/artificial-intelligence/ai](https://www.workday.com/en-us/artificial-intelligence/ai-agents.html) | | [Runs onboarding end to end](/capabilities/runs_onboarding/) | partial | Onboarding workflows in HCM; agents for HR service and admin. | [www.workday.com/en-us/artificial-intelligence/ai](https://www.workday.com/en-us/artificial-intelligence/ai-agents/admin.html) | | [Runs a compensation cycle](/capabilities/runs_comp_cycle/) | none found | Not described as an agent on the pages read. | [www.workday.com/en-us/artificial-intelligence/ai](https://www.workday.com/en-us/artificial-intelligence/ai-agents.html) | | [Runs offboarding](/capabilities/runs_offboarding/) | none found | Not described as an agent. | [www.workday.com/en-us/artificial-intelligence/ai](https://www.workday.com/en-us/artificial-intelligence/ai-agents.html) | | [Acts in other systems (Slack, Teams, email, ATS, payroll)](/capabilities/acts_other_systems/) | yes | Agent Gateway supports MCP and A2A across systems. | [blog.workday.com/en-us/managing-ai-powered-futur](https://blog.workday.com/en-us/managing-ai-powered-future-of-work.html) | | [Approval before anything goes live](/capabilities/approval_before_live/) | yes | 'bring people in to review or approve sensitive actions before anything changes'; sandboxed what-ifs. | [www.workday.com/en-us/artificial-intelligence/wo](https://www.workday.com/en-us/artificial-intelligence/workday-sana.html) | | [Audit log of agent actions](/capabilities/agent_audit_log/) | yes | Agent System of Record with full auditability; Agent Analytics. | [www.workday.com/en-us/ai-platform/agent-app-buil](https://www.workday.com/en-us/ai-platform/agent-app-builder.html) | | [Off switch for AI features](/capabilities/off_switch/) | partial | Agents registered and governed in the Agent System of Record; no single off switch named. | [www.workday.com/en-us/ai-platform/agent-system-o](https://www.workday.com/en-us/ai-platform/agent-system-of-record.html) | | [Sensitive-field controls](/capabilities/sensitive_controls/) | yes | Actions scoped to the user's identity with safeguards for sensitive enterprise data. | [www.workday.com/en-us/ai-platform/agent-tools-mc](https://www.workday.com/en-us/ai-platform/agent-tools-mcp.html) | | [Official MCP server or connector, with access level](/capabilities/mcp_official/) | yes | Hundreds of Agent-Ready Tools over MCP; no release label. | [www.workday.com/en-us/ai-platform/agent-tools-mc](https://www.workday.com/en-us/ai-platform/agent-tools-mcp.html) | | [AI governance certification held or in progress (ISO 42001)](/capabilities/ai_governance_cert/) | yes | ISO 42001 certificate; NIST AI RMF attestation. | [www.workday.com/en-us/why-workday/trust/complian](https://www.workday.com/en-us/why-workday/trust/compliance.html) | Read 8 Nov 2026. A cell says "none found" about the pages read on that date, never about the product. ## In the Agent Diary - [Workday: onboarding a hiring wave](/agents/onboarding/workday/): runs part of it - [Workday: employee questions from policy](/agents/employee-questions/workday/): runs the process - [Workday MCP: status, access and what it does](/mcp/workday/) ## The seven criteria for Workday | Criterion | Weight | Score | Why | | --- | --- | --- | --- | | Build without the vendor | 15 | 5 | Agent & App Builder and Developer Agent build apps and agents in plain language and business process templates, triggers and conditions exist, but the builders target developers and prompt-built business processes are not documented. | | Agents that act, under control | 30 | 7 | Agents draft and execute within guardrails, with human review of sensitive actions, sandboxed what-ifs, an Agent System of Record with auditability and identity-scoped data safeguards, but comp cycle and offboarding agents and a single off switch are not documented. | | Openness to the AI stack | 25 | 6 | Official MCP tools with write actions for Claude, Gemini, OpenAI and Copilot and webhooks are documented, but only by vendor pages with no release stage, plan or audit detail. | | Time to value | 5 | 6 | Published 10 to 20 plus week go-live, fixed price with prescribed scope, and a what-if sandbox are documented, but only on one vendor page and no amount is published. | | Pricing transparency | 5 | 2 | The Flex Credits usage model with top-ups and complimentary credits is described on a marketing page only, with no prices, plan contents or minimum. | | Fit to the organization's shape | 5 | 7 | Multi-entity, multi-country and granular security groups with approvers by role are documented, but only one summary source and no custom fields or language detail. | | Trust | 15 | 8 | ISO 27001 family, ISO 42001, SOC 1 and SOC 2, NIST AI RMF and audit trail documented on two pages; backup and export not documented. | Weighted total 6.3, rank 5 of 14 on Best AI Agents for HR. ## Breadth, reported and not weighted - **Modules:** 8: HCM, Payroll, Recruiting, Talent, Learning, Workforce management, Planning, Financial management - **Native payroll:** Workday Payroll (US, UK, Canada, France and partners) - **ATS:** Workday Recruiting - **IT and devices:** none found - **Integrations:** not stated as a number - **Countries:** global ## Where it is strongest, and where it is weak Enterprise finance and HR in one suite with an Agent System of Record and ISO 42001 held. No published price, a ten-to-twenty-plus-week launch, and builders aimed at platform teams rather than an HR admin. ## Best for An enterprise that will run an agent program with a system of record and an IT team, and wants its outside agents to execute actions. **Where Shapes is weak.** Shapes is first on this site's ranking at 8.6. Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Workday: www.workday.com/en-us/ai-platform/agent-app-builder.html: https://www.workday.com/en-us/ai-platform/agent-app-builder.html (read 10 Nov 2026) - Workday: www.workday.com/en-us/ai-platform/augmented-analytics-software.html: https://www.workday.com/en-us/ai-platform/augmented-analytics-software.html (read 10 Nov 2026) - Workday: www.workday.com/en-us/artificial-intelligence/ai-agents.html: https://www.workday.com/en-us/artificial-intelligence/ai-agents.html (read 10 Nov 2026) - Workday: www.workday.com/en-us/artificial-intelligence/workday-sana.html: https://www.workday.com/en-us/artificial-intelligence/workday-sana.html (read 10 Nov 2026) - Workday: www.workday.com/en-us/ai-platform/agent-tools-mcp.html: https://www.workday.com/en-us/ai-platform/agent-tools-mcp.html (read 10 Nov 2026) - Workday: blog.workday.com/en-us/managing-ai-powered-future-of-work.html: https://blog.workday.com/en-us/managing-ai-powered-future-of-work.html (read 10 Nov 2026) - Workday: www.workday.com/en-us/customer-experience/deployment.html: https://www.workday.com/en-us/customer-experience/deployment.html (read 10 Nov 2026) - Workday: www.workday.com/en-us/products/human-capital-management.html: https://www.workday.com/en-us/products/human-capital-management.html (read 10 Nov 2026) - Workday: www.workday.com/en-us/why-workday/trust/compliance.html: https://www.workday.com/en-us/why-workday/trust/compliance.html (read 10 Nov 2026) - Workday: www.workday.com/en-us/why-workday/trust/security.html: https://www.workday.com/en-us/why-workday/trust/security.html (read 10 Nov 2026) - Workday: www.workday.com/en-us/artificial-intelligence/ai-agents/hr-service.html: https://www.workday.com/en-us/artificial-intelligence/ai-agents/hr-service.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/artificial-intelligence/ai-agents.html: https://www.workday.com/en-us/artificial-intelligence/ai-agents.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/artificial-intelligence/ai-agents/admin.html: https://www.workday.com/en-us/artificial-intelligence/ai-agents/admin.html (read 8 Nov 2026) - Workday: blog.workday.com/en-us/managing-ai-powered-future-of-work.html: https://blog.workday.com/en-us/managing-ai-powered-future-of-work.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/artificial-intelligence/workday-sana.html: https://www.workday.com/en-us/artificial-intelligence/workday-sana.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/ai-platform/agent-app-builder.html: https://www.workday.com/en-us/ai-platform/agent-app-builder.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/ai-platform/agent-system-of-record.html: https://www.workday.com/en-us/ai-platform/agent-system-of-record.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/ai-platform/agent-tools-mcp.html: https://www.workday.com/en-us/ai-platform/agent-tools-mcp.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/why-workday/trust/compliance.html: https://www.workday.com/en-us/why-workday/trust/compliance.html (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/platforms/workday/ --- # HiBob's agents: twelve named Bob AI agents and an MCP server that writes, with partial controls Agents, rank 6 of 14, read 8 Nov 2026 **The short answer.** HiBob scores 5 on agents and ranks sixth at 6.1. Twelve named Bob AI agents answer and act within permissions and the MCP server reads and writes where Bob allows; no page describes an agent running onboarding or a comp cycle end to end, the approval step and the audit log are partial, and no agent off switch was found. HiBob is the clearest case of openness ahead of control: its MCP server is live for all customers with read and write actions, which gives it 8 on openness, while the agent controls are what the matrix calls partial. Admins can require confirmation before writes in some AI tools, and the MCP page describes AI logs of data, but no agent action log or approval step for agents is stated. The diary records HiBob as partial on the comp cycle (multi-currency cycles with AI inside) and on offboarding, and none found on onboarding end to end. - **Acts on:** onboarding a hiring wave: none found; a compensation cycle: partial; employee questions from policy: yes - **Approval model:** partial: Admins can require confirmation before writes in some AI tools; no approval step stated for agents. - **Log:** partial: 'AI logs of data in MCP'; no agent action log stated. - **Off switch:** partial: MCP is turned on in Bob company-wide; no AI off switch stated for agents. - **Sensitive fields:** yes: Sensitive data protected by role-based permissions; agents cannot create a new path to restricted information. - **MCP:** live for all customers, read and write where Bob allows: HiBob MCP server, read and write where Bob already allows. - **AI governance:** none found: ISO 42001 not named; GDPR, SOC 2, ISO 27001, HIPAA badges. ## The 12 capabilities for HiBob, with the evidence | Capability | Found | Evidence | Source | | --- | --- | --- | --- | | [Answers employee questions from policy](/capabilities/answers_policy/) | yes | Bob AI gives employees and managers instant self-service answers. | [www.hibob.com/platform/employee-experience-softw](https://www.hibob.com/platform/employee-experience-software/) | | [Drafts changes](/capabilities/drafts_changes/) | yes | Agents take actions such as updating goals, creating development plans. | [www.hibob.com/platform/core/ai/agents/](https://www.hibob.com/platform/core/ai/agents/) | | [Runs onboarding end to end](/capabilities/runs_onboarding/) | none found | No agent described as running onboarding end to end on 2026-11-08. | [www.hibob.com/platform/core/ai/agents/](https://www.hibob.com/platform/core/ai/agents/) | | [Runs a compensation cycle](/capabilities/runs_comp_cycle/) | partial | Compensation agent shows employees below band minimum; the cycle itself runs in the Compensation module with configurable workflows. | [www.hibob.com/platform/core/ai/agents/](https://www.hibob.com/platform/core/ai/agents/) | | [Runs offboarding](/capabilities/runs_offboarding/) | partial | Offboarding tasks start when a departure is confirmed (automation, not an agent). | [www.hibob.com/platform/core/offboarding/](https://www.hibob.com/platform/core/offboarding/) | | [Acts in other systems (Slack, Teams, email, ATS, payroll)](/capabilities/acts_other_systems/) | yes | MCP runs supported HR actions from Claude, ChatGPT, Cursor and Slack. | [www.hibob.com/platform/core/ai/mcp](https://www.hibob.com/platform/core/ai/mcp) | | [Approval before anything goes live](/capabilities/approval_before_live/) | partial | Admins can require confirmation before writes in some AI tools; no approval step stated for agents. | [www.hibob.com/platform/core/ai/mcp](https://www.hibob.com/platform/core/ai/mcp) | | [Audit log of agent actions](/capabilities/agent_audit_log/) | partial | 'AI logs of data in MCP'; no agent action log stated. | [www.hibob.com/platform/core/ai/mcp](https://www.hibob.com/platform/core/ai/mcp) | | [Off switch for AI features](/capabilities/off_switch/) | partial | MCP is turned on in Bob company-wide; no AI off switch stated for agents. | [www.hibob.com/platform/core/ai/mcp](https://www.hibob.com/platform/core/ai/mcp) | | [Sensitive-field controls](/capabilities/sensitive_controls/) | yes | Sensitive data protected by role-based permissions; agents cannot create a new path to restricted information. | [www.hibob.com/platform/core/ai/agents/](https://www.hibob.com/platform/core/ai/agents/) | | [Official MCP server or connector, with access level](/capabilities/mcp_official/) | yes | HiBob MCP server, read and write where Bob already allows. | [www.hibob.com/platform/core/ai/mcp](https://www.hibob.com/platform/core/ai/mcp) | | [AI governance certification held or in progress (ISO 42001)](/capabilities/ai_governance_cert/) | none found | ISO 42001 not named; GDPR, SOC 2, ISO 27001, HIPAA badges. | [www.hibob.com/platform/core/ai/mcp](https://www.hibob.com/platform/core/ai/mcp) | Read 8 Nov 2026. A cell says "none found" about the pages read on that date, never about the product. ## In the Agent Diary - [HiBob: a compensation cycle](/agents/compensation-cycle/hibob/): runs part of it - [HiBob: employee questions from policy](/agents/employee-questions/hibob/): runs the process - [HiBob MCP: status, access and what it does](/mcp/hibob/) ## The seven criteria for HiBob | Criterion | Weight | Score | Why | | --- | --- | --- | --- | | Build without the vendor | 15 | 5 | No-code workflows with conditional logic and branching, lifecycle triggers and natural-language analytics are described, but only on marketing pages and with the AI builder not yet available. | | Agents that act, under control | 30 | 5 | Human approval of write actions, an admin AI off switch, role-bound access and MCP AI logs are stated, but only on marketing pages and with no whole-process agent or detailed agent audit log. | | Openness to the AI stack | 25 | 8 | An official MCP server with read and write where permitted, admin controls, logs, several named clients and a public API with webhooks are documented, with minor gaps on plan and audit detail. | | Time to value | 5 | 5 | A weeks-not-months claim, a separately sold sandbox and implementation fees in partner terms are documented, but no fee amount or guided implementation detail. | | Pricing transparency | 5 | 3 | Core contents and a per-employee pricing basis with separate modules are described by marketing pages, but no amounts and the minimum is only set in the Order Form. | | Fit to the organization's shape | 5 | 8 | Custom fields and tables, group-based permissions, policy visibility settings, location policies and multi-entity are documented, with language support from marketing only and no per-policy approver choice. | | Trust | 15 | 7 | SOC 2 Type 2, ISO 27001, ISO 27018, GDPR, daily backups, export routes, encryption and no AI training are stated on marketing pages; no ISO 42001 and audit log is partial. | Weighted total 6.1, rank 6 of 14 on Best AI Agents for HR. ## Breadth, reported and not weighted - **Modules:** 10: Core HR, Payroll (US, UK), Hiring, Performance, Compensation, Workforce planning, Learning, Surveys, Time and attendance, Benefits administration - **Native payroll:** US and UK payroll - **ATS:** Bob Hiring - **IT and devices:** none found - **Integrations:** 30+ pre-built (finance page) - **Countries:** global; customer story cites 36 countries ## Where it is strongest, and where it is weak An MCP server with read and write actions that any Bob user can switch on, inside a talent and payroll suite. No published price and no agent documented as running onboarding or a comp cycle end to end. ## Best for A company that runs its AI tools outside the HRIS and wants them to write to Bob, and will build its own approval discipline around them. **Where Shapes is weak.** Shapes is first on this site's ranking at 8.6. Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - HiBob: www.hibob.com/platform/hr-reporting-software/: https://www.hibob.com/platform/hr-reporting-software/ (read 10 Nov 2026) - HiBob: www.hibob.com/platform/core/ai/: https://www.hibob.com/platform/core/ai/ (read 10 Nov 2026) - HiBob: www.hibob.com/platform/core/ai/agents/: https://www.hibob.com/platform/core/ai/agents/ (read 10 Nov 2026) - HiBob: www.hibob.com/platform/core/ai/mcp: https://www.hibob.com/platform/core/ai/mcp (read 10 Nov 2026) - HiBob: apidocs.hibob.com/changelog: https://apidocs.hibob.com/changelog (read 10 Nov 2026) - HiBob: www.hibob.com/platform/finance/insights/: https://www.hibob.com/platform/finance/insights/ (read 10 Nov 2026) - HiBob: www.hibob.com/privacy/hibob-partnership-terms/: https://www.hibob.com/privacy/hibob-partnership-terms/ (read 10 Nov 2026) - HiBob: www.hibob.com/pricing-plans/: https://www.hibob.com/pricing-plans/ (read 10 Nov 2026) - HiBob: www.hibob.com/privacy/customer-subscription-terms/: https://www.hibob.com/privacy/customer-subscription-terms/ (read 10 Nov 2026) - HiBob: www.hibob.com/platform/payroll/time-and-attendance/: https://www.hibob.com/platform/payroll/time-and-attendance/ (read 10 Nov 2026) - HiBob: www.hibob.com/platform/planning/compensation/: https://www.hibob.com/platform/planning/compensation/ (read 10 Nov 2026) - HiBob: www.hibob.com/platform: https://www.hibob.com/platform (read 10 Nov 2026) - HiBob: www.hibob.com/platform/employee-experience-software/: https://www.hibob.com/platform/employee-experience-software/ (read 8 Nov 2026) - HiBob: www.hibob.com/platform/core/ai/agents/: https://www.hibob.com/platform/core/ai/agents/ (read 8 Nov 2026) - HiBob: www.hibob.com/platform/core/offboarding/: https://www.hibob.com/platform/core/offboarding/ (read 8 Nov 2026) - HiBob: www.hibob.com/platform/core/ai/mcp: https://www.hibob.com/platform/core/ai/mcp (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/platforms/hibob/ --- # BambooHR's agents: Bamboo AI with an Audit Trail and an off switch, and an MCP server in beta Agents, rank 7 of 14, read 8 Nov 2026 **The short answer.** BambooHR scores 6 on agents and ranks seventh at 5.8. Agentic workflows 'still go through the appropriate person before they're applied or sent', the Audit Trail records whether a person, Bamboo AI or both made a change, and admins turn Bamboo AI off under Account Settings. No agent runs onboarding or a comp cycle; the MCP server is in beta. BambooHR publishes the three controls most of the card lacks, and its Audit Trail sentence is the most specific on the matrix: it names the agent as an actor. The capabilities around the controls are narrower: Bamboo AI drafts schedules, reviews payroll changes and follows up on missing data, with several items marked coming soon. The MCP server is documented in beta over OAuth 2.0 with read, write and delete tools and a Claude connector in beta, which the matrix records as partial until it is general. - **Acts on:** onboarding a hiring wave: none found; a compensation cycle: none found; employee questions from policy: yes - **Approval model:** yes: 'still go through the appropriate person before they're applied or sent'. - **Log:** yes: Audit Trail records whether a person, Bamboo AI or both made a change. - **Off switch:** yes: Admins turn Bamboo AI off or on under Account Settings > Ask BambooHR. - **Sensitive fields:** partial: Access levels gate the connectors; no sensitive-field control named. - **MCP:** beta or early access, read, write and delete tools (beta): MCP server and AI connectors in beta, OAuth 2.0, mostly read-only with Write and Delete tools. - **AI governance:** none found: Not named on the security page. ## The 12 capabilities for BambooHR, with the evidence | Capability | Found | Evidence | Source | | --- | --- | --- | --- | | [Answers employee questions from policy](/capabilities/answers_policy/) | yes | Chat answers questions using documents administrators select. | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | [Drafts changes](/capabilities/drafts_changes/) | yes | Drafts schedules, organises job titles, prepares work for the appropriate person. | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | [Runs onboarding end to end](/capabilities/runs_onboarding/) | none found | Not described on the Bamboo AI update. | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | [Runs a compensation cycle](/capabilities/runs_comp_cycle/) | none found | Not described. | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | [Runs offboarding](/capabilities/runs_offboarding/) | none found | Not described. | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | [Acts in other systems (Slack, Teams, email, ATS, payroll)](/capabilities/acts_other_systems/) | partial | Employee outreach and payroll runs prepared inside BambooHR; Claude connector in beta. | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | [Approval before anything goes live](/capabilities/approval_before_live/) | yes | 'still go through the appropriate person before they're applied or sent'. | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | [Audit log of agent actions](/capabilities/agent_audit_log/) | yes | Audit Trail records whether a person, Bamboo AI or both made a change. | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | [Off switch for AI features](/capabilities/off_switch/) | yes | Admins turn Bamboo AI off or on under Account Settings > Ask BambooHR. | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | [Sensitive-field controls](/capabilities/sensitive_controls/) | partial | Access levels gate the connectors; no sensitive-field control named. | [documentation.bamboohr.com/docs/mcp-server](https://documentation.bamboohr.com/docs/mcp-server) | | [Official MCP server or connector, with access level](/capabilities/mcp_official/) | partial | MCP server and AI connectors in beta, OAuth 2.0, mostly read-only with Write and Delete tools. | [documentation.bamboohr.com/docs/mcp-server](https://documentation.bamboohr.com/docs/mcp-server) | | [AI governance certification held or in progress (ISO 42001)](/capabilities/ai_governance_cert/) | none found | Not named on the security page. | [www.bamboohr.com/security](https://www.bamboohr.com/security) | Read 8 Nov 2026. A cell says "none found" about the pages read on that date, never about the product. ## In the Agent Diary - [BambooHR: employee questions from policy](/agents/employee-questions/bamboohr/): runs the process - [BambooHR MCP: status, access and what it does](/mcp/bamboohr/) ## The seven criteria for BambooHR | Criterion | Weight | Score | Why | | --- | --- | --- | --- | | Build without the vendor | 15 | 4 | Custom reports built and refined from a prompt but several capabilities are beta or coming soon, custom dashboards need Elite, and prompt-built flows and conditional tasks are not documented. | | Agents that act, under control | 30 | 6 | Prepared work goes through the appropriate person, the Audit Trail records agent changes and admins can turn the feature off, but onboarding and comp cycle agents are absent, several items are beta or coming soon, and sensitive-field controls are not named. | | Openness to the AI stack | 25 | 7 | A documented beta MCP server with OAuth 2.0, read and write tools and admin access control, plus signed webhooks and a public API, with beta status and no dynamic client registration as limits. | | Time to value | 5 | 5 | A free trial, 30 extra implementation days on Elite, a dedicated payroll project manager and self-serve setup are documented, but no amount, timeline or customer sandbox. | | Pricing transparency | 5 | 8 | Per-employee prices for Core, Pro and Elite, flat rates for 25 or fewer employees, plan contents and billing options are published, but add-on prices and any in-product billing detail are not. | | Fit to the organization's shape | 5 | 5 | Custom tabs, tables and fields, role visibility, eight languages and localized settings are listed, mostly by marketing, with only a beta MCP docs page and no multi-entity detail. | | Trust | 15 | 5 | SOC 1 and SOC 2 Type 2 plus a concrete audit trail on two pages; no ISO 27001, GDPR, ISO 42001, sensitive fields or backup detail. | Weighted total 5.8, rank 7 of 14 on Best AI Agents for HR. ## Breadth, reported and not weighted - **Modules:** 10: Employee records, Hiring, Onboarding, Time off, Performance, Compensation (Elite), Payroll (US), Benefits administration (US), Time and attendance, Compliance training - **Native payroll:** US payroll add-on - **ATS:** native hiring - **IT and devices:** none found - **Integrations:** marketplace (count not stated) - **Countries:** US-centric with EOR via Remote ## Where it is strongest, and where it is weak Hiring and a larger marketplace, with published per-employee prices and flat rates for very small teams. US-only payroll and benefits, an MCP server still in beta, and several Bamboo AI capabilities marked coming soon. ## Best for A US company that wants the controls published before the agents arrive, and a vendor whose audit trail names the agent. **Where Shapes is weak.** Shapes is first on this site's ranking at 8.6. Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - BambooHR: www.bamboohr.com/product-updates/bamboo-ai: https://www.bamboohr.com/product-updates/bamboo-ai (read 10 Nov 2026) - BambooHR: www.bamboohr.com/pricing: https://www.bamboohr.com/pricing (read 10 Nov 2026) - BambooHR: documentation.bamboohr.com/docs/mcp-server: https://documentation.bamboohr.com/docs/mcp-server (read 10 Nov 2026) - BambooHR: www.bamboohr.com/legal/terms-of-service: https://www.bamboohr.com/legal/terms-of-service (read 10 Nov 2026) - BambooHR: www.bamboohr.com/platform/: https://www.bamboohr.com/platform/ (read 10 Nov 2026) - BambooHR: www.bamboohr.com/security: https://www.bamboohr.com/security (read 10 Nov 2026) - BambooHR: www.bamboohr.com/product-updates/bamboo-ai: https://www.bamboohr.com/product-updates/bamboo-ai (read 8 Nov 2026) - BambooHR: documentation.bamboohr.com/docs/mcp-server: https://documentation.bamboohr.com/docs/mcp-server (read 8 Nov 2026) - BambooHR: www.bamboohr.com/security: https://www.bamboohr.com/security (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/platforms/bamboohr/ --- # Deel's agents: drafts with a human in the loop and a documented MCP server with read and write Agents, rank 8 of 14, read 8 Nov 2026 **The short answer.** Deel scores 5 on agents and ranks eighth at 5.8. Deel's AI drafts job descriptions, policies and courses and answers questions, the AI terms require a human in the loop, and agentic workflows run from triggers. Documentation read on 10 Nov 2026 describes an official MCP server with read and write; no off switch was found. The human-in-the-loop requirement in Deel's AI terms is a partial approval step on the matrix: it is a contractual rule rather than a product control. The acting column is partial (a Slack plug-in and reports). The matrix MCP cell, read on 8 Nov 2026, reads none found; Deel's developer documentation read on 10 Nov 2026 describes an official MCP server over HTTP with read and write OAuth scopes checked on each tool call, which raised its openness score from 3 to 8. The matrix cell is unchanged until the next edition. The diary has no Deel page: none of the three processes is described as agent-run. - **Acts on:** onboarding a hiring wave: none found; a compensation cycle: none found; employee questions from policy: yes - **Approval model:** partial: AI terms: 'Always keep a human in the loop'. - **Log:** none found: Not stated. - **Off switch:** none found: Not stated. - **Sensitive fields:** none found: Not stated on the AI pages read. - **MCP:** live for all customers, read and write (OAuth scopes per tool): Official MCP server over HTTP with read and write tools, OAuth scopes checked per call (developer docs, read 2026-11-10). - **AI governance:** none found: Not named. ## The 12 capabilities for Deel, with the evidence | Capability | Found | Evidence | Source | | --- | --- | --- | --- | | [Answers employee questions from policy](/capabilities/answers_policy/) | yes | Ask workforce data a question in plain language; asking Deel questions live in Slack. | [www.deel.com/hr/](https://www.deel.com/hr/) | | [Drafts changes](/capabilities/drafts_changes/) | yes | Draft a job description, policy or training course in minutes. | [www.deel.com/hr/](https://www.deel.com/hr/) | | [Runs onboarding end to end](/capabilities/runs_onboarding/) | none found | Onboarding workflows exist; no agent described. | [www.deel.com/hr/](https://www.deel.com/hr/) | | [Runs a compensation cycle](/capabilities/runs_comp_cycle/) | none found | Not described. | [www.deel.com/hr/compensation-management/](https://www.deel.com/hr/compensation-management/) | | [Runs offboarding](/capabilities/runs_offboarding/) | none found | Not described. | [www.deel.com/hr/](https://www.deel.com/hr/) | | [Acts in other systems (Slack, Teams, email, ATS, payroll)](/capabilities/acts_other_systems/) | partial | Slack plugin answers questions; Teams next. | [www.deel.com/hr/plugin/](https://www.deel.com/hr/plugin/) | | [Approval before anything goes live](/capabilities/approval_before_live/) | partial | AI terms: 'Always keep a human in the loop'. | [www.deel.com/legal/ai-terms-of-service/](https://www.deel.com/legal/ai-terms-of-service/) | | [Audit log of agent actions](/capabilities/agent_audit_log/) | none found | Not stated. | [www.deel.com/security/](https://www.deel.com/security/) | | [Off switch for AI features](/capabilities/off_switch/) | none found | Not stated. | [www.deel.com/hr/](https://www.deel.com/hr/) | | [Sensitive-field controls](/capabilities/sensitive_controls/) | none found | Not stated on the AI pages read. | [www.deel.com/hr/](https://www.deel.com/hr/) | | [Official MCP server or connector, with access level](/capabilities/mcp_official/) | yes | Official MCP server over HTTP with read and write tools, OAuth scopes checked per call (developer docs, read 2026-11-10). | [developer.deel.com/mcp/introduction](https://developer.deel.com/mcp/introduction) | | [AI governance certification held or in progress (ISO 42001)](/capabilities/ai_governance_cert/) | none found | Not named. | [www.deel.com/security/](https://www.deel.com/security/) | Read 8 Nov 2026. A cell says "none found" about the pages read on that date, never about the product. ## In the Agent Diary - [Deel: employee questions from policy](/agents/employee-questions/deel/): runs the process - [Deel MCP: status, access and what it does](/mcp/deel/) ## The seven criteria for Deel | Criterion | Weight | Score | Why | | --- | --- | --- | --- | | Build without the vendor | 15 | 3 | A no-code workflow builder with triggers, filters, conditions, delay steps and a simulation tool is described on a marketing page, but building by description and sharing are not documented. | | Agents that act, under control | 30 | 5 | Agentic workflows from triggers with human approval and logged actions cover about as many elements as another platform at 5, also on one marketing page, so 4 understates it. | | Openness to the AI stack | 25 | 8 | A documented official MCP server with read and write OAuth scopes checked per call, signed webhooks with retries, REST API and named AI clients, with gaps on plan detail and audit. | | Time to value | 5 | 6 | A weeks-not-months claim, described implementation steps with an onboarding manager and a documented API sandbox, but no fee and the claim is marketing only. | | Pricing transparency | 5 | 6 | Core from $5 per employee per month, contents of three tiers, and EOR, PEO, contractor, payroll and IT prices are published with month-to-month terms, but Advanced, Elite and recruiting prices are not. | | Fit to the organization's shape | 5 | 6 | Multi-country entities, job architecture, country time off policies and approvals plus payment in 120+ currencies are documented, but no custom fields, roles or per-policy approvers. | | Trust | 15 | 6 | SOC 2 Type II, ISO 27001, ISO 42001, GDPR alignment and daily backups are stated on marketing pages only; no audit log or customer export found. | Weighted total 5.8, rank 8 of 14 on Best AI Agents for HR. ## Breadth, reported and not weighted - **Modules:** 10: Deel HR, Deel Engage, Deel Payroll, EOR, Contractors, Deel IT, Benefits, Hire, Compensation, Workforce planning - **Native payroll:** global payroll and EOR in 150 countries - **ATS:** Deel Hire - **IT and devices:** Deel IT - **Integrations:** not stated as a number on the pages read - **Countries:** 150 countries (EOR) ## Where it is strongest, and where it is weak EOR and payroll in 150 countries with an HRIS that starts at $5 per employee per month. Its agent workflows with approval steps are described only on marketing pages, and no off switch or field-level control for agents was found on 10 Nov 2026. ## Best for A company whose first problem is employing people in 150 countries and whose AI need is drafting and answers. **Where Shapes is weak.** Shapes is first on this site's ranking at 8.6. Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Deel: www.deel.com/hr/: https://www.deel.com/hr/ (read 10 Nov 2026) - Deel: www.deel.com/hr/workforce-planning/: https://www.deel.com/hr/workforce-planning/ (read 10 Nov 2026) - Deel: www.deel.com/legal/ai-terms-of-service/: https://www.deel.com/legal/ai-terms-of-service/ (read 10 Nov 2026) - Deel: www.deel.com/hr/plugin/: https://www.deel.com/hr/plugin/ (read 10 Nov 2026) - Deel: developer.deel.com/api/stable/sandbox.md: https://developer.deel.com/api/stable/sandbox.md (read 10 Nov 2026) - Deel: www.deel.com/pricing/: https://www.deel.com/pricing/ (read 10 Nov 2026) - Deel: www.deel.com/hr/engage/: https://www.deel.com/hr/engage/ (read 10 Nov 2026) - Deel: www.deel.com/security/: https://www.deel.com/security/ (read 10 Nov 2026) - Deel: www.deel.com/hr-platform/ai/: https://www.deel.com/hr-platform/ai/ (read 10 Nov 2026) - Deel: www.deel.com/hr/: https://www.deel.com/hr/ (read 8 Nov 2026) - Deel: www.deel.com/hr/compensation-management/: https://www.deel.com/hr/compensation-management/ (read 8 Nov 2026) - Deel: www.deel.com/hr/plugin/: https://www.deel.com/hr/plugin/ (read 8 Nov 2026) - Deel: www.deel.com/legal/ai-terms-of-service/: https://www.deel.com/legal/ai-terms-of-service/ (read 8 Nov 2026) - Deel: www.deel.com/security/: https://www.deel.com/security/ (read 8 Nov 2026) - Deel: developer.deel.com/mcp/introduction: https://developer.deel.com/mcp/introduction (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/platforms/deel/ --- # Lattice's agents: reviews drafted and submitted over MCP, with the controls not stated Agents, rank 9 of 14, read 8 Nov 2026 **The short answer.** Lattice scores 5 on agents and ranks ninth at 5.8. The AI Agent coaches, drafts reviews, summarizes 1:1s and answers help-desk questions, and the MCP server, labeled New, drafts, edits and submits reviews with the app's permissions. No onboarding or comp-cycle agent, approval step, agent audit log or off switch was found; admins activate the agent in a setup experience. Lattice is the mirror of BambooHR: strong on what the agent does to performance data and silent on the controls. The trust center names no agent log; the July 2026 product update describes activation through setup and role access left to the system's admin, which the matrix records as a partial off switch. The HRIS is in limited US beta, so the agent acts on performance data rather than the system of record for most customers. - **Acts on:** onboarding a hiring wave: none found; a compensation cycle: none found; employee questions from policy: yes - **Approval model:** none found: Not stated for the agent. - **Log:** none found: Not stated. - **Off switch:** partial: Admins activate the AI Agent in a setup experience; role access left to the AI system's admin. - **Sensitive fields:** partial: Answers scoped to each user's Lattice access. - **MCP:** live for all customers, read and write (reviews, updates): Lattice MCP server, labelled New. - **AI governance:** none found: Not named. ## The 12 capabilities for Lattice, with the evidence | Capability | Found | Evidence | Source | | --- | --- | --- | --- | | [Answers employee questions from policy](/capabilities/answers_policy/) | yes | HR Help Desk answers benefits, policy and PTO questions. | [lattice.com/platform/ai-agent/hr-help-desk](https://lattice.com/platform/ai-agent/hr-help-desk) | | [Drafts changes](/capabilities/drafts_changes/) | yes | Drafts evidence-backed reviews; MCP drafts, edits and submits reviews. | [lattice.com/platform/mcp](https://lattice.com/platform/mcp) | | [Runs onboarding end to end](/capabilities/runs_onboarding/) | none found | Not described. | [lattice.com/platform/ai-agent](https://lattice.com/platform/ai-agent) | | [Runs a compensation cycle](/capabilities/runs_comp_cycle/) | none found | Compensation cycle management exists; no agent described. | [lattice.com/platform/compensation/cycle-manageme](https://lattice.com/platform/compensation/cycle-management) | | [Runs offboarding](/capabilities/runs_offboarding/) | none found | Not described. | [lattice.com/platform/ai-agent](https://lattice.com/platform/ai-agent) | | [Acts in other systems (Slack, Teams, email, ATS, payroll)](/capabilities/acts_other_systems/) | partial | Slack, ChatGPT and Codex resources; Zoom, Meet, Teams in 1:1s. | [lattice.com/platform/mcp](https://lattice.com/platform/mcp) | | [Approval before anything goes live](/capabilities/approval_before_live/) | none found | Not stated for the agent. | [lattice.com/platform/ai-agent](https://lattice.com/platform/ai-agent) | | [Audit log of agent actions](/capabilities/agent_audit_log/) | none found | Not stated. | [trustcenter.lattice.com/](https://trustcenter.lattice.com/) | | [Off switch for AI features](/capabilities/off_switch/) | partial | Admins activate the AI Agent in a setup experience; role access left to the AI system's admin. | [lattice.com/blog/july-2026-product-updates](https://lattice.com/blog/july-2026-product-updates) | | [Sensitive-field controls](/capabilities/sensitive_controls/) | partial | Answers scoped to each user's Lattice access. | [lattice.com/platform/mcp](https://lattice.com/platform/mcp) | | [Official MCP server or connector, with access level](/capabilities/mcp_official/) | yes | Lattice MCP server, labelled New. | [lattice.com/platform/mcp](https://lattice.com/platform/mcp) | | [AI governance certification held or in progress (ISO 42001)](/capabilities/ai_governance_cert/) | none found | Not named. | [trustcenter.lattice.com/](https://trustcenter.lattice.com/) | Read 8 Nov 2026. A cell says "none found" about the pages read on that date, never about the product. ## In the Agent Diary - [Lattice: employee questions from policy](/agents/employee-questions/lattice/): runs the process - [Lattice MCP: status, access and what it does](/mcp/lattice/) ## The seven criteria for Lattice | Criterion | Weight | Score | Why | | --- | --- | --- | --- | | Build without the vendor | 15 | 4 | Docs show admin-built custom reports shared with chosen employees and scheduled exports, adding to limited-beta AI and review rules, but there is no prompt-built report or flow. | | Agents that act, under control | 30 | 5 | Docs show a full AI off switch, opt-in per feature, draft review before actions, sensitive-field roles and an agent dashboard log with gaps, but the agent takes single actions and runs no whole process. | | Openness to the AI stack | 25 | 7 | A documented MCP server with read and write scopes, same permissions, a Claude directory connector with a read-only option, a public API and SCIM, but no webhooks and API write-back limited to goals. | | Time to value | 5 | 7 | A sandbox on request adds the one element the current reason lacked, which puts it level with or above another platform at 7 with no sandbox. | | Pricing transparency | 5 | 7 | Per-seat prices for each talent module and add-on and a $4,000 minimum annual agreement are published, but the HRIS is unpriced, enterprise is by quote and in-product billing is not shown. | | Fit to the organization's shape | 5 | 6 | Custom fields, targeted custom roles with per-field permissions, 24 languages and multi-currency compensation cycles are documented, but there is no multi-entity or per-office policy and the HRIS is a US limited beta. | | Trust | 15 | 6 | SOC 2 Type II, GDPR, CCPA, audit logging and backups listed, CSV export and deletion terms documented; the audit log detail covers review cycles only and there is no ISO 27001 or ISO 42001. | Weighted total 5.8, rank 9 of 14 on Best AI Agents for HR. ## Breadth, reported and not weighted - **Modules:** 7: Performance, Goals and OKRs, Engagement, Compensation, Grow, Analytics, HRIS (limited beta, US) - **Native payroll:** none - **ATS:** none - **IT and devices:** none - **Integrations:** not stated as a number - **Countries:** HRIS US-only (limited beta) ## Where it is strongest, and where it is weak Deeper performance tooling and an MCP server that drafts and submits reviews from Claude or ChatGPT. The HRIS is in limited beta for US companies and is not on the pricing page. ## Best for A US company that wants reviews drafted and submitted from Claude or ChatGPT and will set its own approval rule. **Where Shapes is weak.** Shapes is first on this site's ranking at 8.6. Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Lattice: lattice.com/platform/ai-agent: https://lattice.com/platform/ai-agent (read 10 Nov 2026) - Lattice: backup.lattice.com/hris/report-builder.html: https://backup.lattice.com/hris/report-builder.html (read 10 Nov 2026) - Lattice: lattice.com/platform/ai-agent/hr-help-desk: https://lattice.com/platform/ai-agent/hr-help-desk (read 10 Nov 2026) - Lattice: lattice.com/platform/mcp: https://lattice.com/platform/mcp (read 10 Nov 2026) - Lattice: help.lattice.com/en-us/articles/15878584-lattice-mcp-security-and-permissions: https://help.lattice.com/en-us/articles/15878584-lattice-mcp-security-and-permissions (read 10 Nov 2026) - Lattice: lattice.com/pricing: https://lattice.com/pricing (read 10 Nov 2026) - Lattice: lattice.com/blog/july-2026-product-updates: https://lattice.com/blog/july-2026-product-updates (read 10 Nov 2026) - Lattice: backup.lattice.com/pricing.html: https://backup.lattice.com/pricing.html (read 10 Nov 2026) - Lattice: lattice.com/blog/september-2026-product-updates: https://lattice.com/blog/september-2026-product-updates (read 10 Nov 2026) - Lattice: trustcenter.lattice.com/: https://trustcenter.lattice.com/ (read 10 Nov 2026) - Lattice: lattice.com/hris: https://lattice.com/hris (read 10 Nov 2026) - Lattice: lattice.com/platform/ai-agent/hr-help-desk: https://lattice.com/platform/ai-agent/hr-help-desk (read 8 Nov 2026) - Lattice: lattice.com/platform/mcp: https://lattice.com/platform/mcp (read 8 Nov 2026) - Lattice: lattice.com/platform/ai-agent: https://lattice.com/platform/ai-agent (read 8 Nov 2026) - Lattice: lattice.com/platform/compensation/cycle-management: https://lattice.com/platform/compensation/cycle-management (read 8 Nov 2026) - Lattice: trustcenter.lattice.com/: https://trustcenter.lattice.com/ (read 8 Nov 2026) - Lattice: lattice.com/blog/july-2026-product-updates: https://lattice.com/blog/july-2026-product-updates (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/platforms/lattice/ --- # Personio's agents: a screening agent with a human decision, an AI Transformation Hub, and an MCP that is announced Agents, rank 10 of 14, read 8 Nov 2026 **The short answer.** Personio scores 4 on agents and ranks tenth at 5.4. A Talent Screening Agent shortlists with 'a person always makes the decision', the AI Transformation Hub gives every agent a named owner and an approval status, and the Analytics Agent builds reports from a description. No agent for onboarding, comp cycles or employee questions was found; the MCP is announced. Personio's governance story is organizational rather than technical: the Hub names an owner and a status per agent, which the matrix records as a partial approval step and no audit log. AI Studio and inclusive AI credits are CoreEnterprise-only. The MCP status is 'announced': a card with no status page, and live use goes through third-party wrappers, which the method does not accept as a status. - **Acts on:** onboarding a hiring wave: none found; a compensation cycle: none found; employee questions from policy: yes - **Approval model:** partial: 'A person always makes the decision' (screening); every agent gets an approval status in the AI Transformation Hub. - **Log:** none found: Not stated. - **Off switch:** none found: Not stated. - **Sensitive fields:** partial: Unlimited access and permissions on CorePro. - **MCP:** beta or early access, read, limited write off by default: Official MCP server in open beta on Core Pro: read access, limited write that is off by default; named clients Claude, ChatGPT, VS Code with GitHub Copilot (help center, read 2026-11-10). - **AI governance:** none found: Responsible AI principles; ISO 42001 not named. ## The 12 capabilities for Personio, with the evidence | Capability | Found | Evidence | Source | | --- | --- | --- | --- | | [Answers employee questions from policy](/capabilities/answers_policy/) | yes | Personio Assistant answers people questions. | [www.personio.com/product/assistant/](https://www.personio.com/product/assistant/) | | [Drafts changes](/capabilities/drafts_changes/) | partial | Analytics Agent builds reports from a description; screening agent shortlists. | [www.personio.com/whats-new/](https://www.personio.com/whats-new/) | | [Runs onboarding end to end](/capabilities/runs_onboarding/) | none found | Onboarding workflows exist; no agent described. | [www.personio.com/product/onboarding/](https://www.personio.com/product/onboarding/) | | [Runs a compensation cycle](/capabilities/runs_comp_cycle/) | none found | Not described. | [www.personio.com/whats-new/](https://www.personio.com/whats-new/) | | [Runs offboarding](/capabilities/runs_offboarding/) | none found | Not described. | [www.personio.com/whats-new/](https://www.personio.com/whats-new/) | | [Acts in other systems (Slack, Teams, email, ATS, payroll)](/capabilities/acts_other_systems/) | none found | Not described on the pages read. | [www.personio.com/whats-new/](https://www.personio.com/whats-new/) | | [Approval before anything goes live](/capabilities/approval_before_live/) | partial | 'A person always makes the decision' (screening); every agent gets an approval status in the AI Transformation Hub. | [www.personio.com/whats-new/](https://www.personio.com/whats-new/) | | [Audit log of agent actions](/capabilities/agent_audit_log/) | none found | Not stated. | [www.personio.com/security/](https://www.personio.com/security/) | | [Off switch for AI features](/capabilities/off_switch/) | none found | Not stated. | [www.personio.com/whats-new/](https://www.personio.com/whats-new/) | | [Sensitive-field controls](/capabilities/sensitive_controls/) | partial | Unlimited access and permissions on CorePro. | [www.personio.com/pricing/](https://www.personio.com/pricing/) | | [Official MCP server or connector, with access level](/capabilities/mcp_official/) | partial | Official MCP server in open beta on Core Pro: read access, limited write that is off by default; named clients Claude, ChatGPT, VS Code with GitHub Copilot (help center, read 2026-11-10). | [support.personio.de/hc/en-us/articles/3831465667](https://support.personio.de/hc/en-us/articles/38314656674333) | | [AI governance certification held or in progress (ISO 42001)](/capabilities/ai_governance_cert/) | none found | Responsible AI principles; ISO 42001 not named. | [www.personio.com/security/](https://www.personio.com/security/) | Read 8 Nov 2026. A cell says "none found" about the pages read on that date, never about the product. ## In the Agent Diary - [Personio: employee questions from policy](/agents/employee-questions/personio/): runs the process - [Personio MCP: status, access and what it does](/mcp/personio/) ## The seven criteria for Personio | Criterion | Weight | Score | Why | | --- | --- | --- | --- | | Build without the vendor | 15 | 6 | Docs show a no-code workflow builder with templates, conditions and draft or activate, and a Report Builder with permission-respecting data, but natural-language workflow building and a library of built items are not documented. | | Agents that act, under control | 30 | 4 | Docs add central and per-feature AI off switches with a spend limit, advisory-only screening, audited MCP setting changes and hidden person attributes, but no process-running agent or agent audit log. | | Openness to the AI stack | 25 | 6 | A documented official MCP server (beta, Core Pro) with read, opt-in write, per-tool switches and named clients, a webhooks API and Slack and Teams integrations, but beta status and the Core Pro API restriction limit it. | | Time to value | 5 | 6 | A guided Implementation Specialist, phased payroll implementation, a 14-day trial, import tools and a one-time fee are documented, but no timeline or amount and the sandbox is on a paid support plan. | | Pricing transparency | 5 | 3 | Three plans with contents, named add-ons and a 12-month minimum run-time are published, but no prices and no implementation amount (a third-party figure is only a weak hint). | | Fit to the organization's shape | 5 | 8 | Unlimited entities, permissions, custom attributes, per-process approvers, per-workplace currency and per-user language are documented in docs, with policies per office not shown. | | Trust | 15 | 6 | ISO 27001, GDPR, a documented uneditable audit log with CSV export, company data export and retention policies are documented; no SOC 2, ISO 42001 or backup detail. | Weighted total 5.4, rank 10 of 14 on Best AI Agents for HR. ## Breadth, reported and not weighted - **Modules:** 10: Core HR, Recruiting, Performance and development, Surveys, Compensation management, Payroll (DE and others), Absence, Attendance, Workforce planning, Whistleblowing - **Native payroll:** Personio Payroll (Germany and selected EU countries) - **ATS:** Recruiting add-on - **IT and devices:** none found - **Integrations:** marketplace (count not stated) - **Countries:** EU-centric ## Where it is strongest, and where it is weak Payroll in Germany and a deep EU compliance footprint with unlimited legal entities on CorePro. No published per-seat price, a one-time setup fee by company size, a 12-month minimum, and an MCP server still in open beta on Core Pro. ## Best for An EU company with German payroll whose AI need is screening and reporting, and whose governance is a named owner per agent. **Where Shapes is weak.** Shapes is first on this site's ranking at 8.6. Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Personio: www.personio.com/whats-new/: https://www.personio.com/whats-new/ (read 10 Nov 2026) - Personio: www.personio.com/pricing/: https://www.personio.com/pricing/ (read 10 Nov 2026) - Personio: www.personio.com/product/ai-transformation/: https://www.personio.com/product/ai-transformation/ (read 10 Nov 2026) - Personio: www.personio.com/faq/contract-pricing/: https://www.personio.com/faq/contract-pricing/ (read 10 Nov 2026) - Personio: support.personio.de/hc/en-us/articles/38318101338013: https://support.personio.de/hc/en-us/articles/38318101338013 (read 10 Nov 2026) - Personio: www.compono.com/articles/hr-software-pricing-guide: https://www.compono.com/articles/hr-software-pricing-guide (read 10 Nov 2026) - Personio: www.personio.com/product/workflow-automation/: https://www.personio.com/product/workflow-automation/ (read 10 Nov 2026) - Personio: www.personio.com/security/: https://www.personio.com/security/ (read 10 Nov 2026) - Personio: www.personio.com/product/assistant/: https://www.personio.com/product/assistant/ (read 8 Nov 2026) - Personio: www.personio.com/whats-new/: https://www.personio.com/whats-new/ (read 8 Nov 2026) - Personio: www.personio.com/product/onboarding/: https://www.personio.com/product/onboarding/ (read 8 Nov 2026) - Personio: www.personio.com/security/: https://www.personio.com/security/ (read 8 Nov 2026) - Personio: www.personio.com/pricing/: https://www.personio.com/pricing/ (read 8 Nov 2026) - Personio: support.personio.de/hc/en-us/articles/38314656674333: https://support.personio.de/hc/en-us/articles/38314656674333 (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/platforms/personio/ --- # Gusto's agents: no HR agent, and an MCP server where every write comes back for confirmation Agents, rank 11 of 14, read 8 Nov 2026 **The short answer.** Gusto scores 4 on agents and ranks eleventh at 4.3. No HR agent was found; the Premium plan's 'Ask an HR Expert' answers are reviewed by a human; the MCP server, with 31,000 companies connected, returns every write for confirmation, lets the admin choose data categories, and tells users to disable automatic tool execution. Gusto's control story lives entirely in the MCP page: 'every write comes back to you for confirmation before anything moves' is the clearest confirmation sentence on the card, and the data-category scoping is a partial off switch. The agent columns are partial or none found because the product is payroll first. The diary records Gusto as partial on onboarding (the MCP onboards employees or contractors with confirmation). - **Acts on:** onboarding a hiring wave: partial; a compensation cycle: none found; employee questions from policy: partial - **Approval model:** yes: 'Every write comes back to you for confirmation before anything moves.' - **Log:** none found: No audit statement on the MCP page. - **Off switch:** partial: Admin chooses data categories; users told to disable automatic tool execution. - **Sensitive fields:** partial: Data categories chosen per connection. - **MCP:** live for all customers, read and write with confirmation: mcp.api.gusto.com, OAuth2, 31,000+ companies connected. - **AI governance:** none found: Not named. ## The 12 capabilities for Gusto, with the evidence | Capability | Found | Evidence | Source | | --- | --- | --- | --- | | [Answers employee questions from policy](/capabilities/answers_policy/) | partial | Ask an HR Expert on Premium, human-reviewed; AI-driven suggestions. | [gusto.com/pricing](https://gusto.com/pricing) | | [Drafts changes](/capabilities/drafts_changes/) | partial | MCP can log time, onboard employees, prepare payroll. | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | [Runs onboarding end to end](/capabilities/runs_onboarding/) | partial | MCP onboards employees or contractors with confirmation. | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | [Runs a compensation cycle](/capabilities/runs_comp_cycle/) | none found | Not described. | [gusto.com/pricing](https://gusto.com/pricing) | | [Runs offboarding](/capabilities/runs_offboarding/) | none found | Not described. | [gusto.com/pricing](https://gusto.com/pricing) | | [Acts in other systems (Slack, Teams, email, ATS, payroll)](/capabilities/acts_other_systems/) | partial | Slack connector. | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | [Approval before anything goes live](/capabilities/approval_before_live/) | yes | 'Every write comes back to you for confirmation before anything moves.' | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | [Audit log of agent actions](/capabilities/agent_audit_log/) | none found | No audit statement on the MCP page. | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | [Off switch for AI features](/capabilities/off_switch/) | partial | Admin chooses data categories; users told to disable automatic tool execution. | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | [Sensitive-field controls](/capabilities/sensitive_controls/) | partial | Data categories chosen per connection. | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | [Official MCP server or connector, with access level](/capabilities/mcp_official/) | yes | mcp.api.gusto.com, OAuth2, 31,000+ companies connected. | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | [AI governance certification held or in progress (ISO 42001)](/capabilities/ai_governance_cert/) | none found | Not named. | [gusto.com/security](https://gusto.com/security) | Read 8 Nov 2026. A cell says "none found" about the pages read on that date, never about the product. ## In the Agent Diary - [Gusto: onboarding a hiring wave](/agents/onboarding/gusto/): runs part of it - [Gusto: employee questions from policy](/agents/employee-questions/gusto/): runs part of it - [Gusto MCP: status, access and what it does](/mcp/gusto/) ## The seven criteria for Gusto | Criterion | Weight | Score | Why | | --- | --- | --- | --- | | Build without the vendor | 15 | 1 | Reports and AI suggestions with onboarding templates and employee tasks, but no natural-language builder, any-field triggers or conditions found. | | Agents that act, under control | 30 | 4 | MCP writes come back for confirmation and admins choose data categories, but there is no HR agent running processes, no audit statement and only partial off-switch controls. | | Openness to the AI stack | 25 | 6 | A documented MCP server for customers exists, but the public API and webhooks are restricted to partners and the developer MCP server is beta. | | Time to value | 5 | 7 | No setup fee, guided setup included and self-serve are documented across two sources, but no timeline or sandbox. | | Pricing transparency | 5 | 9 | Base and per-person prices for three plans, priced add-ons and month-to-month terms with no plan minimum are published; in-product billing is not documented. | | Fit to the organization's shape | 5 | 3 | Custom fields through the API, time off approvals and report filters by department and address are documented, with no multi-entity, currency or language detail. | | Trust | 15 | 3 | SOC 1 and SOC 2 reports on request on one page; no ISO, GDPR, audit log, backup or export. | Weighted total 4.3, rank 11 of 14 on Best AI Agents for HR. ## Breadth, reported and not weighted - **Modules:** 5: Payroll, Benefits, Time tracking, HR guidance (Premium), Hiring and onboarding basics - **Native payroll:** US payroll - **ATS:** basic hiring tools; partners for ATS - **IT and devices:** partners - **Integrations:** marketplace by category - **Countries:** US employees; global contractor payments ## Where it is strongest, and where it is weak US payroll and benefits with every price published and an MCP server 31,000 companies have connected. A payroll-first product with HR tools only on Premium, no HR agent and no multi-country HR. ## Best for A US company whose AI use is an outside client writing payroll changes with a confirmation on each. **Where Shapes is weak.** Shapes is first on this site's ranking at 8.6. Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Gusto: gusto.com/pricing: https://gusto.com/pricing (read 10 Nov 2026) - Gusto: gusto.com/product/hr: https://gusto.com/product/hr (read 10 Nov 2026) - Gusto: gusto.com/product/integrations/gusto-mcp: https://gusto.com/product/integrations/gusto-mcp (read 10 Nov 2026) - Gusto: gusto.com/product/compare/gusto-vs-quickbooks-payroll: https://gusto.com/product/compare/gusto-vs-quickbooks-payroll (read 10 Nov 2026) - Gusto: mcp.api.gusto.com;: https://mcp.api.gusto.com; (read 10 Nov 2026) - Gusto: gusto.com/security: https://gusto.com/security (read 10 Nov 2026) - Gusto: gusto.com/pricing: https://gusto.com/pricing (read 8 Nov 2026) - Gusto: gusto.com/product/integrations/gusto-mcp: https://gusto.com/product/integrations/gusto-mcp (read 8 Nov 2026) - Gusto: gusto.com/security: https://gusto.com/security (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/platforms/gusto/ --- # Paylocity's agents: proactive assistance in the flow of work, and no controls stated Agents, rank 12 of 14, read 8 Nov 2026 **The short answer.** Paylocity scores 4 on agents and ranks twelfth at 3.6. 'Proactive assistance right into your flow of work' and AI-powered touchless reports are described; onboarding is partial through the Workflows engine; no agent that runs a process, approval model, agent audit log, off switch or MCP server was found on 8 Nov 2026. Paylocity's row is partial on answers, drafts and onboarding and none found on the controls; API libraries are the route for outside tools. The diary records Paylocity as partial on onboarding through its self-serve Workflows automation. - **Acts on:** onboarding a hiring wave: partial; a compensation cycle: none found; employee questions from policy: partial - **Approval model:** none found: Not found. - **Log:** none found: Not found. - **Off switch:** none found: Not found. - **Sensitive fields:** partial: Role-based security in the HCM. - **MCP:** none found: No MCP server found; API libraries. - **AI governance:** none found: Not named. ## The 12 capabilities for Paylocity, with the evidence | Capability | Found | Evidence | Source | | --- | --- | --- | --- | | [Answers employee questions from policy](/capabilities/answers_policy/) | partial | 'Proactive assistance right into your flow of work'; on-demand HR resource library. | [www.paylocity.com/pricing/](https://www.paylocity.com/pricing/) | | [Drafts changes](/capabilities/drafts_changes/) | partial | AI-powered touchless reports and contract insights. | [www.paylocity.com/pricing/](https://www.paylocity.com/pricing/) | | [Runs onboarding end to end](/capabilities/runs_onboarding/) | partial | Workflows self-serve automation engine; onboarding module. | [www.paylocity.com/products/hr/talent-management-](https://www.paylocity.com/products/hr/talent-management-system/onboarding/) | | [Runs a compensation cycle](/capabilities/runs_comp_cycle/) | none found | Compensation module; no agent. | [www.paylocity.com/products/hr/talent-management-](https://www.paylocity.com/products/hr/talent-management-system/compensation-management/) | | [Runs offboarding](/capabilities/runs_offboarding/) | none found | Not found. | [www.paylocity.com/products/hr/](https://www.paylocity.com/products/hr/) | | [Acts in other systems (Slack, Teams, email, ATS, payroll)](/capabilities/acts_other_systems/) | none found | Not found. | [www.paylocity.com/products/hr/](https://www.paylocity.com/products/hr/) | | [Approval before anything goes live](/capabilities/approval_before_live/) | none found | Not found. | [www.paylocity.com/products/hr/](https://www.paylocity.com/products/hr/) | | [Audit log of agent actions](/capabilities/agent_audit_log/) | none found | Not found. | [trust.paylocity.com/](https://trust.paylocity.com/) | | [Off switch for AI features](/capabilities/off_switch/) | none found | Not found. | [www.paylocity.com/products/hr/](https://www.paylocity.com/products/hr/) | | [Sensitive-field controls](/capabilities/sensitive_controls/) | partial | Role-based security in the HCM. | [www.paylocity.com/products/hr/](https://www.paylocity.com/products/hr/) | | [Official MCP server or connector, with access level](/capabilities/mcp_official/) | none found | No MCP server found; API libraries. | [www.paylocity.com/pricing/](https://www.paylocity.com/pricing/) | | [AI governance certification held or in progress (ISO 42001)](/capabilities/ai_governance_cert/) | none found | Not named. | [trust.paylocity.com/](https://trust.paylocity.com/) | Read 8 Nov 2026. A cell says "none found" about the pages read on that date, never about the product. ## In the Agent Diary - [Paylocity: onboarding a hiring wave](/agents/onboarding/paylocity/): runs part of it - [Paylocity: employee questions from policy](/agents/employee-questions/paylocity/): runs part of it - [Paylocity MCP: status, access and what it does](/mcp/paylocity/) ## The seven criteria for Paylocity | Criterion | Weight | Score | Why | | --- | --- | --- | --- | | Build without the vendor | 15 | 3 | Ready-made and custom workflow templates with filters, conditional routing, manager creation rights and progress dashboards are described on a marketing page, but no natural-language builder is documented. | | Agents that act, under control | 30 | 4 | Purpose-built agents for onboarding, payroll and expenses with an enable choice and monitoring of actions are described on a marketing page, but no approval step, audit log or field-level control is stated. | | Openness to the AI stack | 25 | 3 | A documented MCP server exists but only for API documentation, and webhooks and developer access are limited to early adopters, partners or custom-integration agreements. | | Time to value | 5 | 2 | A 4 to 16 week integration estimate and a sandbox for approved customers and partners are documented, but nothing on HR platform go-live, fee or guided implementation. | | Pricing transparency | 5 | 2 | Contents are listed by module, but there are no prices, tiers or minimum, only a request for pricing through a demo. | | Fit to the organization's shape | 5 | 5 | Unlimited custom fields, location-based workflow filters and global payroll currencies are described, but only by marketing pages, with no entity or role detail. | | Trust | 15 | 5 | ISO 27001, SOC 1 and SOC 2 Type II plus a DPA are listed on marketing pages and audit trail reports are mentioned; no audit log detail, ISO 42001, backup or export. | Weighted total 3.6, rank 12 of 14 on Best AI Agents for HR. ## Breadth, reported and not weighted - **Modules:** 11: Payroll, HR, Benefits, Time and labor, Recruiting, Onboarding, Performance, Learning, Compensation, Finance, IT - **Native payroll:** US payroll and global payroll in named countries - **ATS:** Recruiting - **IT and devices:** IT module - **Integrations:** API libraries and marketplace - **Countries:** US plus global payroll countries ## Where it is strongest, and where it is weak Payroll, finance and IT modules under one contract with ISO 27001 and SOC 2 Type II. No published price; its onboarding and payroll agents are described on a marketing page with no approval step or audit log stated, and no natural-language builder was found. ## Best for A US company that wants HR, finance and IT modules under one contract and an assistant rather than an agent. **Where Shapes is weak.** Shapes is first on this site's ranking at 8.6. Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Paylocity: www.paylocity.com/products/hr/: https://www.paylocity.com/products/hr/ (read 10 Nov 2026) - Paylocity: www.paylocity.com/pricing/: https://www.paylocity.com/pricing/ (read 10 Nov 2026) - Paylocity: www.paylocity.com/products/hr/talent-management-system/onboarding/: https://www.paylocity.com/products/hr/talent-management-system/onboarding/ (read 10 Nov 2026) - Paylocity: developer.paylocity.com/integrations/docs/integration-assistant-mcp.md: https://developer.paylocity.com/integrations/docs/integration-assistant-mcp.md (read 10 Nov 2026) - Paylocity: developer.paylocity.com/integrations/docs/integrations-faq.md: https://developer.paylocity.com/integrations/docs/integrations-faq.md (read 10 Nov 2026) - Paylocity: www.paylocity.com/products/hr/payroll/global-payroll/: https://www.paylocity.com/products/hr/payroll/global-payroll/ (read 10 Nov 2026) - Paylocity: trust.paylocity.com/: https://trust.paylocity.com/ (read 10 Nov 2026) - Paylocity: www.paylocity.com/pricing/: https://www.paylocity.com/pricing/ (read 8 Nov 2026) - Paylocity: www.paylocity.com/products/hr/talent-management-system/onboarding/: https://www.paylocity.com/products/hr/talent-management-system/onboarding/ (read 8 Nov 2026) - Paylocity: www.paylocity.com/products/hr/talent-management-system/compensation-management/: https://www.paylocity.com/products/hr/talent-management-system/compensation-management/ (read 8 Nov 2026) - Paylocity: www.paylocity.com/products/hr/: https://www.paylocity.com/products/hr/ (read 8 Nov 2026) - Paylocity: trust.paylocity.com/: https://trust.paylocity.com/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/platforms/paylocity/ --- # ADP Workforce Now's agents: ADP Assist answers, and no process agent, log, switch or MCP was found Agents, rank 13 of 14, read 8 Nov 2026 **The short answer.** ADP scores 2 on agents and ranks thirteenth at 2.9. ADP Assist's virtual assistant addresses HR issues and action cards surface analytics; no agent that runs a process, approval model, agent audit log, off switch or MCP server was found on 8 Nov 2026. Trust scores 4 on SOC 1, SOC 2, ISO 27001 and PCI DSS. ADP's row is answers yes, drafts partial, everything else none found, with sensitive controls partial. APIs through the ADP Marketplace are the route for outside tools. The diary has no ADP page. - **Acts on:** onboarding a hiring wave: none found; a compensation cycle: none found; employee questions from policy: yes - **Approval model:** none found: Not found. - **Log:** none found: Not found. - **Off switch:** none found: Not found. - **Sensitive fields:** partial: Role-based access in Workforce Now. - **MCP:** none found: No MCP server found; APIs via ADP Marketplace. - **AI governance:** none found: Not named. ## The 12 capabilities for ADP, with the evidence | Capability | Found | Evidence | Source | | --- | --- | --- | --- | | [Answers employee questions from policy](/capabilities/answers_policy/) | yes | ADP Assist: 'Action cards and a virtual assistant help employees by addressing most HR issues.' | [www.adp.com/what-we-offer/products/adp-workforce](https://www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx) | | [Drafts changes](/capabilities/drafts_changes/) | partial | Action cards surface tasks; no drafting described. | [www.adp.com/what-we-offer/products/adp-workforce](https://www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx) | | [Runs onboarding end to end](/capabilities/runs_onboarding/) | none found | Onboarding module; no agent. | [www.adp.com/what-we-offer/products/adp-workforce](https://www.adp.com/what-we-offer/products/adp-workforce-now/hr-management) | | [Runs a compensation cycle](/capabilities/runs_comp_cycle/) | none found | Not found. | [www.adp.com/what-we-offer/products/adp-workforce](https://www.adp.com/what-we-offer/products/adp-workforce-now/talent) | | [Runs offboarding](/capabilities/runs_offboarding/) | none found | Not found. | [www.adp.com/what-we-offer/products/adp-workforce](https://www.adp.com/what-we-offer/products/adp-workforce-now/hr-management) | | [Acts in other systems (Slack, Teams, email, ATS, payroll)](/capabilities/acts_other_systems/) | none found | Not found. | [www.adp.com/what-we-offer/products/adp-workforce](https://www.adp.com/what-we-offer/products/adp-workforce-now/capabilities) | | [Approval before anything goes live](/capabilities/approval_before_live/) | none found | Not found. | [www.adp.com/what-we-offer/products/adp-workforce](https://www.adp.com/what-we-offer/products/adp-workforce-now/capabilities) | | [Audit log of agent actions](/capabilities/agent_audit_log/) | none found | Not found. | [www.adp.com/what-we-offer/products/adp-workforce](https://www.adp.com/what-we-offer/products/adp-workforce-now/capabilities) | | [Off switch for AI features](/capabilities/off_switch/) | none found | Not found. | [www.adp.com/what-we-offer/products/adp-workforce](https://www.adp.com/what-we-offer/products/adp-workforce-now/capabilities) | | [Sensitive-field controls](/capabilities/sensitive_controls/) | partial | Role-based access in Workforce Now. | [www.adp.com/what-we-offer/products/adp-workforce](https://www.adp.com/what-we-offer/products/adp-workforce-now/capabilities) | | [Official MCP server or connector, with access level](/capabilities/mcp_official/) | none found | No MCP server found; APIs via ADP Marketplace. | [www.adp.com/what-we-offer/products/adp-workforce](https://www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx) | | [AI governance certification held or in progress (ISO 42001)](/capabilities/ai_governance_cert/) | none found | Not named. | [www.adp.com/knowledge/payroll/small-business/sol](https://www.adp.com/knowledge/payroll/small-business/solutions/security) | Read 8 Nov 2026. A cell says "none found" about the pages read on that date, never about the product. ## In the Agent Diary - [ADP: employee questions from policy](/agents/employee-questions/adp/): runs the process - [ADP MCP: status, access and what it does](/mcp/adp/) ## The seven criteria for ADP | Criterion | Weight | Score | Why | | --- | --- | --- | --- | | Build without the vendor | 15 | 3 | Drag and drop report building with sharing and a prebuilt report library are described, plus guided workflows, but only on marketing pages and with no natural-language building or admin-editable triggers. | | Agents that act, under control | 30 | 2 | A virtual assistant answers HR questions and role-based access exists, but no process agent, approval model, audit log or off switch is documented. | | Openness to the AI stack | 25 | 3 | Docs-type developer pages show HR, time and payroll APIs with OAuth 2.0, but API Central is a separate purchase, the MCP mention is marketing only, and no webhooks were found. | | Time to value | 5 | 3 | An implementation team and Client Success Executive from day one are named, with no timeline, fee or sandbox. | | Pricing transparency | 5 | 2 | Three package names with contents are published, but every tier is Get Pricing with no amounts, add-on prices or minimum. | | Fit to the organization's shape | 5 | 4 | Role approvals, time off policies and multi-country support with a customizable portal are described, but mostly by marketing pages and with no custom fields or entity detail. | | Trust | 15 | 4 | SOC 1 and SOC 2, ISO 27001 and PCI DSS on one page; no GDPR, ISO 42001 or audit log. | Weighted total 2.9, rank 13 of 14 on Best AI Agents for HR. ## Breadth, reported and not weighted - **Modules:** 8: Payroll, HR management, Talent, Benefits, Time, Learning, IT management, Insights - **Native payroll:** US payroll with global partners - **ATS:** talent acquisition - **IT and devices:** IT management - **Integrations:** ADP Marketplace - **Countries:** US with global payroll partners ## Where it is strongest, and where it is weak Payroll, tax and benefits depth for US employers with an implementation team from day one. No published price and no agent, MCP server or natural-language builder found. ## Best for A US employer whose AI need is an assistant inside payroll and benefits, with governance handled by contract. **Where Shapes is weak.** Shapes is first on this site's ranking at 8.6. Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - ADP: www.adp.com/what-we-offer/products/adp-workforce-now/insights: https://www.adp.com/what-we-offer/products/adp-workforce-now/insights (read 10 Nov 2026) - ADP: www.adp.com/what-we-offer/products/adp-workforce-now/hr-management: https://www.adp.com/what-we-offer/products/adp-workforce-now/hr-management (read 10 Nov 2026) - ADP: www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx: https://www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx (read 10 Nov 2026) - ADP: www.adp.com/what-we-offer/products/adp-workforce-now/capabilities: https://www.adp.com/what-we-offer/products/adp-workforce-now/capabilities (read 10 Nov 2026) - ADP: developers.adp.com/: https://developers.adp.com/ (read 10 Nov 2026) - ADP: www.adp.com/what-we-offer/products/adp-workforce-now/the-adp-difference: https://www.adp.com/what-we-offer/products/adp-workforce-now/the-adp-difference (read 10 Nov 2026) - ADP: www.adp.com/what-we-offer/products/adp-workforce-now/enterprise: https://www.adp.com/what-we-offer/products/adp-workforce-now/enterprise (read 10 Nov 2026) - ADP: www.adp.com/knowledge/payroll/small-business/solutions/security: https://www.adp.com/knowledge/payroll/small-business/solutions/security (read 10 Nov 2026) - ADP: www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx: https://www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx (read 8 Nov 2026) - ADP: www.adp.com/what-we-offer/products/adp-workforce-now/hr-management: https://www.adp.com/what-we-offer/products/adp-workforce-now/hr-management (read 8 Nov 2026) - ADP: www.adp.com/what-we-offer/products/adp-workforce-now/talent: https://www.adp.com/what-we-offer/products/adp-workforce-now/talent (read 8 Nov 2026) - ADP: www.adp.com/what-we-offer/products/adp-workforce-now/capabilities: https://www.adp.com/what-we-offer/products/adp-workforce-now/capabilities (read 8 Nov 2026) - ADP: www.adp.com/knowledge/payroll/small-business/solutions/security: https://www.adp.com/knowledge/payroll/small-business/solutions/security (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/platforms/adp/ --- # Sage HR's agents: none found on the pages read Agents, rank 14 of 14, read 8 Nov 2026 **The short answer.** Sage HR scores 1 on agents and ranks last at 2.2. No agent, approval model, AI off switch, agent audit log or MCP server was found on the Sage HR pages read on 8 Nov 2026; the matrix row is none found on eleven of twelve columns and partial on sensitive controls. The row is about the pages. Sage HR is the small-business product; Sage People, implemented through partners, is a different product and is not on the matrix. The diary has no Sage page. The one partial cell, sensitive controls, reflects a permission model described on the product pages rather than a field-level AI rule. - **Acts on:** onboarding a hiring wave: none found; a compensation cycle: none found; employee questions from policy: none found - **Approval model:** none found: Not found. - **Log:** none found: Not found. - **Off switch:** none found: Not found. - **Sensitive fields:** partial: Permission levels in Sage HR. - **MCP:** none found: No MCP server found. - **AI governance:** none found: Not named. ## The 12 capabilities for Sage, with the evidence | Capability | Found | Evidence | Source | | --- | --- | --- | --- | | [Answers employee questions from policy](/capabilities/answers_policy/) | none found | Not found on the Sage HR pages read on 2026-11-08. | [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) | | [Drafts changes](/capabilities/drafts_changes/) | none found | Not found. | [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) | | [Runs onboarding end to end](/capabilities/runs_onboarding/) | none found | Onboarding checklists exist; no agent. | [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) | | [Runs a compensation cycle](/capabilities/runs_comp_cycle/) | none found | Not found. | [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) | | [Runs offboarding](/capabilities/runs_offboarding/) | none found | Not found. | [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) | | [Acts in other systems (Slack, Teams, email, ATS, payroll)](/capabilities/acts_other_systems/) | none found | Not found. | [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) | | [Approval before anything goes live](/capabilities/approval_before_live/) | none found | Not found. | [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) | | [Audit log of agent actions](/capabilities/agent_audit_log/) | none found | Not found. | [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) | | [Off switch for AI features](/capabilities/off_switch/) | none found | Not found. | [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) | | [Sensitive-field controls](/capabilities/sensitive_controls/) | partial | Permission levels in Sage HR. | [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) | | [Official MCP server or connector, with access level](/capabilities/mcp_official/) | none found | No MCP server found. | [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) | | [AI governance certification held or in progress (ISO 42001)](/capabilities/ai_governance_cert/) | none found | Not named. | [www.sage.com/en-us/trust-security/security/techn](https://www.sage.com/en-us/trust-security/security/technical/standards-compliance/) | Read 8 Nov 2026. A cell says "none found" about the pages read on that date, never about the product. ## In the Agent Diary No agent for any of the three diary processes was found on Sage's pages on 8 Nov 2026. - [Sage MCP: status, access and what it does](/mcp/sage/) ## The seven criteria for Sage | Criterion | Weight | Score | Why | | --- | --- | --- | --- | | Build without the vendor | 15 | 1 | Reports and checklist templates with employee task lists, but no natural-language builder, triggers or conditions found and only one source. | | Agents that act, under control | 30 | 1 | No agent, approval model, audit log or AI off switch was found; only permission levels are mentioned. | | Openness to the AI stack | 25 | 2 | Only a Sage API was found, with no MCP server, no webhooks and no AI integrations. | | Time to value | 5 | 4 | A 30-day free trial and modular self-serve setup are documented, with implementation through partners and no fee, timeline or sandbox. | | Pricing transparency | 5 | 7 | Per-employee module prices and priced add-ons are published for the US, but no minimum is stated, the enterprise product is by quote and in-product billing is not shown. | | Fit to the organization's shape | 5 | 2 | Leave policies and manager approvals are documented, but multi-entity and multi-country configuration are not shown as available. | | Trust | 15 | 4 | ISO 27001/27017/27018 and SOC 2 Type II from one source; no GDPR, ISO 42001, audit log, backup or export. | Weighted total 2.2, rank 14 of 14 on Best AI Agents for HR. ## Breadth, reported and not weighted - **Modules:** 7: Core HR and leave, Performance, Shift scheduling, Timesheets, Expenses, Recruitment, HR advice - **Native payroll:** Sage Payroll products (separate) - **ATS:** Recruitment module - **IT and devices:** none - **Integrations:** not stated - **Countries:** UK, US, Canada and others ## Where it is strongest, and where it is weak Module-by-module published prices and a 30-day trial for small UK and US teams. No agent, builder or MCP server on the Sage HR pages read on 8 Nov 2026. ## Best for A small UK or US team that wants module prices and a trial and has no agent requirement yet. **Where Shapes is weak.** Shapes is first on this site's ranking at 8.6. Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Sage: www.sage.com/en-gb/sage-business-cloud/hr/product-updates/: https://www.sage.com/en-gb/sage-business-cloud/hr/product-updates/ (read 10 Nov 2026) - Sage: www.sage.com/en-us/sage-business-cloud/hr/: https://www.sage.com/en-us/sage-business-cloud/hr/ (read 10 Nov 2026) - Sage: www.sage.com/en-gb/sage-business-cloud/hr/: https://www.sage.com/en-gb/sage-business-cloud/hr/ (read 10 Nov 2026) - Sage: www.sage.com/en-us/sage-business-cloud/hr/pricing/: https://www.sage.com/en-us/sage-business-cloud/hr/pricing/ (read 10 Nov 2026) - Sage: www.sage.com/en-us/trust-security/security/technical/standards-compliance/: https://www.sage.com/en-us/trust-security/security/technical/standards-compliance/ (read 10 Nov 2026) - Sage: www.sage.com/en-us/sage-business-cloud/hr/: https://www.sage.com/en-us/sage-business-cloud/hr/ (read 8 Nov 2026) - Sage: www.sage.com/en-us/trust-security/security/technical/standards-compliance/: https://www.sage.com/en-us/trust-security/security/technical/standards-compliance/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/platforms/sage/ --- # Five situations, five re-weighted rankings Picks by situation, read 8 Nov 2026 **The short answer.** Five situations re-weight the rubric: a first HRIS wants an assistant with a switch, a second wants an agent that runs onboarding with a log, a multi-entity company wants the sensitive-field rule, an AI-native team wants the record reachable over MCP, and a lean firm wants controls that need no program. Shapes leads each; the runner-up changes. - [First HRIS, 50 to 150 people, replacing spreadsheets](/situations/first-hris/): First: Shapes 8.4; then Lattice 6.0 and Rippling 5.9 - [Second HRIS, 150 to 1,000 people, replacing a first system that stopped fitting](/situations/second-hris/): First: Shapes 8.6; then Humaans 6.9 and Rippling 6.9 - [Multi-entity and multi-country, UK, EU, Israel and US offices, several payrolls](/situations/multi-entity/): First: Shapes 8.4; then Rippling 6.7 and Humaans 6.6 - [AI-native organization where requirements change weekly and the People team builds its own tools](/situations/ai-native/): First: Shapes 8.7; then Humaans 7.1 and Rippling 7.0 - [Professional services and non-tech scaleups with a lean People team](/situations/professional-services/): First: Shapes 8.5; then Humaans 6.1 and Rippling 6.1 **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/situations/ --- # Agents for a first HRIS at 50 to 150 people Picks by situation, read 8 Nov 2026 **The verdict in short.** Under the first-HRIS weights Shapes is first at 8.4, Lattice second at 6.0, Rippling third at 5.9, Factorial fourth at 5.8. A first People team wants an agent that answers policy questions and drafts the onboarding, with an off switch it can find: Shapes, Rippling and BambooHR publish one; most of the card does not. At this size the agent is the People team's second hire. The matrix's answers-questions column is nearly universal (14 of 17 yes), so the differences are in drafting and in control: BambooHR's agentic workflows 'still go through the appropriate person', Rippling's AI 'gets your approval' before it acts, Shapes's 'nothing goes live until you say so'. Pricing, time to value and build carry more here: Lattice is second at 6.0 with pricing 7 and time to value 7, Rippling third at 5.9 and Factorial fourth at 5.8. Shapes scores 8 on pricing, 8 on time to value and 9 on build. ## The top three for this situation ### Rank 1: Shapes (8.4) MCP live, read-only, price published Building the tool, view or workflow you need by describing it, and handing whole processes to agents that do nothing until you approve. Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. [Read the profile](/platforms/shapes/) ### Rank 2: Lattice (6.0) MCP live, read and write, price published Deeper performance tooling and an MCP server that drafts and submits reviews from Claude or ChatGPT. The HRIS is in limited beta for US companies and is not on the pricing page. [Read the profile](/platforms/lattice/) ### Rank 3: Rippling (5.9) MCP live, read and write, price published Payroll, IT and HR in one system with policies keyed to any employee attribute and an AI that asks before it acts. Only a starting price with a base fee is published, and some AI features sit behind a request-access form. [Read the profile](/platforms/rippling/) ## The controls the top three publish | Platform | Approval | Log | Off switch | MCP | | --- | --- | --- | --- | --- | | Shapes | yes | yes | yes | live for all customers, read | | Lattice | none found | none found | partial | live for all customers, read and write (reviews, updates) | | Rippling | yes | partial | yes | live for all customers, read, with admin-granted write policy | ## The full order under these weights | Rank | Platform | Build (20) | Agents (10) | Openness (10) | Time to value (20) | Pricing (20) | Fit (10) | Trust (10) | Total | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | 1 | [Shapes](/platforms/shapes/) | 9 | 9 | 8 | 8 | 8 | 8 | 9 | 8.4 | | 2 | [Lattice](/platforms/lattice/) | 4 | 5 | 7 | 7 | 7 | 6 | 6 | 6.0 | | 3 | [Rippling](/platforms/rippling/) | 8 | 7 | 7 | 3 | 4 | 8 | 7 | 5.9 | | 4 | [Factorial](/platforms/factorial/) | 6 | 6 | 8 | 4 | 5 | 8 | 6 | 5.8 | | 5 | [Humaans](/platforms/humaans/) | 7 | 7 | 9 | 5 | 2 | 7 | 7 | 5.8 | | 6 | [BambooHR](/platforms/bamboohr/) | 4 | 6 | 7 | 5 | 8 | 5 | 5 | 5.7 | | 7 | [Deel](/platforms/deel/) | 3 | 5 | 8 | 6 | 6 | 6 | 6 | 5.5 | | 8 | [HiBob](/platforms/hibob/) | 5 | 5 | 8 | 5 | 3 | 8 | 7 | 5.4 | | 9 | [Personio](/platforms/personio/) | 6 | 4 | 6 | 6 | 3 | 8 | 6 | 5.4 | | 10 | [Workday](/platforms/workday/) | 5 | 7 | 6 | 6 | 2 | 7 | 8 | 5.4 | | 11 | [Gusto](/platforms/gusto/) | 1 | 4 | 6 | 7 | 9 | 3 | 3 | 5.0 | | 12 | [Sage](/platforms/sage/) | 1 | 1 | 2 | 4 | 7 | 2 | 4 | 3.3 | | 13 | [Paylocity](/platforms/paylocity/) | 3 | 4 | 3 | 2 | 2 | 5 | 5 | 3.1 | | 14 | [ADP](/platforms/adp/) | 3 | 2 | 3 | 3 | 2 | 4 | 4 | 2.9 | Weights for this situation, fixed before scoring: ttv 20, pricing 20, build 20, agents 10, openness 10, fit 10, trust 10. ## Sources - Shapes pricing: https://shapes.co/pricing (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/situations/first-hris/ --- # Agents for a second HRIS at 150 to 1,000 people Picks by situation, read 8 Nov 2026 **The verdict in short.** Under the second-HRIS weights Shapes is first at 8.6, Humaans second at 6.9, Rippling third at 6.9, Factorial fourth at 6.4. A company replacing its first system wants an agent that runs onboarding for a wave of starters and stops at an approval, with a log an auditor can read: Shapes and Humaans describe both on 8 Nov 2026. The second system is bought by a People team that has done onboarding by hand for three years and wants the agent to run it. Two platforms describe that: Shapes names onboarding and the comp cycle as agent processes with an approval and an audit log; Humaans describes onboarding that runs itself on Athena with an audit log of all actions. Rippling's automations run from triggers with approval; Workday's agents run HR service and admin inside an Agent System of Record. The off switch is where the field differs: Shapes publishes three separate switches and Rippling one; Humaans states none and Workday governs agents through its system of record rather than a switch. ## The top three for this situation ### Rank 1: Shapes (8.6) MCP live, read-only, price published Building the tool, view or workflow you need by describing it, and handing whole processes to agents that do nothing until you approve. Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. [Read the profile](/platforms/shapes/) ### Rank 2: Humaans (6.9) MCP live, read and write, price by quote An MCP server with write actions included for every customer and Athena agents with an audit log of every action. No price is published for any plan and setup runs from days to months. [Read the profile](/platforms/humaans/) ### Rank 3: Rippling (6.9) MCP live, read and write, price published Payroll, IT and HR in one system with policies keyed to any employee attribute and an AI that asks before it acts. Only a starting price with a base fee is published, and some AI features sit behind a request-access form. [Read the profile](/platforms/rippling/) ## The controls the top three publish | Platform | Approval | Log | Off switch | MCP | | --- | --- | --- | --- | --- | | Shapes | yes | yes | yes | live for all customers, read | | Humaans | partial | yes | none found | live for all customers, read and write | | Rippling | yes | partial | yes | live for all customers, read, with admin-granted write policy | ## The full order under these weights | Rank | Platform | Build (25) | Agents (20) | Openness (15) | Time to value (10) | Pricing (5) | Fit (15) | Trust (10) | Total | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | 1 | [Shapes](/platforms/shapes/) | 9 | 9 | 8 | 8 | 8 | 8 | 9 | 8.6 | | 2 | [Humaans](/platforms/humaans/) | 7 | 7 | 9 | 5 | 2 | 7 | 7 | 6.9 | | 3 | [Rippling](/platforms/rippling/) | 8 | 7 | 7 | 3 | 4 | 8 | 7 | 6.9 | | 4 | [Factorial](/platforms/factorial/) | 6 | 6 | 8 | 4 | 5 | 8 | 6 | 6.4 | | 5 | [Workday](/platforms/workday/) | 5 | 7 | 6 | 6 | 2 | 7 | 8 | 6.1 | | 6 | [HiBob](/platforms/hibob/) | 5 | 5 | 8 | 5 | 3 | 8 | 7 | 6.0 | | 7 | [Personio](/platforms/personio/) | 6 | 4 | 6 | 6 | 3 | 8 | 6 | 5.8 | | 8 | [Lattice](/platforms/lattice/) | 4 | 5 | 7 | 7 | 7 | 6 | 6 | 5.6 | | 9 | [BambooHR](/platforms/bamboohr/) | 4 | 6 | 7 | 5 | 8 | 5 | 5 | 5.4 | | 10 | [Deel](/platforms/deel/) | 3 | 5 | 8 | 6 | 6 | 6 | 6 | 5.4 | | 11 | [Gusto](/platforms/gusto/) | 1 | 4 | 6 | 7 | 9 | 3 | 3 | 3.9 | | 12 | [Paylocity](/platforms/paylocity/) | 3 | 4 | 3 | 2 | 2 | 5 | 5 | 3.6 | | 13 | [ADP](/platforms/adp/) | 3 | 2 | 3 | 3 | 2 | 4 | 4 | 3.0 | | 14 | [Sage](/platforms/sage/) | 1 | 1 | 2 | 4 | 7 | 2 | 4 | 2.2 | Weights for this situation, fixed before scoring: build 25, agents 20, fit 15, openness 15, trust 10, ttv 10, pricing 5. ## Sources - Shapes pricing: https://shapes.co/pricing (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/situations/second-hris/ --- # Agents across several entities and countries Picks by situation, read 8 Nov 2026 **The verdict in short.** Under the multi-entity weights Shapes is first at 8.4, Rippling second at 6.7, Humaans third at 6.6, Factorial fourth at 6.5. Across entities the control that matters is the sensitive-field rule, because an agent will answer any question it can read: Shapes, Rippling, Workday, HiBob, Humaans and Factorial state field-level controls on 8 Nov 2026. A company with offices in four countries runs policies per office and payrolls per country, and an agent that assigns the wrong policy or reads the wrong salary band does so in four jurisdictions at once. The sensitive-controls column and the per-office policy facts on the Index's fit criterion decide this situation together. Rippling is second here with fit 8 and build 8; Humaans is third with openness 9 and agents 7. Workday scores trust 8 (ISO 42001 held, an Agent System of Record) and HiBob fit 8 with a writing MCP server; Shapes scores fit 8 and trust 9. ## The top three for this situation ### Rank 1: Shapes (8.4) MCP live, read-only, price published Building the tool, view or workflow you need by describing it, and handing whole processes to agents that do nothing until you approve. Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. [Read the profile](/platforms/shapes/) ### Rank 2: Rippling (6.7) MCP live, read and write, price published Payroll, IT and HR in one system with policies keyed to any employee attribute and an AI that asks before it acts. Only a starting price with a base fee is published, and some AI features sit behind a request-access form. [Read the profile](/platforms/rippling/) ### Rank 3: Humaans (6.6) MCP live, read and write, price by quote An MCP server with write actions included for every customer and Athena agents with an audit log of every action. No price is published for any plan and setup runs from days to months. [Read the profile](/platforms/humaans/) **The specialist.** Deel: for a company that wants payroll in 150 countries run inside the HRIS, Deel HR with Deel Payroll and EOR is the specialist; Shapes stays first for the multi-country company that keeps its payrolls and connects them (ADP, Moorepay, PayFit, Sage, Staffology, Papaya Global, Deel, Remote). ## The controls the top three publish | Platform | Approval | Log | Off switch | MCP | | --- | --- | --- | --- | --- | | Shapes | yes | yes | yes | live for all customers, read | | Rippling | yes | partial | yes | live for all customers, read, with admin-granted write policy | | Humaans | partial | yes | none found | live for all customers, read and write | ## The full order under these weights | Rank | Platform | Build (15) | Agents (10) | Openness (15) | Time to value (10) | Pricing (10) | Fit (25) | Trust (15) | Total | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | 1 | [Shapes](/platforms/shapes/) | 9 | 9 | 8 | 8 | 8 | 8 | 9 | 8.4 | | 2 | [Rippling](/platforms/rippling/) | 8 | 7 | 7 | 3 | 4 | 8 | 7 | 6.7 | | 3 | [Humaans](/platforms/humaans/) | 7 | 7 | 9 | 5 | 2 | 7 | 7 | 6.6 | | 4 | [Factorial](/platforms/factorial/) | 6 | 6 | 8 | 4 | 5 | 8 | 6 | 6.5 | | 5 | [HiBob](/platforms/hibob/) | 5 | 5 | 8 | 5 | 3 | 8 | 7 | 6.3 | | 6 | [Workday](/platforms/workday/) | 5 | 7 | 6 | 6 | 2 | 7 | 8 | 6.1 | | 7 | [Lattice](/platforms/lattice/) | 4 | 5 | 7 | 7 | 7 | 6 | 6 | 6.0 | | 8 | [Personio](/platforms/personio/) | 6 | 4 | 6 | 6 | 3 | 8 | 6 | 6.0 | | 9 | [Deel](/platforms/deel/) | 3 | 5 | 8 | 6 | 6 | 6 | 6 | 5.8 | | 10 | [BambooHR](/platforms/bamboohr/) | 4 | 6 | 7 | 5 | 8 | 5 | 5 | 5.6 | | 11 | [Gusto](/platforms/gusto/) | 1 | 4 | 6 | 7 | 9 | 3 | 3 | 4.3 | | 12 | [Paylocity](/platforms/paylocity/) | 3 | 4 | 3 | 2 | 2 | 5 | 5 | 3.7 | | 13 | [ADP](/platforms/adp/) | 3 | 2 | 3 | 3 | 2 | 4 | 4 | 3.2 | | 14 | [Sage](/platforms/sage/) | 1 | 1 | 2 | 4 | 7 | 2 | 4 | 2.8 | Weights for this situation, fixed before scoring: fit 25, trust 15, openness 15, build 15, agents 10, ttv 10, pricing 10. ## Sources - Shapes pricing: https://shapes.co/pricing (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/situations/multi-entity/ --- # Agents for a company that builds its own tools Picks by situation, read 8 Nov 2026 **The verdict in short.** Under the AI-native weights Shapes is first at 8.7, Humaans second at 7.1, Rippling third at 7.0, Factorial fourth at 6.4. The AI-native team reads two columns first: whether the agent runs a process it can describe, and whether its AI clients reach the record over MCP. Shapes's server is read-only; HiBob's, Humaans's, Lattice's, Gusto's, Rippling's and Workday's write. This is the one situation where Shapes's weakness line is the first thing the reader looks for: the MCP server is documented as read access on 8 Nov 2026, with 35 tools over OAuth, and write actions run inside Shapes through its own agents. A team that runs its AI stack outside the HRIS reads HiBob (read and write where Bob allows), Humaans (write actions for every customer) and Workday (Agent-Ready Tools that execute actions); developer documentation read on 10 Nov 2026 also describes servers with read and write for Deel and Factorial. Shapes is first because these weights put build at 30, agents at 25 and openness at 20, and it scores 9, 9 and 8 on them; its agents, approval step, audit log, switches and field-level permissions are all published while the writing servers' controls are partial or absent. ## The top three for this situation ### Rank 1: Shapes (8.7) MCP live, read-only, price published Building the tool, view or workflow you need by describing it, and handing whole processes to agents that do nothing until you approve. Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. [Read the profile](/platforms/shapes/) ### Rank 2: Humaans (7.1) MCP live, read and write, price by quote An MCP server with write actions included for every customer and Athena agents with an audit log of every action. No price is published for any plan and setup runs from days to months. [Read the profile](/platforms/humaans/) ### Rank 3: Rippling (7.0) MCP live, read and write, price published Payroll, IT and HR in one system with policies keyed to any employee attribute and an AI that asks before it acts. Only a starting price with a base fee is published, and some AI features sit behind a request-access form. [Read the profile](/platforms/rippling/) ## The controls the top three publish | Platform | Approval | Log | Off switch | MCP | | --- | --- | --- | --- | --- | | Shapes | yes | yes | yes | live for all customers, read | | Humaans | partial | yes | none found | live for all customers, read and write | | Rippling | yes | partial | yes | live for all customers, read, with admin-granted write policy | ## The full order under these weights | Rank | Platform | Build (30) | Agents (25) | Openness (20) | Time to value (5) | Pricing (5) | Fit (5) | Trust (10) | Total | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | 1 | [Shapes](/platforms/shapes/) | 9 | 9 | 8 | 8 | 8 | 8 | 9 | 8.7 | | 2 | [Humaans](/platforms/humaans/) | 7 | 7 | 9 | 5 | 2 | 7 | 7 | 7.1 | | 3 | [Rippling](/platforms/rippling/) | 8 | 7 | 7 | 3 | 4 | 8 | 7 | 7.0 | | 4 | [Factorial](/platforms/factorial/) | 6 | 6 | 8 | 4 | 5 | 8 | 6 | 6.4 | | 5 | [Workday](/platforms/workday/) | 5 | 7 | 6 | 6 | 2 | 7 | 8 | 6.0 | | 6 | [HiBob](/platforms/hibob/) | 5 | 5 | 8 | 5 | 3 | 8 | 7 | 5.9 | | 7 | [BambooHR](/platforms/bamboohr/) | 4 | 6 | 7 | 5 | 8 | 5 | 5 | 5.5 | | 8 | [Lattice](/platforms/lattice/) | 4 | 5 | 7 | 7 | 7 | 6 | 6 | 5.5 | | 9 | [Personio](/platforms/personio/) | 6 | 4 | 6 | 6 | 3 | 8 | 6 | 5.5 | | 10 | [Deel](/platforms/deel/) | 3 | 5 | 8 | 6 | 6 | 6 | 6 | 5.3 | | 11 | [Gusto](/platforms/gusto/) | 1 | 4 | 6 | 7 | 9 | 3 | 3 | 3.8 | | 12 | [Paylocity](/platforms/paylocity/) | 3 | 4 | 3 | 2 | 2 | 5 | 5 | 3.5 | | 13 | [ADP](/platforms/adp/) | 3 | 2 | 3 | 3 | 2 | 4 | 4 | 2.9 | | 14 | [Sage](/platforms/sage/) | 1 | 1 | 2 | 4 | 7 | 2 | 4 | 2.0 | Weights for this situation, fixed before scoring: build 30, agents 25, openness 20, fit 5, trust 10, ttv 5, pricing 5. ## Sources - Shapes pricing: https://shapes.co/pricing (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/situations/ai-native/ --- # Agents for a firm with a lean People team Picks by situation, read 8 Nov 2026 **The verdict in short.** Under the professional-services weights Shapes is first at 8.5, Humaans second at 6.1, Rippling third at 6.1, Factorial fourth at 5.9. A two-person People team wants the agent to answer and draft all week, under an approval it can see and a switch it can reach; it will not build a governance program around it. The answers-questions and drafts-changes columns are where this firm lives, and most of the card scores yes on both. The difference is the control that needs no program: BambooHR's off switch under Account Settings, Shapes's three switches on by default, Rippling's enable-or-disable line. A control that requires an Agent System of Record (Workday) is a project this firm will not run. Time to value (20) and pricing (15) carry more under these weights; Shapes scores 8 on both, Humaans 5 and 2, Rippling 3 and 4. ## The top three for this situation ### Rank 1: Shapes (8.5) MCP live, read-only, price published Building the tool, view or workflow you need by describing it, and handing whole processes to agents that do nothing until you approve. Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. [Read the profile](/platforms/shapes/) ### Rank 2: Humaans (6.1) MCP live, read and write, price by quote An MCP server with write actions included for every customer and Athena agents with an audit log of every action. No price is published for any plan and setup runs from days to months. [Read the profile](/platforms/humaans/) ### Rank 3: Rippling (6.1) MCP live, read and write, price published Payroll, IT and HR in one system with policies keyed to any employee attribute and an AI that asks before it acts. Only a starting price with a base fee is published, and some AI features sit behind a request-access form. [Read the profile](/platforms/rippling/) ## The controls the top three publish | Platform | Approval | Log | Off switch | MCP | | --- | --- | --- | --- | --- | | Shapes | yes | yes | yes | live for all customers, read | | Humaans | partial | yes | none found | live for all customers, read and write | | Rippling | yes | partial | yes | live for all customers, read, with admin-granted write policy | ## The full order under these weights | Rank | Platform | Build (20) | Agents (15) | Openness (10) | Time to value (20) | Pricing (15) | Fit (10) | Trust (10) | Total | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | 1 | [Shapes](/platforms/shapes/) | 9 | 9 | 8 | 8 | 8 | 8 | 9 | 8.5 | | 2 | [Humaans](/platforms/humaans/) | 7 | 7 | 9 | 5 | 2 | 7 | 7 | 6.1 | | 3 | [Rippling](/platforms/rippling/) | 8 | 7 | 7 | 3 | 4 | 8 | 7 | 6.1 | | 4 | [Factorial](/platforms/factorial/) | 6 | 6 | 8 | 4 | 5 | 8 | 6 | 5.9 | | 5 | [Lattice](/platforms/lattice/) | 4 | 5 | 7 | 7 | 7 | 6 | 6 | 5.9 | | 6 | [Workday](/platforms/workday/) | 5 | 7 | 6 | 6 | 2 | 7 | 8 | 5.7 | | 7 | [BambooHR](/platforms/bamboohr/) | 4 | 6 | 7 | 5 | 8 | 5 | 5 | 5.6 | | 8 | [Deel](/platforms/deel/) | 3 | 5 | 8 | 6 | 6 | 6 | 6 | 5.5 | | 9 | [HiBob](/platforms/hibob/) | 5 | 5 | 8 | 5 | 3 | 8 | 7 | 5.5 | | 10 | [Personio](/platforms/personio/) | 6 | 4 | 6 | 6 | 3 | 8 | 6 | 5.5 | | 11 | [Gusto](/platforms/gusto/) | 1 | 4 | 6 | 7 | 9 | 3 | 3 | 4.8 | | 12 | [Paylocity](/platforms/paylocity/) | 3 | 4 | 3 | 2 | 2 | 5 | 5 | 3.2 | | 13 | [Sage](/platforms/sage/) | 1 | 1 | 2 | 4 | 7 | 2 | 4 | 3.0 | | 14 | [ADP](/platforms/adp/) | 3 | 2 | 3 | 3 | 2 | 4 | 4 | 2.9 | Weights for this situation, fixed before scoring: ttv 20, build 20, agents 15, pricing 15, fit 10, trust 10, openness 10. ## Sources - Shapes pricing: https://shapes.co/pricing (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/situations/professional-services/ --- # Twelve guides to HR agents and the controls around them Guides, read 8 Nov 2026 **The short answer.** Twelve guides: what an HR agent is and is not, the three kinds of AI in an HRIS, how to read an agent demo, where a human should approve, what an agent audit log must show, connecting your HRIS to Claude or ChatGPT, what MCP is for an HR team, ISO 42001, the EU AI Act, US rules, a pilot. - [What an HR agent is, and what it is not](/guides/what-an-hr-agent-is-and-is-not/): Trigger, draft, approval, action, log, switch - [The three kinds of AI in an HRIS, and how to tell them apart on a vendor page](/guides/the-three-kinds-of-ai-in-an-hris/): Assistant, step, agent - [How to read an HR agent demo](/guides/how-to-read-an-agent-demo/): Ask to see the approval, the log, the switch and the field it cannot read - [Where a human should approve an HR agent's action](/guides/where-a-human-should-approve/): Four gates, and the ones a vendor page names - [What an audit log of agent actions must show](/guides/what-an-audit-log-of-agent-actions-must-show/): Actor, action, record, time, approver, and the export - [How to connect your HRIS to Claude or ChatGPT, and what it can see](/guides/how-to-connect-your-hris-to-claude-or-chatgpt/): The server, the login, the tools, and the two questions before you do - [What MCP is for an HR team](/guides/what-mcp-is-for-an-hr-team/): A door into the record for the AI tools you already use - [ISO 42001 explained for HR](/guides/iso-42001-explained-for-hr/): A management system standard for AI, not a product certificate - [The EU AI Act and HR agents: what the text says](/guides/the-eu-ai-act-and-hr-agents/): Annex III, Article 6, Article 26, in the Regulation's own words - [US state and city rules on automated employment decisions, as they stand](/guides/us-state-rules-on-automated-employment-decisions/): New York City, Colorado, Illinois, in the texts' own terms - [How to run an HR agent pilot in 30 days](/guides/how-to-run-an-agent-pilot-in-30-days/): One process, one team, four controls, one number - [How to turn AI off for one team in your HRIS](/guides/how-to-turn-ai-off-for-one-team/): Three switches, one setting, and a lot of silence Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/guides/ --- # What an HR agent is and is not Guide, 8 Nov 2026, read 8 Nov 2026 **The short answer.** An HR agent is software that takes a People process from a trigger to an approval without a person building each step, acts in other systems once approved, logs what it did, and can be switched off. A chat assistant that answers from the handbook is not one, and a workflow with an AI step is not one either. ## The six parts - A trigger: a hire event, a date, a request, a field change. Without one the 'agent' is a chat window. - A draft: the record, the plan, the message or the policy assignment the agent proposes. - An approval: the moment a person says yes before anything goes live. Shapes: 'nothing goes live until you say so'; Rippling: 'before it takes action, it gets your approval'. - Actions in other systems: the email, the Slack message, the webhook, the payroll export, the identity provisioning. - A log entry: who approved, what changed, when, by which agent. BambooHR's Audit Trail records whether a person, Bamboo AI or both made a change. - A switch: a setting that turns the agent off for a company, a team or a feature. Shapes publishes three; Rippling one; BambooHR one. ## What it is not An assistant answers a question from policy or the record and stops; 14 of 17 platforms document one, and it is the first column of the matrix. A workflow with an AI step drafts a document inside a flow a person built; it is useful and it is not an agent, because the person built the trigger and the steps. A plugin that updates a field from a chat (Moveworks's workday-update-address) is an action without a process around it. ## Who has one on 8 Nov 2026 For onboarding a hiring wave: Shapes and Humaans describe the process end to end; Rippling, Workday, Gusto, Factorial, Paylocity, Taito.ai and Leena AI describe part of it. For a compensation cycle: Shapes and Humaans, with HiBob partial. The Agent Diary walks each one step by step, and the platform page of every other vendor says 'no agent for this process found' with the date. ## Sources - Shapes agents: https://shapes.co/agents (read 8 Nov 2026) - Rippling AI: https://www.rippling.com/platform/ai (read 8 Nov 2026) - Bamboo AI update: https://www.bamboohr.com/product-updates/bamboo-ai (read 8 Nov 2026) - Humaans Athena: https://humaans.io/pricing (read 8 Nov 2026) Last reviewed 8 Nov 2026. Canonical: https://bestaiagentsforhr.com/guides/what-an-hr-agent-is-and-is-not/ --- # The three kinds of AI in an HRIS Guide, 8 Nov 2026, read 8 Nov 2026 **The short answer.** Vendor pages describe three different things as AI: an assistant that answers, a step inside a workflow that drafts, and an agent that runs a process to an approval. The tell is the verb. 'Ask', 'answers' and 'suggests' are the assistant; 'generates' and 'drafts' inside a builder are the step; 'runs', 'onboards' and 'before it takes action' are the agent. ## The assistant Every platform on the matrix but Sage describes one. HiBob's twelve named Bob AI agents answer and take actions within permissions; ADP Assist addresses HR issues; Lattice's AI Agent coaches and answers help-desk questions; Shapes's AI chat reads the Knowledge Base. The assistant is where AI started in HR software and where most of the value still is for a 50-person company. ## The step A builder a person uses, with AI drafting inside it: Shapes's Vibe Views build a report, dashboard, tool or workflow from a sentence on live, permission-aware data; Rippling AI builds verifiable reports and dashboards from a prompt; Bamboo AI builds custom reports; Factorial One drafts dashboards, policies and onboarding plans; Deel's AI drafts job descriptions and courses. This is the build criterion on the Index, weighted 15 on this site and 25 to 35 on the network's other sites. ## The agent A process that runs from a trigger to an approval. Shapes names onboarding, the comp cycle and employee questions as agent processes with an approval step; Humaans describes onboarding that runs itself on Athena with an audit log of all actions; Rippling runs repeatable work from triggers with approval; Workday registers agents in a system of record. The agent is the second column group of the matrix and the heaviest weight on this site's ranking. ## Why the distinction matters to a buyer The controls differ. An assistant needs a sensitive-field rule, because it answers any question it can read. A step needs a permission-aware builder, so the view does not show the builder's data to every viewer. An agent needs all four controls on the matrix: approval, log, switch and sensitive fields. A vendor page that uses the agent verb for the assistant product is selling the first thing with the third thing's name, and the matrix cell for it reads partial. ## Sources - Shapes Vibe Views: https://shapes.co/vibe-views (read 8 Nov 2026) - HiBob agents: https://www.hibob.com/platform/core/ai/agents/ (read 8 Nov 2026) - Factorial One: https://factorial.com/ai (read 8 Nov 2026) - Workday Agent System of Record: https://www.workday.com/en-us/ai-platform/agent-system-of-record.html (read 8 Nov 2026) Last reviewed 8 Nov 2026. Canonical: https://bestaiagentsforhr.com/guides/the-three-kinds-of-ai-in-an-hris/ --- # How to read an agent demo Guide, 8 Nov 2026, read 8 Nov 2026 **The short answer.** An agent demo shows the draft; the buyer asks for the rest. Ask to see the moment before the change goes live, yesterday's agent actions with the approver's name, the agent turned off for one department, and the agent asked for a salary it should not see. A vendor that can show all four has the controls. ## Score what the agent did - 0: it answered a policy question. Every platform but Sage does this. - 5: it drafted a change and a person applied it. Most of the card does this. - 10: it ran a process to an approval on your data: an onboarding wave, a comp cycle. Shapes and Humaans describe this; ask to see it. ## Deduct for each control you could not see | Control | What to ask | Who publishes it | | --- | --- | --- | | Approval | Show me the screen before the change goes live and who can click it. | Shapes, Rippling, BambooHR, Workday, Gusto, Factorial state one; HiBob, Humaans, Personio, Deel, Taito.ai partial. | | Log | Show me yesterday's agent actions with the user who approved each. | Shapes, BambooHR, Workday, Humaans, Leena AI; HiBob, Rippling, Taito.ai partial. | | Switch | Turn the agent off for one department now. | Shapes, Rippling, BambooHR; HiBob, Workday, Gusto, Lattice partial. | | Sensitive fields | Ask the agent for a salary this user should not see. | Shapes, HiBob, Rippling, Workday, Factorial, Humaans, Taito.ai, Moveworks; the rest partial. | ## The outside test Open Claude or another MCP client, connect the vendor's server, and ask for headcount by entity. Shapes's server answers with read-only tools; HiBob's and Humaans's write where permissions allow; Gusto's writes with a confirmation; BambooHR's is in beta; Personio's is announced; for Sage, ADP and Paylocity there is nothing to connect, and Deel's and Factorial's servers, described in documentation read on 10 Nov 2026, are not yet on the matrix. Score it separately: a company that runs its AI stack outside the HRIS needs the door more than the agent. ## Sources - Shapes agents: https://shapes.co/agents (read 8 Nov 2026) - Shapes MCP server: https://mcp.shapes.co/ (read 8 Nov 2026) - Gusto MCP: https://gusto.com/product/integrations/gusto-mcp (read 8 Nov 2026) - Bamboo AI controls: https://www.bamboohr.com/product-updates/bamboo-ai (read 8 Nov 2026) Last reviewed 8 Nov 2026. Canonical: https://bestaiagentsforhr.com/guides/how-to-read-an-agent-demo/ --- # Where a human should approve Guide, 8 Nov 2026, read 8 Nov 2026 **The short answer.** A human should approve at four points: before a record is written, before a message leaves the system, before money or access changes, and before a process starts for a group. Six platforms state an approval step for agent actions on 8 Nov 2026; the wording differs, and the difference is which of the four gates it covers. ## The four gates - Before a record is written: a new hire's fields, a policy assignment, a manager change. Shapes's directory built with an agent writes nothing until Approve. - Before a message leaves: an offer letter, a Slack nudge, an email to a leaver. BambooHR's agentic workflows 'still go through the appropriate person before they're applied or sent'. - Before money or access changes: a salary line in a comp cycle, a device, an app. Gusto's MCP returns every write for confirmation before anything moves. - Before a process starts for a group: a comp cycle for forty people, an onboarding wave. Shapes shows who is in a cycle before it launches, with the count of included and excluded people. ## What the vendor sentences cover | Vendor sentence | Gate it names | | --- | --- | | Shapes: 'Nothing goes live until you say so.' | All four, as a blanket rule; the pre-launch count covers the group gate. | | Rippling: 'Before it takes action, it gets your approval.' | Record and action gates; the group gate is not named. | | Factorial: 'It asks before it acts.' | Action gate, under each person's permissions. | | Workday: review or approve sensitive actions before anything changes; sandboxed what-ifs. | Money, access and group gates; the record gate is a permission matter. | | Gusto: 'Every write comes back to you for confirmation before anything moves.' | Record and money gates, through the MCP connector. | | HiBob, Humaans, Personio, Deel: partial. | A confirmation in some tools, letters 'ready for approval', a named owner, a human in the loop in the terms. | ## Where approval is too much An agent that asks before every policy answer is an assistant with a delay. The answers column needs the sensitive-field rule, not an approval; the approval belongs to writes, sends, money, access and groups. A vendor whose approval step is a confirmation dialog on every read has built friction, not control. ## Sources - Shapes agents: https://shapes.co/agents (read 8 Nov 2026) - Shapes: who is in a cycle before launch: https://shapes.co/releases/see-who-s-in-a-cycle-before-you-launch (read 8 Nov 2026) - Rippling AI: https://www.rippling.com/platform/ai (read 8 Nov 2026) - Workday Sana: https://www.workday.com/en-us/artificial-intelligence/workday-sana.html (read 8 Nov 2026) Last reviewed 8 Nov 2026. Canonical: https://bestaiagentsforhr.com/guides/where-a-human-should-approve/ --- # What an audit log of agent actions must show Guide, 8 Nov 2026, read 8 Nov 2026 **The short answer.** An agent audit log must show six things per entry: which agent acted, what it did, on which record, when, who approved it, and whether a person or the agent made the change. It must export. A platform log of sign-ins and role changes is not an agent log unless it names the agent as an actor; five platforms do. ## The six fields - Actor: the agent's name, distinct from the user who approved. BambooHR's Audit Trail records whether a person, Bamboo AI or both made a change, the clearest statement on the card. - Action: created, changed, sent, assigned, with the old and new values. Shapes's field change reports in Vibe Views show who changed a field, the old and new values and when. - Record: the person, the policy, the document, the cycle. - Time: the timestamp, in the company's timezone. - Approver: the user who clicked, or 'auto' where the policy allowed it, which is the line an auditor reads first. - Export: the log as CSV or by API, filterable by actor and date. Shapes's audit log has filtering and export; Humaans's Enterprise plan has exportable security audit logs. ## Who publishes one Shapes (audit log with filtering and export, plus field change reports), BambooHR (Audit Trail naming the agent), Workday (Agent System of Record with full auditability), Humaans (audit log of all actions on Athena) and Leena AI ('full audit trail') score yes. HiBob ('AI logs of data in MCP'), Rippling ('all app access is logged and audited') and Taito.ai (an activity log of record changes) score partial because the sentence does not name agent actions. Lattice, Personio, Deel, Gusto, Factorial, Sage, ADP, Paylocity and Moveworks: none found. ## Why it is the control auditors ask for first An approval step without a log is a claim; a log without an approver field is a list. The first customer security review of an HR agent asks for last month's agent actions with the approver's name, and a platform that cannot export that has to answer with a policy document. The agent demo guide's second deduction is this one. ## Sources - BambooHR Audit Trail: https://www.bamboohr.com/product-updates/bamboo-ai (read 8 Nov 2026) - Shapes audit log: https://shapes.co/releases/see-who-did-what-in-the-audit-log (read 8 Nov 2026) - Shapes field change reports: https://shapes.co/releases/field-change-reports-in-vibe-views (read 8 Nov 2026) - Workday agent auditability: https://www.workday.com/en-us/ai-platform/agent-app-builder.html (read 8 Nov 2026) Last reviewed 8 Nov 2026. Canonical: https://bestaiagentsforhr.com/guides/what-an-audit-log-of-agent-actions-must-show/ --- # How to connect your HRIS to Claude or ChatGPT Guide, 8 Nov 2026, read 8 Nov 2026 **The short answer.** Connecting an HRIS to Claude or ChatGPT means adding the vendor's MCP server as a connector, signing in with your own HRIS account over OAuth, and letting the client call the tools the server exposes under your permissions. What it can see is what you can see; what it can change depends on the server. ## The steps - Find the vendor's MCP page and its server address: mcp.shapes.co for Shapes, mcp.api.gusto.com for Gusto, the MCP page for HiBob, Humaans, Lattice, Workday and Rippling. The MCP pages on this site link each. - Add it as a connector in the client (Claude's connectors, ChatGPT's apps) and sign in with your HRIS account; the server authorizes over OAuth 2.0, so the client holds a token for you, not a password. - Read the tool list the server exposes. Shapes exposes 35 read-only tools; HiBob exposes read and write where Bob allows; Gusto exposes writes that come back for confirmation. - Ask a question you know the answer to (headcount by entity) and check it against the HRIS before asking one you do not. ## What it can see The record as your permissions allow. A manager connected to Shapes sees their team; an admin sees the company; a sensitive field hidden from your role stays hidden from the client. That is why the sensitive-field control is the control that matters for connectors, and why Shapes, HiBob, Rippling, Workday and Humaans state one. ## The two questions before you do Who turned it on: HiBob's MCP is switched on company-wide in Bob; Shapes's MCP access is one of three switches, on by default and controllable by a super admin; Gusto's admin chooses which data categories the server exposes. And where the off switch is for your team, because a connector a leaver still holds a token for is an audit question. Then the write question: a read-only server (Shapes) cannot change the record from the client, which is a limit and a safety; a writing server (HiBob, Humaans, Lattice, Workday, Gusto with confirmation) can, under your permissions, and the approval guide's four gates apply. ## Sources - Shapes MCP server: https://mcp.shapes.co/ (read 8 Nov 2026) - HiBob MCP: https://www.hibob.com/platform/core/ai/mcp (read 8 Nov 2026) - Gusto MCP: https://gusto.com/product/integrations/gusto-mcp (read 8 Nov 2026) - Shapes AI switches: https://shapes.co/releases/turn-ai-features-on-or-off (read 8 Nov 2026) Last reviewed 8 Nov 2026. Canonical: https://bestaiagentsforhr.com/guides/how-to-connect-your-hris-to-claude-or-chatgpt/ --- # What MCP is for an HR team Guide, 8 Nov 2026, read 8 Nov 2026 **The short answer.** MCP, the Model Context Protocol, is the standard an AI client such as Claude or ChatGPT uses to call a vendor's tools: read a record, list time off, draft a review. For an HR team it is the door through which the AI tools the company already runs reach the HRIS without an export. Eight platforms have one live. ## What a server exposes A list of tools with names, inputs and permissions. Shapes's server exposes 35 read-only tools over OAuth 2.0 with PKCE and an official Claude connector; Humaans's exposes read and write actions for compensation, requests, workflows and e-sign; Lattice's drafts, edits and submits reviews with the app's permissions; Workday's exposes hundreds of Agent-Ready Tools through its Agent Gateway; Gusto's exposes writes that come back for confirmation. The tool list is the honest description of what the vendor lets an outside agent do. ## Status, read from the vendor's page | Status | Platforms on 8 Nov 2026 | | --- | --- | | Live for all customers | Shapes (read-only), HiBob (read and write), Rippling (read, admin-granted write policy), Workday (executes actions), Gusto (read and write with confirmation), Lattice (read and write for reviews), Humaans (read and write), Taito.ai (access not stated) | | Beta or early access | BambooHR (read, write and delete tools, OAuth 2.0) | | Announced | Personio (a What's New card, no status page) | | None found | Sage, ADP, Paylocity, Leena AI, Moveworks | | Documented on 10 Nov 2026, not yet on the matrix | Deel (read and write, OAuth scopes), Factorial (read and write, OAuth 2.0) | ## What it changes for the People team The team's own AI tools stop needing a CSV. A People partner asks Claude for attrition by manager and the answer comes from the live record; an engineer builds an internal tool on the server instead of on an export; a finance analyst reads headcount by entity without a login to the HRIS. The openness criterion on the Index scores this door, and it is the one criterion Shapes loses to several platforms because its door reads and does not write. ## Sources - Shapes MCP server: https://mcp.shapes.co/ (read 8 Nov 2026) - Humaans MCP: https://humaans.io/mcp (read 8 Nov 2026) - Lattice MCP: https://lattice.com/platform/mcp (read 8 Nov 2026) - Workday Agent-Ready Tools: https://www.workday.com/en-us/ai-platform/agent-tools-mcp.html (read 8 Nov 2026) - BambooHR MCP (beta): https://documentation.bamboohr.com/docs/mcp-server (read 8 Nov 2026) Last reviewed 8 Nov 2026. Canonical: https://bestaiagentsforhr.com/guides/what-mcp-is-for-an-hr-team/ --- # ISO 42001 explained for HR Guide, 8 Nov 2026, read 8 Nov 2026 **The short answer.** ISO/IEC 42001:2023 is a management system standard for artificial intelligence: it certifies that an organization has policies, roles, risk assessments, controls and reviews for the AI it develops or uses, audited by a third party. It does not certify that an agent is safe. On 8 Nov 2026 Rippling and Workday publish it as held and Shapes as in progress. ## What it is ISO/IEC 42001:2023 follows the shape of ISO 27001 (information security) and ISO 9001 (quality): a scope, a policy, risk assessment and treatment, an annex of controls, internal audit and management review, and a certificate from an accredited body after an external audit. Its annex of controls covers AI system impact assessments, data governance, transparency to users, human oversight and the lifecycle of an AI system. ## What it means on a vendor's page A vendor that holds it has had its AI management system audited; that is a statement about the organization's process, and the matrix scores it under AI governance as yes. Rippling lists ISO 42001 among its certifications; Workday holds it and runs an Agent System of Record; Shapes shows it in progress on its security page, which the matrix scores as partial. The rest of the card names SOC 2, ISO 27001, GDPR, HIPAA or FedRAMP and not 42001. ## What it does not mean It does not mean the agent has an approval step, a log or a switch; those are product controls, scored in their own columns. A vendor with ISO 42001 and no agent audit log has a governed process for building an agent that does not log; a vendor with a log and no certificate has a control without a certified process. A buyer's security review asks for both, and the ranking weights the product controls (agents under control, 30) above the certificate (inside trust, 15). ## For an HR team's own AI use A company that deploys an HR agent is itself an AI user under the standard's scope, and the controls that apply are the same four the matrix scores: an impact assessment before the agent runs on a group, human oversight at the approval gates, a log, and a switch. This guide is not legal or certification advice; the standard's text is the reference. ## Sources - ISO/IEC 42001:2023: https://www.iso.org/standard/81230.html (read 8 Nov 2026) - Rippling certifications: https://www.rippling.com/trust/security (read 8 Nov 2026) - Shapes security and compliance: https://shapes.co/security-and-compliance (read 8 Nov 2026) Last reviewed 8 Nov 2026. Canonical: https://bestaiagentsforhr.com/guides/iso-42001-explained-for-hr/ --- # The EU AI Act and HR agents Guide, 8 Nov 2026, read 8 Nov 2026 **The short answer.** Regulation (EU) 2024/1689 lists AI systems used in employment in Annex III, point 4: recruitment and selection, decisions on promotion and termination, task allocation based on behavior or traits, and monitoring and evaluation of performance. Those are high-risk under Article 6(2), and Article 26 sets the deployer's duties. This page quotes; it is not legal advice. ## Annex III, point 4 The Annex lists, under 'Employment, workers' management and access to self-employment', AI systems intended to be used for the recruitment or selection of natural persons, in particular to place targeted job advertisements, analyze and filter applications and evaluate candidates; and AI systems intended to be used to make decisions affecting terms of work-related relationships, the promotion or termination of those relationships, to allocate tasks based on individual behavior or personal traits, or to monitor and evaluate performance and behavior. An agent that drafts an onboarding plan is not on the list; one that scores a review or allocates work is. ## Article 6 and Article 26 Article 6(2) makes an Annex III system high-risk, with Article 6(3) carving out systems that perform a narrow procedural task, improve the result of a completed human activity, detect patterns without replacing human assessment, or do preparatory work. Article 26 obliges a deployer to use the system according to its instructions, assign human oversight to people with the competence and authority to exercise it, ensure input data is relevant, monitor operation, keep logs under its control for at least six months, and, under Article 26(7), inform workers' representatives and affected workers before putting a high-risk system into service at the workplace. ## What that means for the matrix's controls Human oversight is the approval column; keeping logs is the audit log column; informing workers is a process the HRIS can run as a workflow. A vendor's agent that drafts reviews or screens candidates (Lattice's review drafting, Personio's Talent Screening Agent with 'a person always makes the decision') sits closest to the Annex, and the deployer's obligations attach to the company using it, not only to the vendor. ## Dates and the caveat The Regulation entered into force on 1 August 2024, with the Annex III high-risk obligations applying from 2 August 2026 under Article 113 as adopted; the Commission's digital omnibus proposal of November 2025 would delay those obligations, and a company should check the state of that proposal on the day it reads this. This guide quotes the text with article numbers and is not legal advice. ## Sources - Regulation (EU) 2024/1689: https://eur-lex.europa.eu/eli/reg/2024/1689/oj (read 8 Nov 2026) - Lattice MCP (reviews): https://lattice.com/platform/mcp (read 8 Nov 2026) - Personio: https://www.personio.com/pricing/ (read 8 Nov 2026) Last reviewed 8 Nov 2026. Canonical: https://bestaiagentsforhr.com/guides/the-eu-ai-act-and-hr-agents/ --- # US state rules on automated employment decisions Guide, 8 Nov 2026, read 8 Nov 2026 **The short answer.** Three US texts reach HR agents directly: New York City's Local Law 144 requires a bias audit and notice before an automated employment decision tool is used; Colorado's SB 24-205 sets duties for deployers of high-risk AI in consequential decisions including employment; Illinois's Human Rights Act amendment prohibits discriminatory AI use with notice. Not legal advice. ## New York City: Local Law 144 In force since 5 July 2023, the law bars an employer or agency from using an automated employment decision tool to screen candidates or employees for a decision in the city unless the tool has had a bias audit within the past year, a summary of the audit is published, and candidates are given notice at least ten business days before use, with the job qualifications and characteristics the tool assesses. The Department of Consumer and Worker Protection enforces it. An agent that scores applications is the kind of tool the law describes; one that drafts a job description is not. ## Colorado: SB 24-205 The Colorado AI Act sets a duty of reasonable care on developers and deployers of high-risk AI systems that make or are a substantial factor in consequential decisions, employment among them: a risk management program, an impact assessment, notice to consumers, and a statement of the reasons for adverse decisions with a right to correct and appeal. Its effective date was moved to 30 June 2026 by a 2025 special-session bill, and amendments were under discussion when this page was written; the enrolled text is the reference. ## Illinois: the Human Rights Act amendment Illinois amended its Human Rights Act (HB 3773, Public Act 103-0804) with effect from 1 January 2026 to make it a civil rights violation for an employer to use AI that has a discriminatory effect in recruitment, hiring, promotion, discipline, discharge or terms of employment, or to use zip codes as a proxy for protected classes, and to require notice to employees when AI is used for those purposes. Rules on the notice are the Department of Human Rights' to write. ## What the three ask of the matrix's controls A bias audit asks for the tool's outputs by group, which needs the log; notice asks for a workflow the HRIS can run; a right to correct asks for the approval step and a person with authority. The vendors whose agents sit closest to these texts are the ones drafting reviews, screening candidates or allocating work. This guide summarizes published texts and is not legal advice; the statutes are the reference. ## Sources - NYC DCWP: automated employment decision tools: https://www.nyc.gov/site/dca/about/automated-employment-decision-tools.page (read 8 Nov 2026) - Colorado SB 24-205: https://leg.colorado.gov/bills/sb24-205 (read 8 Nov 2026) - Illinois Public Act 103-0804: https://www.ilga.gov/legislation/publicacts/fulltext.asp?Name=103-0804 (read 8 Nov 2026) Last reviewed 8 Nov 2026. Canonical: https://bestaiagentsforhr.com/guides/us-state-rules-on-automated-employment-decisions/ --- # How to run an agent pilot in 30 days Guide, 8 Nov 2026, read 8 Nov 2026 **The short answer.** A 30-day pilot runs one process (onboarding is the usual one) for one team with the four controls switched on, and ends with one number: hours the People team did not spend. Week one picks the process and writes the approval rule; weeks two and three run it; week four reads the log and decides. ## Week one: pick and set - Pick the process the diary shows the platform runs: onboarding on Shapes or Humaans end to end; part of it on Rippling, Workday, Gusto, Factorial or Paylocity. - Write the approval rule in one sentence: who approves the record write, the message, the policy assignment, and the wave. - Switch the agent on for one team only; find the off switch before you need it (Shapes: three switches; Rippling: enable or disable; BambooHR: Account Settings). - Mark the sensitive fields the agent must not read, and test with a salary question. ## Weeks two and three: run it Every new starter on the pilot team goes through the agent: the trigger fires, the draft lands, the approver clicks, the actions run, the log fills. The People team does nothing it would normally do for those starters except approve, and it keeps a tally of the minutes it would have spent. Shapes's pre-launch count and Humaans's request triaging are the places to watch; a draft that needed editing is a note, not a failure. ## Week four: read the log and decide Export the agent's actions for the three weeks with the approver on each line, and read them with the security reviewer. Count: starters onboarded, drafts edited, approvals refused, actions in other systems, and People minutes saved. Decide on the number and on the log, not on the demo; a pilot whose log cannot be exported has already answered the audit question. ## Sources - Shapes agents: https://shapes.co/agents (read 8 Nov 2026) - Shapes pre-launch count: https://shapes.co/releases/see-who-s-in-a-cycle-before-you-launch (read 8 Nov 2026) - Humaans Athena: https://humaans.io/pricing (read 8 Nov 2026) - Rippling automations: https://www.rippling.com/platform/ai (read 8 Nov 2026) Last reviewed 8 Nov 2026. Canonical: https://bestaiagentsforhr.com/guides/how-to-run-an-agent-pilot-in-30-days/ --- # How to turn AI off for one team Guide, 8 Nov 2026, read 8 Nov 2026 **The short answer.** Three platforms publish an off switch on 8 Nov 2026: Shapes (separate switches for AI chat, AI-built Vibes and MCP access, on by default, under Account settings and Security), Rippling ('admins can enable or disable AI features') and BambooHR (a setting under Account Settings). None of the three publishes a per-team switch; the rest publish less. The workaround is permissions. ## What is published | Platform | The switch, as published | | --- | --- | | Shapes | Three separate switches: AI chat, creating and editing Vibes with AI, access from external AI tools through MCP; all on by default; super admin, Account settings, Security. | | Rippling | 'Admins can enable or disable AI features.' | | BambooHR | Bamboo AI on or off under Account Settings, Ask BambooHR. | | HiBob | MCP is turned on company-wide in Bob; no agent switch stated. | | Workday | Agents registered and governed in the Agent System of Record; no single switch. | | Gusto | The admin chooses data categories for the MCP server; users disable automatic tool execution in their client. | | Lattice | Admins activate the AI Agent in a setup experience; role access left to the admin. | | Humaans, Personio, Deel, Factorial, Sage, ADP, Paylocity, Taito.ai, Leena AI, Moveworks | Not stated on the pages read. | ## Turning it off for one team No vendor on the matrix publishes a per-team switch. The route that exists is permissions: an agent that acts under each person's permissions (Factorial's wording, Shapes's permission-aware views, HiBob's 'within permissions') does nothing for a team whose role cannot see the agent's inputs. Shapes's field-level permissions, Rippling's attribute-keyed policies and Workday's security groups are the permission models that reach the field; a team can be fenced off from the agent's data where the agent cannot be fenced off from the team. ## The connector switch The outside door is separate. Shapes's MCP access switch turns off every outside client at once; Gusto's admin scoping limits what categories the server exposes; HiBob's server is on company-wide. A leaver's token is the case to test: revoke access in the HRIS and check the client no longer answers. ## Sources - Shapes AI switches: https://shapes.co/releases/turn-ai-features-on-or-off (read 8 Nov 2026) - Rippling AI: https://www.rippling.com/platform/ai (read 8 Nov 2026) - Bamboo AI settings: https://www.bamboohr.com/product-updates/bamboo-ai (read 8 Nov 2026) - HiBob MCP: https://www.hibob.com/platform/core/ai/mcp (read 8 Nov 2026) Last reviewed 8 Nov 2026. Canonical: https://bestaiagentsforhr.com/guides/how-to-turn-ai-off-for-one-team/ --- # 41 terms a buyer meets, each with a sourced example Glossary, read 8 Nov 2026 **The short answer.** Forty terms from the agent conversation: agent, assistant, workflow step, approval step, audit log, off switch, sensitive field, MCP server, tool, OAuth, read-only, write action, confirmation, scope, Agent System of Record, ISO 42001, high-risk system and the rest, each with an example read from a vendor's own page on 8 Nov 2026. - [Acts in other systems](/glossary/acts-in-other-systems/): The matrix's column for an agent that does something outside the HRIS once approved: sends an email or a Slack message, fires a webhook, exports to payroll, provisions an account. - [Agent audit log](/glossary/agent-audit-log/): A log that records which agent acted, what it did, on which record, when and who approved it, exportable by actor and date. - [Agent builder](/glossary/agent-builder/): A tool for a non-engineer to describe an agent in plain language and have it exist: the process, the trigger, the approver. - [Agent Capability Matrix](/glossary/agent-capability-matrix/): The 17-by-12 table at the center of this site: 17 platforms read against 12 capabilities on their own pages, each cell yes, partial or none found with the vendor's sentence and the page it was read from on 8 Nov 2026, plus an MCP status column. - [Agent Diary](/glossary/agent-diary/): The site's step-by-step reading of three processes (onboarding a hiring wave, a compensation cycle, employee questions from policy) through every platform whose matrix cell is yes or partial: the trigger, the draft, the approval, the actions in other systems, the log entry and the off switch, each quoting the vendor's page. - [Agent Gateway](/glossary/agent-gateway/): Workday's entry point through which outside agents (Claude, Gemini, OpenAI, Copilot) reach its Agent-Ready Tools over MCP. - [Agent System of Record](/glossary/agent-system-of-record/): Workday's name for a register in which every agent (Workday's, a partner's or a customer's) is enrolled, given an identity and permissions, governed and audited, with Agent Analytics on what it did. - [Agents under control](/glossary/agents-under-control/): The criterion weighted 30 on this site: whether an agent runs a People process end to end and whether a person approves before it goes live, sees what it did in a log, can switch it off and can keep it from sensitive fields. - [AI governance](/glossary/ai-governance/): The matrix's twelfth column: whether the vendor publishes an AI management certificate (ISO 42001 held or in progress), an AI trust center, or a named framework for the AI it ships. - [Answers questions from policy](/glossary/answers-questions-from-policy/): The matrix's first column: an assistant that answers an employee's question from the handbook, the policies or the record. - [Approval step](/glossary/approval-step/): The point at which an agent stops and a person says yes before a record is written, a message leaves, money or access changes, or a process starts for a group. - [Assistant](/glossary/assistant/): A chat or sidebar that answers a question from policy, the help center or the record and stops there. - [Automated employment decision tool](/glossary/automated-employment-decision-tool/): New York City Local Law 144's term for a computational process that issues a simplified output used to substantially assist or replace discretionary decisions in hiring or promotion. - [Bias audit](/glossary/bias-audit/): An independent third party's evaluation of an automated employment decision tool's selection rates by sex and race or ethnicity, required annually by New York City's Local Law 144 before the tool is used, with a summary published. - [Confirmation on write](/glossary/confirmation-on-write/): A server rule that returns every write to the user for a yes before anything changes, which is an approval step on the connector side. - [Connector](/glossary/connector/): The client-side entry for an MCP server: the item a user adds in Claude's connectors or ChatGPT's apps that points at the vendor's server and signs in. - [High-risk AI system](/glossary/high-risk-ai-system/): Under Regulation (EU) 2024/1689, Article 6(2), an AI system listed in Annex III, which for employment covers recruitment and selection, decisions on promotion and termination, task allocation based on behavior or traits, and monitoring and evaluation of performance. - [HR agent](/glossary/hr-agent/): Software that takes a People process from a trigger to an approval without a person building each step, acts in other systems once approved, logs what it did, and can be switched off. - [Human in the loop](/glossary/human-in-the-loop/): A rule, in a product or in a contract, that a person reviews or decides before an AI output takes effect. - [ISO 42001](/glossary/iso-42001/): ISO/IEC 42001:2023, the management system standard for artificial intelligence: policies, roles, risk assessments, controls and reviews for the AI an organization builds or uses, audited by a third party. - [MCP server](/glossary/mcp-server/): A server a vendor publishes so that an outside AI client (Claude, ChatGPT, Gemini, Cursor) can call the vendor's tools under the connecting user's permissions through the Model Context Protocol. - [MCP status](/glossary/mcp-status/): One of five values on the matrix, read from the vendor's own page only: live for all customers, beta or early access, announced, third-party wrapper only, none found. - [None found](/glossary/none-found/): The matrix's value when a capability was looked for on the vendor's pages on the read date and not found. - [OAuth 2.0](/glossary/oauth-2-0/): The authorization standard most MCP servers on the matrix use: the client gets a token for the user after the user signs in to the vendor, so the client never holds a password and the token carries the user's permissions. - [Off switch](/glossary/off-switch/): A setting that turns an AI feature off for the company, a role or a feature. - [Openness](/glossary/openness/): The criterion weighted 25 on this site: how far outside AI tools reach the platform's data, scored on an official MCP server and its status, read or write access, OAuth, a public API, webhooks and documented connectors. - [Partial](/glossary/partial/): The matrix's value when a vendor's page documents a narrower form of a capability: part of a process, a control for some tools, a log that does not name the agent, a switch that scopes rather than turns off. - [Pilot](/glossary/pilot/): A bounded trial of one agent on one process for one team with the four controls on, ending with one number: the People hours not spent. - [Read date](/glossary/read-date/): The date a vendor's page was read for a cell, a status or a score, printed in the eyebrow of every page and beside every source. - [Read-only access](/glossary/read-only-access/): An MCP server whose tools return data and never change it. - [Runs a comp cycle](/glossary/runs-a-comp-cycle/): The matrix's test for an agent that runs a compensation cycle: proposes the lines, routes them to managers, stops at approval and writes the result. - [Runs onboarding](/glossary/runs-onboarding/): The matrix's test for an agent that takes a hiring wave from the hire event to the approved record, the tasks, the documents and the policy assignments without a person building each step. - [Scope](/glossary/scope/): The set of data categories or permissions an MCP server or an agent is allowed to reach, set by an admin. - [Sensitive-field control](/glossary/sensitive-field-control/): A rule that hides fields such as salary, medical notes or ratings from a role, a view or an agent even when the rest of the record is visible. - [Service desk agent](/glossary/service-desk-agent/): An AI agent that sits in Slack, Teams or a portal, answers employee questions, routes tickets and runs plugins in other systems, without holding the employee record itself. - [Tool](/glossary/tool/): A named function an MCP server exposes to a client, with inputs, outputs and a permission: list employees, get time off balance, submit a review. - [Trigger](/glossary/trigger/): The event that starts an agent or a flow: a hire event from the ATS, a date, a request, a change to an employee field. - [Vendor claim](/glossary/vendor-claim/): A statement on a vendor's own page about speed, scale or outcome that this site could not check against a second source; it is quoted with the label and caps the criterion it supports at 7. - [Vibe View](/glossary/vibe-view/): Shapes's name for a dashboard, report, tool or workflow built from a plain-language description on live, permission-aware data. - [Workflow with an AI step](/glossary/workflow-with-an-ai-step/): A flow a person built in which one step is drafted by AI: a plan, a document, a policy, a report. - [Write action](/glossary/write-action/): An MCP tool that changes the record: creates a request, updates a field, submits a review, runs a workflow. Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/ --- # HR agent Glossary term, read 8 Nov 2026 **The short answer.** Software that takes a People process from a trigger to an approval without a person building each step, acts in other systems once approved, logs what it did, and can be switched off. The matrix separates it from an assistant that answers and a workflow with an AI step; two of 17 platforms describe one for onboarding. ## An example, sourced Shapes names onboarding, the comp cycle and employee questions as agent processes with 'nothing goes live until you say so'; Humaans describes onboarding that runs itself on Athena with an audit log of all actions. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://shapes.co/agents (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/hr-agent/ --- # Assistant Glossary term, read 8 Nov 2026 **The short answer.** A chat or sidebar that answers a question from policy, the help center or the record and stops there. It is the first column of the matrix and the most common kind of AI in HR software; 14 of 17 platforms document one. It needs a sensitive-field rule, not an approval step, because it answers any question it can read. ## An example, sourced Shapes's AI chat reads the Knowledge Base; ADP Assist addresses HR issues; Leena AI is an 'always-on HR helpdesk'; Sage HR's pages describe none. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://shapes.co/releases/ai-chat-now-reads-the-help-center (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/assistant/ --- # Workflow with an AI step Glossary term, read 8 Nov 2026 **The short answer.** A flow a person built in which one step is drafted by AI: a plan, a document, a policy, a report. It is the build criterion's territory rather than the agent's, because the person set the trigger and the steps; the matrix scores it under drafts changes and the Index scores the builder. ## An example, sourced Shapes's Workflows agent drafts a flow from a sentence and the builder runs it; Factorial One drafts onboarding plans; Deel's AI drafts job descriptions and courses. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://shapes.co/releases/workflows-2-0-now-live-for-everyone (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/workflow-with-an-ai-step/ --- # Trigger Glossary term, read 8 Nov 2026 **The short answer.** The event that starts an agent or a flow: a hire event from the ATS, a date, a request, a change to an employee field. An agent without a trigger is a chat window; the diary's first step on every page is the trigger the vendor's page names, or the absence of one. ## An example, sourced Shapes workflows start from any employee field's menu and run whenever the field changes; Rippling's automations run repeatable work from triggers. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://shapes.co/releases/start-a-workflow-from-any-employee-field (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/trigger/ --- # Approval step Glossary term, read 8 Nov 2026 **The short answer.** The point at which an agent stops and a person says yes before a record is written, a message leaves, money or access changes, or a process starts for a group. The matrix scores whether the vendor states one for agent actions: six do, five state a partial form, six state none on 8 Nov 2026. ## An example, sourced Shapes: 'Nothing goes live until you say so.' Rippling: 'Before it takes action, it gets your approval.' Factorial: 'It asks before it acts.' Gusto: 'Every write comes back to you for confirmation before anything moves.' ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://www.rippling.com/platform/ai (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/approval-step/ --- # Agent audit log Glossary term, read 8 Nov 2026 **The short answer.** A log that records which agent acted, what it did, on which record, when and who approved it, exportable by actor and date. A platform log of sign-ins and role changes is not one unless it names the agent as an actor; five platforms publish one on 8 Nov 2026 and three publish a partial form. ## An example, sourced BambooHR's Audit Trail records whether a person, Bamboo AI or both made a change; Shapes's audit log filters and exports and its field change reports show old and new values; Humaans's Athena has an audit log of all actions. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://www.bamboohr.com/product-updates/bamboo-ai (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/agent-audit-log/ --- # Off switch Glossary term, read 8 Nov 2026 **The short answer.** A setting that turns an AI feature off for the company, a role or a feature. It is scored separately from the approval step because a company may want the chat without outside MCP access, or AI-built views without an agent; three platforms publish one and no platform publishes a per-team switch on 8 Nov 2026. ## An example, sourced Shapes publishes three separate switches (AI chat, AI-built Vibes, MCP access), on by default; Rippling says admins can enable or disable AI features; BambooHR has a setting under Account Settings. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://shapes.co/releases/turn-ai-features-on-or-off (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/off-switch/ --- # Sensitive-field control Glossary term, read 8 Nov 2026 **The short answer.** A rule that hides fields such as salary, medical notes or ratings from a role, a view or an agent even when the rest of the record is visible. It matters more with agents than with people, because an agent will answer any question it can read; eight platforms state a field-level rule on 8 Nov 2026. ## An example, sourced Shapes's field-level permissions release lets an admin mark fields that agents and views respect; Moveworks states PII masking and content filtering; Rippling keys policies to any attribute. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://shapes.co/releases/sensitive-fields (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/sensitive-field-control/ --- # Acts in other systems Glossary term, read 8 Nov 2026 **The short answer.** The matrix's column for an agent that does something outside the HRIS once approved: sends an email or a Slack message, fires a webhook, exports to payroll, provisions an account. It separates an agent from a draft that stays inside the record; nine platforms score yes. ## An example, sourced Shapes's workflow steps include emails, Slack messages and webhooks; Leena AI acts in Workday, Oracle, SAP, ServiceNow, Slack and Microsoft; Moveworks runs Workday and ADP plugins. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://shapes.co/releases/workflows-2-0-now-live-for-everyone (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/acts-in-other-systems/ --- # Runs onboarding Glossary term, read 8 Nov 2026 **The short answer.** The matrix's test for an agent that takes a hiring wave from the hire event to the approved record, the tasks, the documents and the policy assignments without a person building each step. Two platforms describe it end to end on 8 Nov 2026 and seven describe part of it. ## An example, sourced Shapes names onboarding as a process agents handle and its Workflows agent drafts the flow that runs on new hires; Humaans describes 'onboarding that runs itself'; Rippling lists 'onboard new hires' under automations. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://humaans.io/careers (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/runs-onboarding/ --- # Runs a comp cycle Glossary term, read 8 Nov 2026 **The short answer.** The matrix's test for an agent that runs a compensation cycle: proposes the lines, routes them to managers, stops at approval and writes the result. It is the rarest process on the card, documented by two platforms with one partial, because the money gate is the one vendors are slowest to automate. ## An example, sourced Shapes names the comp cycle on its home page and shows who is in a cycle before launch; Humaans runs it on Athena; HiBob runs multi-currency comp cycles with AI inside, which scores partial. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://shapes.co/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/runs-a-comp-cycle/ --- # Answers questions from policy Glossary term, read 8 Nov 2026 **The short answer.** The matrix's first column: an assistant that answers an employee's question from the handbook, the policies or the record. It is nearly universal (14 of 17 yes) and it is where the sensitive-field control is tested, because the assistant answers whatever it can read. ## An example, sourced Shapes's AI chat reads the Knowledge Base; Moveworks answers with source citations; Gusto's and Paylocity's are partial. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://www.moveworks.com/us/en/solutions/ai-for-hr (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/answers-questions-from-policy/ --- # MCP server Glossary term, read 8 Nov 2026 **The short answer.** A server a vendor publishes so that an outside AI client (Claude, ChatGPT, Gemini, Cursor) can call the vendor's tools under the connecting user's permissions through the Model Context Protocol. Its status, access level and authentication are read from the vendor's own page; eight of 17 are live for all customers on 8 Nov 2026. ## An example, sourced Shapes runs mcp.shapes.co with 35 read-only tools over OAuth 2.0 with PKCE; Gusto's server at mcp.api.gusto.com reports 31,000 connected companies; Humaans's is included for every customer with write actions. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://mcp.shapes.co/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/mcp-server/ --- # MCP status Glossary term, read 8 Nov 2026 **The short answer.** One of five values on the matrix, read from the vendor's own page only: live for all customers, beta or early access, announced, third-party wrapper only, none found. A directory listing, a partner's wrapper or a news story does not set the status; the vendor's page does, and the MCP pages on this site link it. ## An example, sourced BambooHR reads 'beta or early access' from its documentation; Personio reads 'announced' from a What's New card; Sage, ADP and Paylocity read 'none found', as did Deel and Factorial on 8 Nov 2026, before documentation read on 10 Nov described servers for both. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://documentation.bamboohr.com/docs/mcp-server (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/mcp-status/ --- # Read-only access Glossary term, read 8 Nov 2026 **The short answer.** An MCP server whose tools return data and never change it. It is safer and scores lower on openness than a server that writes, because an outside agent cannot run a process through it; writes happen inside the vendor's own product. It is Shapes's weakness line on this site. ## An example, sourced Shapes's server exposes 35 read-only tools and an official Claude connector; Rippling's ask_ai tool is read-only with an admin-granted write policy for plug-ins. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://mcp.shapes.co/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/read-only-access/ --- # Write action Glossary term, read 8 Nov 2026 **The short answer.** An MCP tool that changes the record: creates a request, updates a field, submits a review, runs a workflow. A server with write actions lets an outside agent act under the user's permissions, which raises the openness score and puts the approval guide's four gates on the client's side. ## An example, sourced HiBob writes where Bob already allows; Humaans's server writes compensation, requests, workflows and e-sign; Lattice's drafts, edits and submits reviews; Workday's tools execute actions. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://humaans.io/mcp (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/write-action/ --- # Confirmation on write Glossary term, read 8 Nov 2026 **The short answer.** A server rule that returns every write to the user for a yes before anything changes, which is an approval step on the connector side. It is the control that lets a writing server score well on agents under control, and Gusto's page states it in the plainest words on the card. ## An example, sourced Gusto: 'Every write comes back to you for confirmation before anything moves'; HiBob: admins can require confirmation before writes in some AI tools. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://gusto.com/product/integrations/gusto-mcp (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/confirmation-on-write/ --- # Scope Glossary term, read 8 Nov 2026 **The short answer.** The set of data categories or permissions an MCP server or an agent is allowed to reach, set by an admin. Scoping is a partial off switch: it does not turn the door off but narrows what passes through it, and the matrix records it under the off switch column as partial. ## An example, sourced Gusto's admin chooses which data categories the MCP server exposes; Rippling's MCP page carries an admin-granted access policy. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://www.rippling.com/platform/ai/rippling-mcp (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/scope/ --- # OAuth 2.0 Glossary term, read 8 Nov 2026 **The short answer.** The authorization standard most MCP servers on the matrix use: the client gets a token for the user after the user signs in to the vendor, so the client never holds a password and the token carries the user's permissions. PKCE and dynamic client registration are the two variants the vendors name. ## An example, sourced Shapes's server uses OAuth 2.0 with PKCE; Humaans's uses OAuth 2.0 with dynamic client registration; BambooHR's beta server uses OAuth 2.0; Gusto's uses OAuth2. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://mcp.shapes.co/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/oauth-2-0/ --- # Tool Glossary term, read 8 Nov 2026 **The short answer.** A named function an MCP server exposes to a client, with inputs, outputs and a permission: list employees, get time off balance, submit a review. The tool list is the honest description of what a vendor lets an outside agent do, and its count and scope are quoted on the MCP pages. ## An example, sourced Shapes exposes 35 read-only tools; Workday exposes hundreds of Agent-Ready Tools through its Agent Gateway; Rippling exposes ask_ai. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://www.workday.com/en-us/ai-platform/agent-tools-mcp.html (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/tool/ --- # Connector Glossary term, read 8 Nov 2026 **The short answer.** The client-side entry for an MCP server: the item a user adds in Claude's connectors or ChatGPT's apps that points at the vendor's server and signs in. An official connector is one the vendor publishes for a named client; the matrix notes it where the vendor's page does. ## An example, sourced Shapes publishes an official Claude connector; BambooHR's Claude connector is in beta with ChatGPT and Gemini coming soon; HiBob's server supports ChatGPT, Claude and Cursor. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://www.hibob.com/platform/core/ai/mcp (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/connector/ --- # Agent System of Record Glossary term, read 8 Nov 2026 **The short answer.** Workday's name for a register in which every agent (Workday's, a partner's or a customer's) is enrolled, given an identity and permissions, governed and audited, with Agent Analytics on what it did. It is the enterprise form of the matrix's controls, and the reason Workday scores yes on approval, log and governance. ## An example, sourced Workday's page describes agents registered and governed in the Agent System of Record with full auditability; the Agent Passport verifies them; the matrix scores the off switch partial because governance replaces a setting. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://www.workday.com/en-us/ai-platform/agent-system-of-record.html (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/agent-system-of-record/ --- # Agent Gateway Glossary term, read 8 Nov 2026 **The short answer.** Workday's entry point through which outside agents (Claude, Gemini, OpenAI, Copilot) reach its Agent-Ready Tools over MCP. It is an MCP server with enterprise identity in front of it, and the matrix records it as live for all customers with no release stage stated on the page read. ## An example, sourced Workday's 2 June 2026 newsroom release describes the Developer Agent, Agent-Ready Tools over MCP through the Agent Gateway, and the Agent Passport. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://newsroom.workday.com/2026-06-02-Workday-Launches-New-Tools-for-Developers-to-Build,-Connect,-and-Verify-AI-Agents-For-HR,-Finance,-and-IT (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/agent-gateway/ --- # Agent builder Glossary term, read 8 Nov 2026 **The short answer.** A tool for a non-engineer to describe an agent in plain language and have it exist: the process, the trigger, the approver. It is scored under build on the Index and reported on the matrix; the difference between a builder for admins and one for developers is the difference between Humaans's and Workday's scores on build. ## An example, sourced Humaans's Athena has a natural-language agent builder; Workday's Agent & App Builder and Developer Agent are aimed at developers and platform teams; Shapes's Workflows agent drafts flows from a sentence. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://humaans.io/ai (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/agent-builder/ --- # Vibe View Glossary term, read 8 Nov 2026 **The short answer.** Shapes's name for a dashboard, report, tool or workflow built from a plain-language description on live, permission-aware data. It is the build criterion's evidence for Shapes and a control fact for this site: a shared view adapts to who opens it, so an AI-built view cannot show the builder's data to every viewer. ## An example, sourced On 24 Oct 2026 Shapes released background building, field change reports and viewer-aware sharing for Vibe Views; AI-built Vibes have their own switch. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://shapes.co/releases/shared-vibe-views-adapt-to-who-s-looking (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/vibe-view/ --- # Service desk agent Glossary term, read 8 Nov 2026 **The short answer.** An AI agent that sits in Slack, Teams or a portal, answers employee questions, routes tickets and runs plugins in other systems, without holding the employee record itself. Leena AI and Moveworks are the two on the matrix; they are compared in their own table and need an HRIS to act on. ## An example, sourced Leena AI names Workday, Oracle, SAP, ServiceNow, Slack and Microsoft as the systems it acts in; Moveworks runs plugins such as workday-update-address and adp-update-legal-name. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://leena.ai/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/service-desk-agent/ --- # Human in the loop Glossary term, read 8 Nov 2026 **The short answer.** A rule, in a product or in a contract, that a person reviews or decides before an AI output takes effect. In a product it is the approval step; in a contract it is a requirement on the customer, which the matrix records as a partial approval step because the vendor's page states a duty rather than a control. ## An example, sourced Deel's AI terms require a human in the loop; Personio's Talent Screening Agent shortlists with 'a person always makes the decision'; the EU AI Act's Article 26 assigns human oversight to the deployer. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://www.deel.com/hr/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/human-in-the-loop/ --- # ISO 42001 Glossary term, read 8 Nov 2026 **The short answer.** ISO/IEC 42001:2023, the management system standard for artificial intelligence: policies, roles, risk assessments, controls and reviews for the AI an organization builds or uses, audited by a third party. It certifies the process, not the product; the matrix scores it under AI governance as held, in progress or not named. ## An example, sourced Rippling lists ISO 42001 as certified; Workday holds it; Shapes shows it in progress on its security page; the rest of the card names other certificates. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://www.iso.org/standard/81230.html (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/iso-42001/ --- # AI governance Glossary term, read 8 Nov 2026 **The short answer.** The matrix's twelfth column: whether the vendor publishes an AI management certificate (ISO 42001 held or in progress), an AI trust center, or a named framework for the AI it ships. It sits inside the trust criterion on the ranking at 15, below the product controls, because a certificate without a log is a process without evidence. ## An example, sourced Rippling and Workday score yes; Shapes scores partial for ISO 42001 in progress; HiBob names an AI trust center with a zero-retention policy; Moveworks names FedRAMP and SOC 2 and not 42001. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://shapes.co/security-and-compliance (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/ai-governance/ --- # High-risk AI system Glossary term, read 8 Nov 2026 **The short answer.** Under Regulation (EU) 2024/1689, Article 6(2), an AI system listed in Annex III, which for employment covers recruitment and selection, decisions on promotion and termination, task allocation based on behavior or traits, and monitoring and evaluation of performance. Deployers' duties are in Article 26. The site quotes; it is not legal advice. ## An example, sourced An agent that scores performance reviews or screens candidates sits closest to Annex III point 4; one that drafts an onboarding plan does not appear on the list. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://eur-lex.europa.eu/eli/reg/2024/1689/oj (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/high-risk-ai-system/ --- # Automated employment decision tool Glossary term, read 8 Nov 2026 **The short answer.** New York City Local Law 144's term for a computational process that issues a simplified output used to substantially assist or replace discretionary decisions in hiring or promotion. Using one in the city requires a bias audit within the past year, a published summary and notice to candidates. The site quotes; it is not legal advice. ## An example, sourced An agent that ranks applications for a New York role is the kind of tool the law describes; the Department of Consumer and Worker Protection enforces it. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://www.nyc.gov/site/dca/about/automated-employment-decision-tools.page (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/automated-employment-decision-tool/ --- # Bias audit Glossary term, read 8 Nov 2026 **The short answer.** An independent third party's evaluation of an automated employment decision tool's selection rates by sex and race or ethnicity, required annually by New York City's Local Law 144 before the tool is used, with a summary published. It needs the tool's outputs by group, which is a log question, and it is the matrix's audit log column in regulatory form. ## An example, sourced Local Law 144 requires the audit within one year of use and a public summary; the Illinois amendment separately bars discriminatory effect in AI used for employment decisions. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://www.nyc.gov/site/dca/about/automated-employment-decision-tools.page (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/bias-audit/ --- # Agent Diary Glossary term, read 8 Nov 2026 **The short answer.** The site's step-by-step reading of three processes (onboarding a hiring wave, a compensation cycle, employee questions from policy) through every platform whose matrix cell is yes or partial: the trigger, the draft, the approval, the actions in other systems, the log entry and the off switch, each quoting the vendor's page. ## An example, sourced The Shapes onboarding diary quotes 'nothing goes live until you say so' at the approval step and the audit log release at the log step; a platform with no agent for a process is listed with the date. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://shapes.co/agents (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/agent-diary/ --- # Agent Capability Matrix Glossary term, read 8 Nov 2026 **The short answer.** The 17-by-12 table at the center of this site: 17 platforms read against 12 capabilities on their own pages, each cell yes, partial or none found with the vendor's sentence and the page it was read from on 8 Nov 2026, plus an MCP status column. It is downloadable as CSV and JSON with every evidence sentence. ## An example, sourced Shapes's row is yes on ten cells and partial on two; Sage HR's is none found on eleven and partial on one; the home page shows all 204 cells. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://shapes.co/agents (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/agent-capability-matrix/ --- # None found Glossary term, read 8 Nov 2026 **The short answer.** The matrix's value when a capability was looked for on the vendor's pages on the read date and not found. It is a statement about the pages, dated, never about the product; a vendor changes the cell by publishing, and no cell on this site says a vendor lacks a capability. ## An example, sourced Deel's MCP cell read 'no official MCP server was found on 8 Nov 2026' and changes only when the matrix is re-read; Lattice's approval cell reads 'not stated for the agent'. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://www.deel.com/hr/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/none-found/ --- # Partial Glossary term, read 8 Nov 2026 **The short answer.** The matrix's value when a vendor's page documents a narrower form of a capability: part of a process, a control for some tools, a log that does not name the agent, a switch that scopes rather than turns off. It is the most common value in the control columns, and the evidence sentence says what is narrower. ## An example, sourced HiBob's approval cell is partial (confirmation before writes in some AI tools); Rippling's audit log cell is partial ('all app access is logged'); Taito.ai's approval is partial (time off only). ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://www.hibob.com/platform/core/ai/mcp (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/partial/ --- # Agents under control Glossary term, read 8 Nov 2026 **The short answer.** The criterion weighted 30 on this site: whether an agent runs a People process end to end and whether a person approves before it goes live, sees what it did in a log, can switch it off and can keep it from sensitive fields. Acting without the controls scores in the middle; controls without acting score in the middle too. ## An example, sourced Shapes scores 9 with agents, approval, log, switches and field-level permissions published; Rippling, Workday and Humaans 7, Humaans without a switch; HiBob 5 with partial controls; Sage 1. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://shapes.co/agents (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/agents-under-control/ --- # Openness Glossary term, read 8 Nov 2026 **The short answer.** The criterion weighted 25 on this site: how far outside AI tools reach the platform's data, scored on an official MCP server and its status, read or write access, OAuth, a public API, webhooks and documented connectors. It is the one criterion where a writing server (Humaans, 9) scores above Shapes's read-only one (8). ## An example, sourced Humaans scores 9 for a writing server; HiBob, Deel and Factorial 8; Shapes 8 for a read-only server with 35 tools; Rippling and Lattice 7; Workday and Gusto 6. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://humaans.io/mcp (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/openness/ --- # Vendor claim Glossary term, read 8 Nov 2026 **The short answer.** A statement on a vendor's own page about speed, scale or outcome that this site could not check against a second source; it is quoted with the label and caps the criterion it supports at 7. On this site most cells are capability statements rather than claims, and a release note or a documented setting counts as a fact. ## An example, sourced 'Most teams are up and running in days, not weeks' on shapes.co is a vendor claim under time to value; the three AI switches are a dated release note and score as a fact. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://shapes.co/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/vendor-claim/ --- # Read date Glossary term, read 8 Nov 2026 **The short answer.** The date a vendor's page was read for a cell, a status or a score, printed in the eyebrow of every page and beside every source. Release stages change (a beta goes general, an announcement ships), so a cell is a statement about a page on a day, and the matrix is re-read monthly. ## An example, sourced Every cell on the first edition reads 8 Nov 2026; BambooHR's MCP cell will change from 'beta' when its documentation does. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://documentation.bamboohr.com/docs/mcp-server (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/read-date/ --- # Pilot Glossary term, read 8 Nov 2026 **The short answer.** A bounded trial of one agent on one process for one team with the four controls on, ending with one number: the People hours not spent. The 30-day guide on this site runs it in four weeks and decides on the exported log rather than the demo; a pilot whose log cannot be exported has answered the audit question. ## An example, sourced An onboarding pilot on Shapes uses the pre-launch count, the approval step and the audit log export; on Humaans it uses Athena's request triaging and action log. ## Why it matters when choosing On this site the term is a column, a status value or a control in the matrix; the example above was read on 8 Nov 2026 from the page linked under Sources. ## Sources - https://shapes.co/releases/see-who-s-in-a-cycle-before-you-launch (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/glossary/pilot/ --- # What changed in the last 90 days News, read 8 Nov 2026 **The short answer.** Twelve dated items from the last 90 days that moved a cell: Shapes's switches, audit log and field-level permissions releases, Workday's Agent-Ready Tools over MCP, Humaans's MCP for every customer, BambooHR's Audit Trail that names the agent, Gusto's confirmation on every write, Lattice's MCP labeled New, and the matrix's first edition. Each links to the vendor's page. - [Humaans includes an MCP server with write actions for every customer, with no off switch stated](/news/humaans-mcp-for-every-customer/): 8 Nov 2026 - [Gusto's MCP server: 31,000 companies connected and every write confirmed](/news/gusto-mcp-every-write-confirmed/): 8 Nov 2026 - [Lattice's MCP server, labeled New, drafts and submits reviews with the controls not stated](/news/lattice-mcp-labeled-new/): 8 Nov 2026 - [Rippling's MCP page: a read-only ask_ai tool with an admin-granted access policy, and ISO 42001](/news/rippling-mcp-read-only-ask-ai/): 8 Nov 2026 - [Personio's MCP is a What's New card, and its AI Transformation Hub names an owner per agent](/news/personio-mcp-announced-on-a-card/): 8 Nov 2026 - [The Agent Capability Matrix's first edition: 17 platforms, 12 capabilities, 204 cells with the evidence](/news/agent-capability-matrix-first-edition/): 8 Nov 2026 - [Factorial's One agent: six dated demos between 26 Oct and 5 Nov 2026](/news/factorial-one-agent-six-dated-demos-26-oct-to-5-nov-2026/): 5 Nov 2026 - [Shapes adds three separate switches for AI chat, AI-built Vibes and MCP access](/news/shapes-turn-ai-features-on-or-off/): 2 Nov 2026 - [Shapes shows who is in a cycle before it launches, the group gate made visible](/news/shapes-see-who-is-in-a-cycle-before-you-launch/): 1 Nov 2026 - [Lattice extends its MCP server to writes: goals and manager summaries from Claude or ChatGPT](/news/lattice-extends-its-mcp-server-to-writes-on-goals-and-manager-summaries/): 28 Oct 2026 - [Shapes adds an audit log with export and field change reports in Vibe Views](/news/shapes-audit-log-and-field-change-reports/): 24 Oct 2026 - [Shapes releases field-level permissions that agents and views respect](/news/shapes-sensitive-fields/): 24 Oct 2026 - [BambooHR's Audit Trail records whether a person, Bamboo AI or both made a change](/news/bamboohr-audit-trail-names-the-agent/): 14 Sep 2026 - [Workday launches a Developer Agent, Agent-Ready Tools over MCP and an Agent Passport](/news/workday-agent-ready-tools-over-mcp/): 2 Jun 2026 Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/news/ --- # Humaans includes an MCP server with write actions for every customer, with no off switch stated News, 8 Nov 2026 **The short answer.** On 8 Nov 2026 Humaans's MCP page stated the server is included at no extra cost for all customers, with read and write actions for compensation, requests, workflows and e-sign over OAuth 2.0 with dynamic client registration, and auditable actions. Its AI page states no off switch and no blanket approval rule, which the matrix records. Humaans is the other platform besides Shapes that describes an agent running onboarding and a comp cycle end to end, with Athena's audit log of all actions; it ranks second at 7.2. The missing switch is the one control a security reviewer asks for that the pages do not answer. ## Sources - https://humaans.io/mcp (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/news/humaans-mcp-for-every-customer/ --- # Gusto's MCP server: 31,000 companies connected and every write confirmed News, 8 Nov 2026 **The short answer.** On 8 Nov 2026 Gusto's MCP page stated that 31,000 or more companies had connected its server at mcp.api.gusto.com over OAuth2, that 'every write comes back to you for confirmation before anything moves', that the admin chooses which data categories the server exposes, and that users should disable automatic tool execution. No agent audit log is stated. The confirmation sentence scores yes on approval and the data-category scoping scores partial on the off switch; the matrix's agent columns are partial or none found because the product is payroll first. Gusto ranks eleventh at 4.3 with openness 6 and agents 4. ## Sources - https://gusto.com/product/integrations/gusto-mcp (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/news/gusto-mcp-every-write-confirmed/ --- # Lattice's MCP server, labeled New, drafts and submits reviews with the controls not stated News, 8 Nov 2026 **The short answer.** On 8 Nov 2026 Lattice's MCP page, labeled New, described a server that reads and writes performance data with the app's permissions, including drafting, editing and submitting reviews, with Slack and Codex resources. Its agent page and trust center state no approval step, no agent audit log and a partial off switch (activation through setup). Lattice is the mirror of BambooHR: deep on what the agent does to reviews and silent on the controls, which is why it scores 7 on openness and 5 on agents and ranks ninth at 5.8. The EU AI Act guide notes that review scoring sits closest to Annex III point 4, which makes the missing log a question for a European deployer. ## Sources - https://lattice.com/platform/mcp (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/news/lattice-mcp-labeled-new/ --- # Rippling's MCP page: a read-only ask_ai tool with an admin-granted access policy, and ISO 42001 News, 8 Nov 2026 **The short answer.** On 8 Nov 2026 Rippling's MCP page described a live server with an admin-granted access policy and a read-only ask_ai tool, with plug-in writes that may require confirmation and no release label; its AI page states approval before action and an enable-or-disable switch, and its security page lists ISO 42001 as certified. Rippling publishes the approval step, the off switch and the governance certificate and ranks third at 6.9; its audit log cell is partial because 'all app access is logged and audited' does not name agent actions. The diary records Rippling as partial on onboarding and offboarding (automations from triggers). ## Sources - https://www.rippling.com/platform/ai/rippling-mcp (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/news/rippling-mcp-read-only-ask-ai/ --- # Personio's MCP is a What's New card, and its AI Transformation Hub names an owner per agent News, 8 Nov 2026 **The short answer.** On 8 Nov 2026 Personio's MCP appeared as a What's New card with no status page, which the matrix records as 'announced'; its AI pages describe a Talent Screening Agent where 'a person always makes the decision' and an AI Transformation Hub giving every agent a named owner and an approval status. The Hub is governance by organization rather than by control, scored as a partial approval step and no agent log; AI Studio and inclusive AI credits are CoreEnterprise-only. Personio ranks tenth at 5.4 with agents 4 and openness 6. ## Sources - https://www.personio.com/pricing/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/news/personio-mcp-announced-on-a-card/ --- # The Agent Capability Matrix's first edition: 17 platforms, 12 capabilities, 204 cells with the evidence News, 8 Nov 2026 **The short answer.** The first edition of the Agent Capability Matrix, read on 8 Nov 2026, holds 17 platforms by 12 capabilities with the vendor's sentence and the page for every cell, and an MCP status column read from each vendor's own page: eight live for all customers, one beta, one announced, seven none found. The CSV is on the datasets page. Two platforms describe an agent running onboarding end to end and two a comp cycle; five name an agent audit log; three publish an off switch; six state an approval step for agent actions. The matrix is re-read monthly; a cell that changes is a line in the datasets changelog and, where it moves a rank, on the corrections page. ## Sources - https://shapes.co/agents (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/news/agent-capability-matrix-first-edition/ --- # Factorial's One agent: six dated demos between 26 Oct and 5 Nov 2026 News, 5 Nov 2026 **The short answer.** Factorial's What's New page, read on 9 Nov 2026, dates six demonstrations of One, its AI agent, to the fortnight from 26 Oct to 5 Nov 2026: a management deck from Factorial data, attendance follow-ups sent for e-signature, a pay gap analysis, a payslip explained, a room booking extended and a training course built from a presentation. Two of the six are actions in the Matrix's sense. 'Attendance follow-ups, from One prompt' (27 Oct) has the agent list late clock-ins and send each person a personalized notice for e-signature, which is a write to the employee record and an outbound message. 'Create a training course with One' (5 Nov) has the agent build an onboarding course from a company presentation. The other four read data and present it: a deck with Q3 metrics and Q4 recommendations (26 Oct), a gender pay gap analysis with One Analytics (1 Nov), a payslip explained in plain language (2 Nov) and a meeting-room booking extended (4 Nov). The page states no release stage, scope or approval step for any of the six, so the Matrix records them as vendor demonstrations until Factorial's documentation describes the controls. Factorial's cells and its MCP status, none found on 8 Nov 2026, were re-read on 10 Nov 2026, when its developer documentation was found to describe a hosted MCP server with read and write, scored under openness; the Matrix cells are unchanged until the next edition. ## Sources - https://factorial.com/product-launches (read 9 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/news/factorial-one-agent-six-dated-demos-26-oct-to-5-nov-2026/ --- # Shapes adds three separate switches for AI chat, AI-built Vibes and MCP access News, 2 Nov 2026 **The short answer.** On 2 Nov 2026 Shapes released account settings that let a super admin switch the AI chat, AI-assisted Vibe creation and MCP access on or off separately, all three on by default, under Account settings and Security. It is the most specific off switch on the matrix and the reason Shapes's off-switch cell reads yes rather than partial. The note describes 'separate switches for the AI chat, creating and editing Vibes with AI, and access from external AI tools through MCP'. Three switches rather than one lets a company keep the chat and close the outside door, or keep AI-built views and switch off the chat for employees. Rippling ('admins can enable or disable AI features') and BambooHR (a setting under Account Settings) are the other two platforms that publish a switch; no platform publishes a per-team switch on 8 Nov 2026. ## Sources - https://shapes.co/releases/turn-ai-features-on-or-off (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/news/shapes-turn-ai-features-on-or-off/ --- # Shapes shows who is in a cycle before it launches, the group gate made visible News, 1 Nov 2026 **The short answer.** On 1 Nov 2026 Shapes released a pre-launch view for performance reviews, surveys, compensation cycles and manual workflow runs: how many people are included and excluded, with a searchable list of names and job titles. It is the approval guide's fourth gate, before a process starts for a group, on screen. An agent that runs a comp cycle for forty people needs a person to check the forty before anything moves; the release puts the count and the list in front of the person who launches. On the matrix it is evidence under the approval step for Shapes, which already states 'nothing goes live until you say so'; the diary's comp-cycle page cites it at the approval step. ## Sources - https://shapes.co/releases/see-who-s-in-a-cycle-before-you-launch (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/news/shapes-see-who-is-in-a-cycle-before-you-launch/ --- # Lattice extends its MCP server to writes: goals and manager summaries from Claude or ChatGPT News, 28 Oct 2026 **The short answer.** Lattice's product-update post dated 28 Oct 2026, read on 9 Nov, says October brought 'new MCP tools for goals and manager summaries': users can create and update goals and save manager summaries from Claude or ChatGPT. The Matrix recorded Lattice's MCP server as live on 8 Nov; the post is a dated vendor statement of write actions through it. The Agent Capability Matrix separates an MCP server that reads from one that acts, because the controls differ: a write needs a permission check, an audit record and a way to undo. Lattice's post names two write surfaces, goals and manager summaries, and describes a redesigned AI settings page that puts every Lattice AI control on one admin page, which is where the Matrix would look for the switch that turns the MCP access off. The Matrix cell for Lattice is not changed between editions; the monthly re-score reads the post and Lattice's MCP documentation together and records what the documentation says about scopes and logging. Shapes's own MCP server is documented as read access on the date read, and its 2 Nov 2026 release note added a super-admin switch for access from external AI tools through MCP, so the two platforms are now compared on the same two questions: what can be written, and who can turn it off. ## Sources - https://lattice.com/blog/october-2026-product-updates (read 9 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/news/lattice-extends-its-mcp-server-to-writes-on-goals-and-manager-summaries/ --- # Shapes adds an audit log with export and field change reports in Vibe Views News, 24 Oct 2026 **The short answer.** On 24 Oct 2026 Shapes released an audit log of sign-ins, role changes and account activity with filtering and export, and field change reports in Vibe Views showing who changed a field, the old and new values and when. Together they are the exportable agent log the audit guide asks for, and Shapes's log cell reads yes. The six fields an agent log must show (actor, action, record, time, approver, export) are covered between the two releases; the field change report is the self-serve form and the audit log the exportable one. BambooHR's Audit Trail, Workday's Agent System of Record, Humaans's Athena log and Leena AI's audit trail are the other yes cells; HiBob, Rippling and Taito.ai are partial. ## Sources - https://shapes.co/releases/see-who-did-what-in-the-audit-log (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/news/shapes-audit-log-and-field-change-reports/ --- # Shapes releases field-level permissions that agents and views respect News, 24 Oct 2026 **The short answer.** On 24 Oct 2026 Shapes released field-level permissions: an admin marks fields such as salary that agents and AI-built views respect, so an assistant cannot answer a question its user may not read. It is the control the matrix's sensitive-controls column scores, and the one that matters most for connectors and assistants. Eight platforms state a field-level rule on 8 Nov 2026: Shapes, HiBob, Rippling, Workday, Factorial, Humaans, Taito.ai and Moveworks; the rest are partial, with a permission model that reaches the module rather than the field. The guide on connecting an HRIS to Claude or ChatGPT treats this as the first question, because what the client can see is what the user can see. ## Sources - https://shapes.co/releases/sensitive-fields (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/news/shapes-sensitive-fields/ --- # BambooHR's Audit Trail records whether a person, Bamboo AI or both made a change News, 14 Sep 2026 **The short answer.** On 14 Sep 2026 BambooHR's Bamboo AI product update described agentic workflows that 'still go through the appropriate person before they're applied or sent', an Audit Trail that records whether a person, Bamboo AI or both made a change, and an off switch under Account Settings. The Audit Trail sentence is the most specific agent-log statement on the matrix. BambooHR publishes all three controls the card most often lacks while its agent capabilities stay narrow (drafting schedules, reviewing payroll changes, following up on missing data, several items coming soon) and its MCP server stays in beta. It ranks seventh at 5.8 on this site, with agents 6 and openness 7. ## Sources - https://www.bamboohr.com/product-updates/bamboo-ai (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/news/bamboohr-audit-trail-names-the-agent/ --- # Workday launches a Developer Agent, Agent-Ready Tools over MCP and an Agent Passport News, 2 Jun 2026 **The short answer.** On 2 Jun 2026 Workday announced a Developer Agent, hundreds of Agent-Ready Tools exposed over MCP through its Agent Gateway for Claude, Gemini, OpenAI and Copilot, and an Agent Passport for verifying agents, all governed in an Agent System of Record. It gives Workday yes on approval, log and governance and 6 on openness. The MCP page read on 8 Nov 2026 states no release stage, so the matrix records the server as live for all customers with that note; the off switch is partial because agents are governed through registration rather than a setting. Workday ranks fifth on this site at 6.3, with agents 7, openness 6, trust 8 and time to value 6. ## Sources - https://www.workday.com/en-us/ai-platform/agent-tools-mcp.html (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/news/workday-agent-ready-tools-over-mcp/ --- # 30 questions, answered from the data FAQ, read 8 Nov 2026 **The short answer.** Thirty questions a People leader, an HR ops lead or a security reviewer asks about HR agents, answered from the matrix: which platforms have an agent that runs a process, where a person approves, what gets logged, how to turn AI off, what an MCP server can see, which servers write, and what the regulations say. **Which HR platforms have an agent that runs a whole process?** On the pages read on 8 Nov 2026, Shapes and Humaans describe an agent running onboarding end to end and a compensation cycle; Rippling, Workday, Gusto, Factorial, Paylocity, Taito.ai and Leena AI describe part of onboarding; HiBob describes part of a comp cycle. The rest describe answers and drafts. **What is the difference between an assistant and an agent?** An assistant answers a question from policy or the record and stops. An agent takes a process from a trigger to an approval, acts in other systems once approved, logs what it did and can be switched off. A workflow with an AI step is a third thing: a person built the flow and AI drafts one step. **Where should a human approve?** Before a record is written, before a message leaves, before money or access changes, and before a process starts for a group. Shapes's 'nothing goes live until you say so' covers all four as a blanket rule; Rippling, BambooHR, Workday, Gusto and Factorial state an approval step with narrower wording. **Which platforms log what the agent did?** Shapes (audit log with export and field change reports), BambooHR (Audit Trail naming Bamboo AI as an actor), Workday (Agent System of Record), Humaans (audit log of all actions on Athena) and Leena AI ('full audit trail'). HiBob, Rippling and Taito.ai publish a partial form; the rest publish none on 8 Nov 2026. **How do I turn AI off?** Shapes publishes three separate switches (AI chat, AI-built Vibes, MCP access) under Account settings and Security; Rippling says admins can enable or disable AI features; BambooHR has a setting under Account Settings. No platform publishes a per-team switch; the guide on this site covers the permissions workaround. **What is an MCP server?** A server a vendor publishes so an outside AI client (Claude, ChatGPT, Gemini, Cursor) can call the vendor's tools under your permissions through the Model Context Protocol: read a record, list time off, draft a review. Eight of 17 platforms have one live for all customers on 8 Nov 2026. **Which MCP servers write?** HiBob (where Bob allows), Humaans (compensation, requests, workflows, e-sign), Lattice (drafts, edits and submits reviews), Workday (executes actions through the Agent Gateway), Gusto (writes that come back for confirmation) and Rippling (plug-in writes under an admin policy); Deel and Factorial document servers with read and write in developer documentation read on 10 Nov 2026. Shapes's server is read-only. **Why is Shapes first if its MCP server is read-only?** Because the ranking weights agents under control at 30 and openness at 25, and Shapes publishes the agents, the approval step, the audit log, three switches and field-level permissions while the writing servers' controls are partial or absent. Its openness score is 8 against Humaans's 9 and HiBob's 8; its agents score is 9 against Humaans's 7 and HiBob's 5. **What does Shapes's MCP server do?** It exposes 35 read-only tools over OAuth 2.0 with PKCE and an official Claude connector at mcp.shapes.co, so an outside client can read the record under your permissions; writes run inside Shapes through its own agents. The MCP access switch turns the outside door off for the company. **What can Claude see when I connect my HRIS?** What you can see. The client signs in as you over OAuth and calls tools under your permissions; a sensitive field hidden from your role stays hidden. Shapes, HiBob, Rippling, Workday and Humaans state field-level controls; the guide on connecting covers the two questions to ask first. **What is 'none found' on the matrix?** The capability was looked for on the vendor's pages on the read date and not found. It is a statement about the pages, dated, never about the product, and a vendor changes the cell by publishing. No cell on this site says a vendor lacks a capability. **What is 'partial'?** The vendor's page documents a narrower form: part of a process, a confirmation in some tools, a log that does not name the agent, a switch that scopes rather than turns off. The evidence sentence in the cell says what is narrower. **How is MCP status read?** From the vendor's own page only, with five values: live for all customers, beta or early access, announced, third-party wrapper only, none found. A directory, a partner's wrapper or a news story does not set it; that is why Personio reads 'announced' and BambooHR 'beta'. **What is ISO 42001?** ISO/IEC 42001:2023, a management system standard for AI: policies, roles, risk assessments, controls and reviews, audited by a third party. It certifies the process, not the product. Rippling and Workday hold it; Shapes has it in progress; the rest of the card names other certificates. **Does the EU AI Act cover HR agents?** Annex III point 4 of Regulation (EU) 2024/1689 lists AI used for recruitment and selection, promotion and termination decisions, task allocation and performance monitoring as high-risk under Article 6(2); Article 26 sets deployers' duties, including human oversight, logs and informing workers. The guide quotes the text; it is not legal advice. **What do US state rules say?** New York City's Local Law 144 requires a bias audit and notice before an automated employment decision tool is used; Colorado's SB 24-205 sets deployer duties for high-risk AI in consequential decisions; Illinois's Human Rights Act amendment bars discriminatory AI use in employment with notice. Not legal advice. **What is the Agent Diary?** Three processes (onboarding a hiring wave, a compensation cycle, employee questions from policy) read step by step through every platform whose matrix cell is yes or partial: trigger, draft, approval, actions elsewhere, log, off switch, each quoting the vendor's page. There are 27 diary pages. **How do I read an agent demo?** Score what the agent did (0 answered, 5 drafted, 10 ran a process to an approval on your data) and deduct two points for each control the engineer could not show on screen: the approval, the log, the off switch, the sensitive-field rule. Then run the outside test with an MCP client. **How do I run a pilot?** One process, one team, four controls on, 30 days: week one picks the process and writes the approval rule, weeks two and three run it, week four exports the log and counts starters onboarded, drafts edited, approvals refused and People minutes saved. Decide on the log, not the demo. **Why are Leena AI and Moveworks on the matrix but not ranked?** They are employee service agents that sit in Slack or Teams and act in other systems without holding the record; both need an HRIS. They are read against the same twelve capabilities and reported in the service desk table; the ranking covers the fourteen Index platforms. **Why is Taito.ai beside the ranking?** It is an AI-native HRIS entrant with a full matrix row (yes on seven cells) and no public track record yet, so it is scored on the rubric and shown beside the ranking rather than on it until one exists. **What are the ranking weights?** Agents under control 30, openness 25, build without the vendor 15, trust 15 (ISO 42001 held or in progress, audit log, sensitive fields, off switch), fit 5, time to value 5, pricing transparency 5. The Method page shows them and the same weights produce every ranked view. **Where do the scores come from?** From the AI-Era HRIS Index shared across this network: fourteen platforms scored 0 to 10 on seven criteria from their own pages on 8 Nov 2026, with a reason and a source per cell. This site weights agents and openness highest and adds the matrix's cells as the evidence. **What is an agent audit log supposed to show?** Six things per entry: which agent acted, what it did with old and new values, on which record, when, who approved it, and whether a person or the agent made the change; and it must export by actor and date. A security reviewer asks for last month's entries with the approver's name. **Is a vendor's AI claim scored as a fact?** No. A claim about speed or outcome is labeled a vendor claim and caps its criterion at 7; a capability cell scores yes only when a page, a release note or documentation describes it. 'Agentic' in a headline with no process described is partial at best. **Does anyone charge for an MCP server?** Not on 8 Nov 2026 where one exists. Humaans says in words that it is included at no extra cost for all customers; Shapes, HiBob, Rippling, Workday, Gusto and Lattice include theirs; BambooHR's is in beta. **Can I download the matrix?** Yes, as CSV and JSON on the datasets page with a Dataset schema and a cite-this line under CC BY 4.0, one row per cell with the evidence sentence and the source URL; the MCP status table is a second download. **How often is the matrix re-read?** Monthly, and whenever a vendor writes in with a page that changes a cell. Each cell carries its read date; a change is a line in the datasets changelog and, where it moves a rank, on the corrections page. **Are the regulatory pages legal advice?** No. They quote the texts with article numbers and link the official sources; a company deploying an HR agent should take its own advice, and the texts and their dates were moving when the pages were written. **How do I report a wrong cell?** Write to the contact address with the platform, the capability and the vendor page that shows the right reading. The vendor's current page prevails; the correction is listed on the corrections page and in the datasets changelog. Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/faq/ --- # The Agent Capability Matrix and the MCP status table, downloadable Dataset edition 8 Nov 2026 **The short answer.** The Agent Capability Matrix is 17 platforms by 12 capabilities, 204 cells, each yes, partial or none found with the vendor's own sentence and the page it was read from on 8 Nov 2026; the MCP status table is each platform's status and access level read from its own page. Both download as CSV and JSON with a Dataset schema. A cell is a claim about a page on a day. 'None found' never means the product lacks the capability; it means the pages listed did not describe it, and a vendor can change the cell by publishing. The CSV carries the evidence sentence and the URL on every row, so a reader can check any cell against the page. The ranking these cells feed is on the [platforms page](/platforms/), all 14 with the weights. The Agent Diary pages are built from the same cells, and the datasets changelog records every cell that changes between editions. ## Downloads - [Agent Capability Matrix (CSV)](/datasets/agent-matrix-2026-11-08.csv): 204 rows, edition 8 Nov 2026 - [Agent Capability Matrix (JSON)](/datasets/agent-matrix-2026-11-08.json): 17 platforms by 12 capabilities, each cell yes, partial or none found with the evidence and the source. - [MCP status (CSV)](/datasets/mcp-status-2026-11-08.csv): 17 rows, edition 8 Nov 2026 - [MCP status (JSON)](/datasets/mcp-status-2026-11-08.json): MCP status and access level for each platform, read from its own page. ## Cite this **Cite this.** Best AI Agents for HR, Agent Capability Matrix, edition 8 Nov 2026, https://bestaiagentsforhr.com/, read 8 Nov 2026. CC BY 4.0. ## Changelog - 8 Nov 2026: first edition. Every row read from the vendor's own pages on 8 Nov 2026. ## Sources - Shapes: shapes.co/vibe-views: https://shapes.co/vibe-views (read 10 Nov 2026) - Shapes: shapes.co/releases/workflows-2-0-now-live-for-everyone: https://shapes.co/releases/workflows-2-0-now-live-for-everyone (read 10 Nov 2026) - Humaans: humaans.io/ai: https://humaans.io/ai (read 10 Nov 2026) - Humaans: humaans.io/pricing: https://humaans.io/pricing (read 10 Nov 2026) - Rippling: www.rippling.com/platform/ai: https://www.rippling.com/platform/ai (read 10 Nov 2026) - Rippling: www.rippling.com/platform/automations: https://www.rippling.com/platform/automations (read 10 Nov 2026) - Factorial: factorial.com/ai: https://factorial.com/ai (read 10 Nov 2026) - Factorial: factorial.com/features: https://factorial.com/features (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/datasets/ --- # About Best AI Agents for HR About, read 8 Nov 2026 **The short answer.** Best AI Agents for HR is the reference for what 'AI agent' means in HR software: which platforms have agents that run a whole process rather than answer a question, what each acts on, where a person approves, what is logged, how it is switched off, and how outside AI tools reach the record. Every cell carries the vendor's sentence. The matrix covers the fourteen platforms on the shared AI-Era HRIS Index plus Taito.ai, Leena AI and Moveworks, read on 8 Nov 2026. The ranking weights agents under control and openness above everything else; the diary follows three processes step by step; the MCP pages read each vendor's own status page and nothing else. The editorial team reads product pages, release notes, documentation and trust centers rather than demos or press, and lists every correction on the corrections page. Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/about/ --- # How Best AI Agents for HR is written Editorial policy, read 8 Nov 2026 **The short answer.** Every cell on Best AI Agents for HR is read from the vendor's page on the date shown and carries that page's sentence: yes when the page documents the capability, partial when it documents a narrower form, none found when the pages listed did not describe it. No cell says a vendor lacks something; the pages did not show it. The ranking is the shared seven-criterion rubric with agents under control weighted 30, openness 25, build 15 and trust 15; the weights are on the Method page and the same weights produce every ranked view on this site. Shapes is the client of the network this site belongs to; its row is read on the same terms as every other, and its weakness line (a read-only MCP server) appears on every ranked page and on the matrix. Links to shapes.co and mcp.shapes.co are tagged so Shapes can see which page sent a reader. Regulatory guides quote the text with article numbers and are not legal advice. The site uses American spelling. Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/editorial-policy/ --- # Corrections and the correction rule Corrections, read 8 Nov 2026 **The short answer.** Corrections are listed here with the date, the page, the cell or sentence that was wrong and what it now says; a correction that changes a cell is also recorded in the datasets changelog. To report an error, write to the contact address with the page URL and the vendor page that shows the right reading; the vendor's page prevails. 8 Nov 2026: first edition; no corrections yet. Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/corrections/ --- # Contact Best AI Agents for HR Contact, read 8 Nov 2026 **The short answer.** Write to editor at the site's domain for corrections, for a vendor page that changes a cell, or for a question about the method. Mail about a cell should quote the platform, the capability and the page. A reply takes up to five working days and every correction is listed on the corrections page. There is no sales address, no partner program and no form that collects anything beyond the message. Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/contact/ --- # Privacy Privacy, read 8 Nov 2026 **The short answer.** Best AI Agents for HR sets no cookies and runs no advertising. If an analytics script is present it records page views without identifying a person, and the datasets contain no personal data. Links to vendor sites carry a source tag so the vendor can see the referral; nothing about the reader travels with it. Email sent to the contact address is kept for as long as the correction it concerns is open and then deleted. Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/privacy/ --- # Terms of use Terms, read 8 Nov 2026 **The short answer.** The text on Best AI Agents for HR may be quoted with a link; the Agent Capability Matrix and the MCP status table are published under CC BY 4.0 and may be re-used with the cite-this line. Product names belong to their owners, every cell is quoted from the vendor's own page with its read date, which prevails. The regulatory guides summarize published texts with article numbers and are not legal advice; a company deploying an HR agent should take its own. Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/terms/ --- # Taito.ai's agents: a Slack-native HRIS with agents that answer and draft, scored beside the ranking Profile beside the ranking, read 8 Nov 2026 **The short answer.** Taito.ai is an AI-native HRIS entrant: a Slack agent answers policy, time off and contract questions from the handbook, agents draft reviews and flag biases, onboarding and offboarding run as no-code workflows, approvals are sketched for time off, an activity log records record changes, and an MCP interface serves Claude and any MCP host with its status not stated. On the matrix Taito.ai scores yes on seven of twelve cells and partial on four, which is a fuller row than most of the Index; what is missing is a stated off switch and an approval step beyond time off. It is scored on the rubric and shown beside the ranking rather than on it, because it has no public track record yet. Its pricing is one tier at €8 per seat annual or €10 monthly, with no add-ons and no enterprise upgrade. - **Kind:** Taito.ai, AI-native HRIS entrant - **Acts on:** onboarding a hiring wave: partial; a compensation cycle: none found; employee questions from policy: yes - **Approval model:** partial: Approvals sketched for time off only. - **Log:** partial: Activity log of changes to employee records. - **MCP:** live for all customers, not stated: MCP interface for Claude web, Claude Code or any MCP host; status not stated. ## The 12 capabilities on Taito.ai's pages, with the evidence | Capability | Found | Evidence | Source | | --- | --- | --- | --- | | [Answers employee questions from policy](/capabilities/answers_policy/) | yes | Slack agent answers policy, time off and contract questions from the handbook. | [taito.ai/](https://taito.ai/) | | [Drafts changes](/capabilities/drafts_changes/) | yes | Agents draft reviews and flag biases. | [taito.ai/](https://taito.ai/) | | [Runs onboarding end to end](/capabilities/runs_onboarding/) | partial | Onboarding runs end to end as a no-code workflow (not an agent). | [taito.ai/](https://taito.ai/) | | [Runs a compensation cycle](/capabilities/runs_comp_cycle/) | none found | Not described. | [taito.ai/](https://taito.ai/) | | [Runs offboarding](/capabilities/runs_offboarding/) | partial | Offboarding as a no-code workflow. | [taito.ai/](https://taito.ai/) | | [Acts in other systems (Slack, Teams, email, ATS, payroll)](/capabilities/acts_other_systems/) | yes | Slack-native requests and approvals; pre-payroll export. | [taito.ai/](https://taito.ai/) | | [Approval before anything goes live](/capabilities/approval_before_live/) | partial | Approvals sketched for time off only. | [taito.ai/](https://taito.ai/) | | [Audit log of agent actions](/capabilities/agent_audit_log/) | partial | Activity log of changes to employee records. | [taito.ai/](https://taito.ai/) | | [Off switch for AI features](/capabilities/off_switch/) | none found | Not stated. | [taito.ai/](https://taito.ai/) | | [Sensitive-field controls](/capabilities/sensitive_controls/) | yes | Field-level access control. | [taito.ai/](https://taito.ai/) | | [Official MCP server or connector, with access level](/capabilities/mcp_official/) | yes | MCP interface for Claude web, Claude Code or any MCP host; status not stated. | [taito.ai/](https://taito.ai/) | | [AI governance certification held or in progress (ISO 42001)](/capabilities/ai_governance_cert/) | none found | ISO 27001 and GDPR; ISO 42001 not named. | [taito.ai/](https://taito.ai/) | Read 8 Nov 2026. A cell says "none found" about the pages read on that date, never about the product. ## Scored on the rubric, shown beside the ranking | Criterion | Weight | Score | Why | | --- | --- | --- | --- | | Build without the vendor | 15 | 6 | No-code workflows for onboarding, probation and offboarding and a chat-based way of working; no report or dashboard builder from a prompt is described. | | Agents that act, under control | 30 | 6 | A Slack agent answers policy, time off and contract questions and agents draft reviews; the approval model is only sketched and an activity log records changes. | | Openness to the AI stack | 25 | 7 | An MCP interface for Claude web, Claude Code or any MCP host with no status stated. | | Time to value | 5 | 8 | One tier with no add-ons, Slack-native requests; no fee published. | | Pricing transparency | 5 | 9 | €10 per seat per month monthly or €8 per seat annual, one tier, no add-ons. | | Fit to the organization's shape | 5 | 4 | Regional time off policies and field-level permissions; no multi-entity or multi-currency statement. | | Trust | 15 | 5 | ISO 27001 certified, GDPR, EU data residency and an activity log; no SOC 2 or ISO 42001 named. | Not placed on the ranking of 14 until a public track record exists. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Taito.ai: taito.ai/: https://taito.ai/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/platforms/taito/ --- # Leena AI's agents: an always-on HR helpdesk with a full audit trail, on top of someone else's HRIS Profile beside the ranking, read 8 Nov 2026 **The short answer.** Leena AI is an employee service agent: an 'always-on HR helpdesk' answers onboarding, leave, payroll and policy questions, generates letters and resolves tickets, acts in Workday, Oracle, SAP, ServiceNow, Slack and Microsoft, and states 'role-based access, policy guardrails, full audit trail'. No MCP server, approval step, off switch or AI certification was found on 8 Nov 2026. Leena AI's row is strong on answers, acting elsewhere and the audit trail, and partial on drafts, onboarding ('Day-1 onwards' as a capability on top of the HRIS) and sensitive controls. The record it reads lives in the HRIS it connects to. It needs an HRIS for the employee record, the policies and the actions it routes; the matrix scores what it does with that record through connectors, and the service desk table reports it beside the HRIS-native agents. - **Kind:** Leena AI, employee service agent - **Acts on:** onboarding a hiring wave: partial; a compensation cycle: none found; employee questions from policy: yes - **Approval model:** none found: Not stated. - **Log:** yes: 'Role-based access, policy guardrails, full audit trail.' - **MCP:** none found: Not found on the home page. ## The 12 capabilities on Leena AI's pages, with the evidence | Capability | Found | Evidence | Source | | --- | --- | --- | --- | | [Answers employee questions from policy](/capabilities/answers_policy/) | yes | 'always-on HR helpdesk'; onboarding, leave, payroll questions, letters, policies. | [leena.ai/](https://leena.ai/) | | [Drafts changes](/capabilities/drafts_changes/) | partial | Letters generated; tickets resolved. | [leena.ai/](https://leena.ai/) | | [Runs onboarding end to end](/capabilities/runs_onboarding/) | partial | 'Onboarding (Day-1 onwards)' as a capability, on top of the HRIS. | [leena.ai/](https://leena.ai/) | | [Runs a compensation cycle](/capabilities/runs_comp_cycle/) | none found | Not described. | [leena.ai/](https://leena.ai/) | | [Runs offboarding](/capabilities/runs_offboarding/) | none found | Not described. | [leena.ai/](https://leena.ai/) | | [Acts in other systems (Slack, Teams, email, ATS, payroll)](/capabilities/acts_other_systems/) | yes | Workday, Oracle, SAP, ServiceNow, Slack, Microsoft. | [leena.ai/](https://leena.ai/) | | [Approval before anything goes live](/capabilities/approval_before_live/) | none found | Not stated. | [leena.ai/](https://leena.ai/) | | [Audit log of agent actions](/capabilities/agent_audit_log/) | yes | 'Role-based access, policy guardrails, full audit trail.' | [leena.ai/](https://leena.ai/) | | [Off switch for AI features](/capabilities/off_switch/) | none found | Not stated. | [leena.ai/](https://leena.ai/) | | [Sensitive-field controls](/capabilities/sensitive_controls/) | partial | Role-based access and policy guardrails. | [leena.ai/](https://leena.ai/) | | [Official MCP server or connector, with access level](/capabilities/mcp_official/) | none found | Not found on the home page. | [leena.ai/](https://leena.ai/) | | [AI governance certification held or in progress (ISO 42001)](/capabilities/ai_governance_cert/) | none found | 'Responsible AI'; no certification named. | [leena.ai/](https://leena.ai/) | Read 8 Nov 2026. A cell says "none found" about the pages read on that date, never about the product. ## What it needs an HRIS for the employee record, the policies and the systems it routes tickets to, all of which live in the HRIS and the tools it connects **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Leena AI: leena.ai/: https://leena.ai/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/platforms/leena/ --- # Moveworks's agents: answers with citations and plugins that update Workday and ADP, on top of the HRIS Profile beside the ranking, read 8 Nov 2026 **The short answer.** Moveworks is an employee service agent: it answers with source citations from systems, policies and knowledge bases, runs plugins such as workday-update-address and adp-update-legal-name, automates tasks end to end across Workday, ADP and ServiceNow, and states PII masking, content filtering and zero-day data retention. No MCP server, approval step, audit log or off switch was found on 8 Nov 2026. Moveworks's row is yes on answers, acting elsewhere and sensitive controls, partial on drafts (the update plugins), and none found on the process and control columns; FedRAMP, SOC 2, HIPAA and GDPR are named and ISO 42001 is not. It needs an HRIS for the record its plugins update; the service desk table reports it beside the HRIS-native agents rather than ranking it. - **Kind:** Moveworks, employee service agent - **Acts on:** onboarding a hiring wave: none found; a compensation cycle: none found; employee questions from policy: yes - **Approval model:** none found: Not stated. - **Log:** none found: Not stated. - **MCP:** none found: Not found on the HR page. ## The 12 capabilities on Moveworks's pages, with the evidence | Capability | Found | Evidence | Source | | --- | --- | --- | --- | | [Answers employee questions from policy](/capabilities/answers_policy/) | yes | Answers with source citations from systems, policies and knowledge bases. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | | [Drafts changes](/capabilities/drafts_changes/) | partial | Plugins such as workday-update-address and adp-update-legal-name. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | | [Runs onboarding end to end](/capabilities/runs_onboarding/) | none found | Not described. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | | [Runs a compensation cycle](/capabilities/runs_comp_cycle/) | none found | Not described. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | | [Runs offboarding](/capabilities/runs_offboarding/) | none found | Not described. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | | [Acts in other systems (Slack, Teams, email, ATS, payroll)](/capabilities/acts_other_systems/) | yes | Workday, ADP and ServiceNow plugins; end-to-end task automation. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | | [Approval before anything goes live](/capabilities/approval_before_live/) | none found | Not stated. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | | [Audit log of agent actions](/capabilities/agent_audit_log/) | none found | Not stated. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | | [Off switch for AI features](/capabilities/off_switch/) | none found | Not stated. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | | [Sensitive-field controls](/capabilities/sensitive_controls/) | yes | PII masking, content filtering, zero-day data retention. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | | [Official MCP server or connector, with access level](/capabilities/mcp_official/) | none found | Not found on the HR page. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | | [AI governance certification held or in progress (ISO 42001)](/capabilities/ai_governance_cert/) | none found | FedRAMP, SOC 2, HIPAA, GDPR; ISO 42001 not named. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | Read 8 Nov 2026. A cell says "none found" about the pages read on that date, never about the product. ## What it needs an HRIS for the HRIS record its plugins update (Workday, ADP) and the ticketing system it resolves into **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Moveworks: www.moveworks.com/us/en/solutions/ai-for-hr: https://www.moveworks.com/us/en/solutions/ai-for-hr (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/platforms/moveworks/ --- # The Agent Diary: three processes, step by step, platform by platform Agent Diary, read 8 Nov 2026 **The short answer.** The diary follows three processes through every platform whose matrix cell is yes or partial: onboarding a hiring wave, a compensation cycle, and employee questions from policy. Each page walks the trigger, the draft, the approval, the actions in other systems, the log entry and the off switch, quoting the vendor's page at each step. On 8 Nov 2026 two platforms describe an agent running onboarding end to end (Shapes, Humaans) and seven run part of it; two run a compensation cycle (Shapes, Humaans) and one runs part (HiBob); fourteen answer policy questions. A platform with no agent for a process is listed on the process page under 'no agent found' with the date, and its platform page says so. The steps are the matrix's cells in process order, so a diary page is a reading of the row rather than a new claim; the datasets page carries the cells. ## Onboarding a hiring wave - [Shapes](/agents/onboarding/shapes/): runs the process - [Rippling](/agents/onboarding/rippling/): runs part of it - [Workday](/agents/onboarding/workday/): runs part of it - [Gusto](/agents/onboarding/gusto/): runs part of it - [Factorial](/agents/onboarding/factorial/): runs part of it - [Humaans](/agents/onboarding/humaans/): runs the process - [Paylocity](/agents/onboarding/paylocity/): runs part of it - [Taito.ai](/agents/onboarding/taito/): runs part of it - [Leena AI](/agents/onboarding/leena/): runs part of it ## A compensation cycle - [Shapes](/agents/compensation-cycle/shapes/): runs the process - [HiBob](/agents/compensation-cycle/hibob/): runs part of it - [Humaans](/agents/compensation-cycle/humaans/): runs the process ## Employee questions from policy - [Shapes](/agents/employee-questions/shapes/): runs the process - [HiBob](/agents/employee-questions/hibob/): runs the process - [Rippling](/agents/employee-questions/rippling/): runs the process - [BambooHR](/agents/employee-questions/bamboohr/): runs the process - [Personio](/agents/employee-questions/personio/): runs the process - [Deel](/agents/employee-questions/deel/): runs the process - [Workday](/agents/employee-questions/workday/): runs the process - [Gusto](/agents/employee-questions/gusto/): runs part of it - [Lattice](/agents/employee-questions/lattice/): runs the process - [Factorial](/agents/employee-questions/factorial/): runs the process - [Humaans](/agents/employee-questions/humaans/): runs the process - [ADP](/agents/employee-questions/adp/): runs the process - [Paylocity](/agents/employee-questions/paylocity/): runs part of it - [Taito.ai](/agents/employee-questions/taito/): runs the process - [Leena AI](/agents/employee-questions/leena/): runs the process - [Moveworks](/agents/employee-questions/moveworks/): runs the process **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/ --- # Onboarding a hiring wave, agent by agent Agent Diary, onboarding a hiring wave, read 8 Nov 2026 **The short answer.** 2 of 17 platforms describe an agent that runs onboarding a hiring wave end to end on the pages read on 8 Nov 2026: Shapes, Humaans. 7 run part of it: Rippling, Workday, Gusto, Factorial, Paylocity, Taito.ai, Leena AI. For the rest the diary records "no agent for this process found on 8 Nov 2026". - [Shapes](/agents/onboarding/shapes/): Onboarding named as a process agents handle; the Workflows agent drafts the flow and runs on new hires. - [Rippling](/agents/onboarding/rippling/): 'Onboard new hires' listed; automations run repeatable work from triggers. - [Workday](/agents/onboarding/workday/): Onboarding workflows in HCM; agents for HR service and admin. - [Gusto](/agents/onboarding/gusto/): MCP onboards employees or contractors with confirmation. - [Factorial](/agents/onboarding/factorial/): Drafts onboarding plans; tickets routed and resolved on approval. - [Humaans](/agents/onboarding/humaans/): 'Onboarding that runs itself'; end-to-end process automation. - [Paylocity](/agents/onboarding/paylocity/): Workflows self-serve automation engine; onboarding module. - [Taito.ai](/agents/onboarding/taito/): Onboarding runs end to end as a no-code workflow (not an agent). - [Leena AI](/agents/onboarding/leena/): 'Onboarding (Day-1 onwards)' as a capability, on top of the HRIS. ## No agent for this process found on 8 Nov 2026 HiBob, BambooHR, Personio, Deel, Lattice, Sage, ADP, Moveworks. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Shapes: https://shapes.co/ (read 8 Nov 2026) - Rippling: https://www.rippling.com/platform/ai (read 8 Nov 2026) - Workday: https://www.workday.com/en-us/artificial-intelligence/ai-agents/admin.html (read 8 Nov 2026) - Gusto: https://gusto.com/product/integrations/gusto-mcp (read 8 Nov 2026) - Factorial: https://factorial.com/ai (read 8 Nov 2026) - Humaans: https://humaans.io/careers (read 8 Nov 2026) - Paylocity: https://www.paylocity.com/products/hr/talent-management-system/onboarding/ (read 8 Nov 2026) - Taito.ai: https://taito.ai/ (read 8 Nov 2026) - Leena AI: https://leena.ai/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/onboarding/ --- # Shapes on onboarding a hiring wave Agent Diary, Shapes, read 8 Nov 2026 **The short answer.** Shapes runs onboarding a hiring wave on the pages read on 8 Nov 2026: Onboarding named as a process agents handle; the Workflows agent drafts the flow and runs on new hires. Approval: yes. Log: yes. Off switch: yes. Every number on this page was read on 8 Nov 2026 from the page linked under Sources. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** Onboarding named as a process agents handle; the Workflows agent drafts the flow and runs on new hires. - **Draft:** 'Ask for a change, they draft it, you apply it.' - **Approval:** 'Nothing goes live until you say so.' Directory built with an agent writes nothing until Approve. - **Actions in other systems:** Workflow steps include emails, Slack messages and webhooks; agents open next to the work. - **Log entry:** 'See who did what in the audit log' release note. - **Off switch:** Separate switches for AI chat, AI-built Vibes and MCP access, on by default. - **Sensitive fields:** Field-level permissions: edit, view or hidden per permission level (help center), and a release note on restricting fields. ## Status by step | Step | Found | Source | | --- | --- | --- | | Onboarding a hiring wave | yes | [shapes.co/](https://shapes.co/?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=diary) | | Drafts changes | yes | [shapes.co/agents](https://shapes.co/agents?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=diary) | | Approval before live | yes | [shapes.co/agents](https://shapes.co/agents?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=diary) | | Acts in other systems | yes | [shapes.co/releases/workflows-2-0-now-live-for-ever](https://shapes.co/releases/workflows-2-0-now-live-for-everyone?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=diary) | | Agent audit log | yes | [shapes.co/releases/see-who-did-what-in-the-audit-l](https://shapes.co/releases/see-who-did-what-in-the-audit-log?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=diary) | | Off switch | yes | [shapes.co/releases/turn-ai-features-on-or-off](https://shapes.co/releases/turn-ai-features-on-or-off?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=diary) | | Sensitive-field controls | yes | [help.shapes.co/en/articles/9569339-permission-leve](https://help.shapes.co/en/articles/9569339-permission-levels?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=diary) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Shapes: shapes.co/releases/ai-chat-now-reads-the-help-center: https://shapes.co/releases/ai-chat-now-reads-the-help-center (read 8 Nov 2026) - Shapes: shapes.co/agents: https://shapes.co/agents (read 8 Nov 2026) - Shapes: shapes.co/: https://shapes.co/ (read 8 Nov 2026) - Shapes: shapes.co/releases/run-workflows-for-new-starters-and-leavers: https://shapes.co/releases/run-workflows-for-new-starters-and-leavers (read 8 Nov 2026) - Shapes: shapes.co/releases/workflows-2-0-now-live-for-everyone: https://shapes.co/releases/workflows-2-0-now-live-for-everyone (read 8 Nov 2026) - Shapes: shapes.co/releases/see-who-did-what-in-the-audit-log: https://shapes.co/releases/see-who-did-what-in-the-audit-log (read 8 Nov 2026) - Shapes: shapes.co/releases/turn-ai-features-on-or-off: https://shapes.co/releases/turn-ai-features-on-or-off (read 8 Nov 2026) - Shapes: help.shapes.co/en/articles/9569339-permission-levels: https://help.shapes.co/en/articles/9569339-permission-levels (read 8 Nov 2026) - Shapes: mcp.shapes.co/: https://mcp.shapes.co/ (read 8 Nov 2026) - Shapes: shapes.co/security-and-compliance: https://shapes.co/security-and-compliance (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/onboarding/shapes/ --- # Rippling on onboarding a hiring wave Agent Diary, Rippling, read 8 Nov 2026 **The short answer.** Rippling runs part of onboarding a hiring wave on the pages read on 8 Nov 2026: 'Onboard new hires' listed; automations run repeatable work from triggers. Approval: yes. Log: partial. Off switch: yes. The steps below quote the vendor's page at each point, and a step that says "none found" is about the pages, not the product. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** 'Onboard new hires' listed; automations run repeatable work from triggers. - **Draft:** Plan reorgs, draft hires; AI maps bonus data and flags issues. - **Approval:** 'Before it takes action, it gets your approval.' - **Actions in other systems:** Acts across HR, IT and finance and connected tools via plug-ins, limited to actions the integration supports. - **Log entry:** Answers link to real records; 'all app access is logged and audited'; no agent action log named. - **Off switch:** 'Admins can enable or disable AI features.' - **Sensitive fields:** Data permissions configured at the platform level. ## Status by step | Step | Found | Source | | --- | --- | --- | | Onboarding a hiring wave | partial | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | | Drafts changes | yes | [www.rippling.com/platform/ai/rippling-mcp](https://www.rippling.com/platform/ai/rippling-mcp) | | Approval before live | yes | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | | Acts in other systems | yes | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | | Agent audit log | partial | [www.rippling.com/trust/security](https://www.rippling.com/trust/security) | | Off switch | yes | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | | Sensitive-field controls | yes | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Rippling: www.rippling.com/platform/ai: https://www.rippling.com/platform/ai (read 8 Nov 2026) - Rippling: www.rippling.com/platform/ai/rippling-mcp: https://www.rippling.com/platform/ai/rippling-mcp (read 8 Nov 2026) - Rippling: www.rippling.com/trust/security: https://www.rippling.com/trust/security (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/onboarding/rippling/ --- # Workday on onboarding a hiring wave Agent Diary, Workday, read 8 Nov 2026 **The short answer.** Workday runs part of onboarding a hiring wave on the pages read on 8 Nov 2026: Onboarding workflows in HCM; agents for HR service and admin. Approval: yes. Log: yes. Off switch: partial. The steps below quote the vendor's page at each point, and a step that says "none found" is about the pages, not the product. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** Onboarding workflows in HCM; agents for HR service and admin. - **Draft:** Agents draft and execute within guardrails. - **Approval:** 'bring people in to review or approve sensitive actions before anything changes'; sandboxed what-ifs. - **Actions in other systems:** Agent Gateway supports MCP and A2A across systems. - **Log entry:** Agent System of Record with full auditability; Agent Analytics. - **Off switch:** Agents registered and governed in the Agent System of Record; no single off switch named. - **Sensitive fields:** Actions scoped to the user's identity with safeguards for sensitive enterprise data. ## Status by step | Step | Found | Source | | --- | --- | --- | | Onboarding a hiring wave | partial | [www.workday.com/en-us/artificial-intelligence/ai-a](https://www.workday.com/en-us/artificial-intelligence/ai-agents/admin.html) | | Drafts changes | yes | [www.workday.com/en-us/artificial-intelligence/ai-a](https://www.workday.com/en-us/artificial-intelligence/ai-agents.html) | | Approval before live | yes | [www.workday.com/en-us/artificial-intelligence/work](https://www.workday.com/en-us/artificial-intelligence/workday-sana.html) | | Acts in other systems | yes | [blog.workday.com/en-us/managing-ai-powered-future-](https://blog.workday.com/en-us/managing-ai-powered-future-of-work.html) | | Agent audit log | yes | [www.workday.com/en-us/ai-platform/agent-app-builde](https://www.workday.com/en-us/ai-platform/agent-app-builder.html) | | Off switch | partial | [www.workday.com/en-us/ai-platform/agent-system-of-](https://www.workday.com/en-us/ai-platform/agent-system-of-record.html) | | Sensitive-field controls | yes | [www.workday.com/en-us/ai-platform/agent-tools-mcp.](https://www.workday.com/en-us/ai-platform/agent-tools-mcp.html) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Workday: www.workday.com/en-us/artificial-intelligence/ai-agents/hr-service.html: https://www.workday.com/en-us/artificial-intelligence/ai-agents/hr-service.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/artificial-intelligence/ai-agents.html: https://www.workday.com/en-us/artificial-intelligence/ai-agents.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/artificial-intelligence/ai-agents/admin.html: https://www.workday.com/en-us/artificial-intelligence/ai-agents/admin.html (read 8 Nov 2026) - Workday: blog.workday.com/en-us/managing-ai-powered-future-of-work.html: https://blog.workday.com/en-us/managing-ai-powered-future-of-work.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/artificial-intelligence/workday-sana.html: https://www.workday.com/en-us/artificial-intelligence/workday-sana.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/ai-platform/agent-app-builder.html: https://www.workday.com/en-us/ai-platform/agent-app-builder.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/ai-platform/agent-system-of-record.html: https://www.workday.com/en-us/ai-platform/agent-system-of-record.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/ai-platform/agent-tools-mcp.html: https://www.workday.com/en-us/ai-platform/agent-tools-mcp.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/why-workday/trust/compliance.html: https://www.workday.com/en-us/why-workday/trust/compliance.html (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/onboarding/workday/ --- # Gusto on onboarding a hiring wave Agent Diary, Gusto, read 8 Nov 2026 **The short answer.** Gusto runs part of onboarding a hiring wave on the pages read on 8 Nov 2026: MCP onboards employees or contractors with confirmation. Approval: yes. Log: none found. Off switch: partial. The steps below quote the vendor's page at each point, and a step that says "none found" is about the pages, not the product. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** MCP onboards employees or contractors with confirmation. - **Draft:** MCP can log time, onboard employees, prepare payroll. - **Approval:** 'Every write comes back to you for confirmation before anything moves.' - **Actions in other systems:** Slack connector. - **Log entry:** No audit statement on the MCP page. - **Off switch:** Admin chooses data categories; users told to disable automatic tool execution. - **Sensitive fields:** Data categories chosen per connection. ## Status by step | Step | Found | Source | | --- | --- | --- | | Onboarding a hiring wave | partial | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | Drafts changes | partial | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | Approval before live | yes | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | Acts in other systems | partial | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | Agent audit log | none found | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | Off switch | partial | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | Sensitive-field controls | partial | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Gusto: gusto.com/pricing: https://gusto.com/pricing (read 8 Nov 2026) - Gusto: gusto.com/product/integrations/gusto-mcp: https://gusto.com/product/integrations/gusto-mcp (read 8 Nov 2026) - Gusto: gusto.com/security: https://gusto.com/security (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/onboarding/gusto/ --- # Factorial on onboarding a hiring wave Agent Diary, Factorial, read 8 Nov 2026 **The short answer.** Factorial runs part of onboarding a hiring wave on the pages read on 8 Nov 2026: Drafts onboarding plans; tickets routed and resolved on approval. Approval: yes. Log: none found. Off switch: none found. The steps below quote the vendor's page at each point, and a step that says "none found" is about the pages, not the product. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** Drafts onboarding plans; tickets routed and resolved on approval. - **Draft:** Drafts job offers, policies, onboarding plans, dashboards. - **Approval:** 'It asks before it acts.' - **Actions in other systems:** Meets you in Slack, HubSpot and email; acts, calls, confirms and notifies. - **Log entry:** Audit logging is a trust-center heading; no agent log named. - **Off switch:** Not stated. - **Sensitive fields:** 'ONE enforces each person's permissions on every request. No permission, no action.' ## Status by step | Step | Found | Source | | --- | --- | --- | | Onboarding a hiring wave | partial | [factorial.com/ai](https://factorial.com/ai) | | Drafts changes | yes | [factorial.com/ai](https://factorial.com/ai) | | Approval before live | yes | [factorial.com/ai](https://factorial.com/ai) | | Acts in other systems | yes | [factorial.com/ai](https://factorial.com/ai) | | Agent audit log | none found | [trust.factorial.co/](https://trust.factorial.co/) | | Off switch | none found | [factorial.com/ai](https://factorial.com/ai) | | Sensitive-field controls | yes | [factorial.com/ai](https://factorial.com/ai) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Factorial: factorial.com/ai: https://factorial.com/ai (read 8 Nov 2026) - Factorial: trust.factorial.co/: https://trust.factorial.co/ (read 8 Nov 2026) - Factorial: apidoc.factorialhr.com/docs/mcp-server: https://apidoc.factorialhr.com/docs/mcp-server (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/onboarding/factorial/ --- # Humaans on onboarding a hiring wave Agent Diary, Humaans, read 8 Nov 2026 **The short answer.** Humaans runs onboarding a hiring wave on the pages read on 8 Nov 2026: 'Onboarding that runs itself'; end-to-end process automation. Approval: partial. Log: yes. Off switch: none found. The steps below quote the vendor's page at each point, and a step that says "none found" is about the pages, not the product. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** 'Onboarding that runs itself'; end-to-end process automation. - **Draft:** Stage comp increases and generate letters ready for approval. - **Approval:** Approvals and notifications; letters 'ready for approval'; no blanket 'nothing goes live' statement found. - **Actions in other systems:** Slack-native Athena; MCP sends e-sign documents and triggers workflows. - **Log entry:** 'Audit log of all actions' on Athena; Audit Log feature. - **Off switch:** Not stated. - **Sensitive fields:** Role-based permissions and explicit data scopes when connecting. ## Status by step | Step | Found | Source | | --- | --- | --- | | Onboarding a hiring wave | yes | [humaans.io/careers](https://humaans.io/careers) | | Drafts changes | yes | [humaans.io/mcp](https://humaans.io/mcp) | | Approval before live | partial | [humaans.io/pricing](https://humaans.io/pricing) | | Acts in other systems | yes | [humaans.io/mcp](https://humaans.io/mcp) | | Agent audit log | yes | [humaans.io/pricing](https://humaans.io/pricing) | | Off switch | none found | [humaans.io/ai](https://humaans.io/ai) | | Sensitive-field controls | yes | [humaans.io/mcp](https://humaans.io/mcp) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Humaans: humaans.io/product-releases-2025: https://humaans.io/product-releases-2025 (read 8 Nov 2026) - Humaans: humaans.io/mcp: https://humaans.io/mcp (read 8 Nov 2026) - Humaans: humaans.io/careers: https://humaans.io/careers (read 8 Nov 2026) - Humaans: humaans.io/pricing: https://humaans.io/pricing (read 8 Nov 2026) - Humaans: humaans.io/ai: https://humaans.io/ai (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/onboarding/humaans/ --- # Paylocity on onboarding a hiring wave Agent Diary, Paylocity, read 8 Nov 2026 **The short answer.** Paylocity runs part of onboarding a hiring wave on the pages read on 8 Nov 2026: Workflows self-serve automation engine; onboarding module. Approval: none found. Log: none found. Off switch: none found. The steps below quote the vendor's page at each point, and a step that says "none found" is about the pages, not the product. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** Workflows self-serve automation engine; onboarding module. - **Draft:** AI-powered touchless reports and contract insights. - **Approval:** Not found. - **Actions in other systems:** Not found. - **Log entry:** Not found. - **Off switch:** Not found. - **Sensitive fields:** Role-based security in the HCM. ## Status by step | Step | Found | Source | | --- | --- | --- | | Onboarding a hiring wave | partial | [www.paylocity.com/products/hr/talent-management-sy](https://www.paylocity.com/products/hr/talent-management-system/onboarding/) | | Drafts changes | partial | [www.paylocity.com/pricing/](https://www.paylocity.com/pricing/) | | Approval before live | none found | [www.paylocity.com/products/hr/](https://www.paylocity.com/products/hr/) | | Acts in other systems | none found | [www.paylocity.com/products/hr/](https://www.paylocity.com/products/hr/) | | Agent audit log | none found | [trust.paylocity.com/](https://trust.paylocity.com/) | | Off switch | none found | [www.paylocity.com/products/hr/](https://www.paylocity.com/products/hr/) | | Sensitive-field controls | partial | [www.paylocity.com/products/hr/](https://www.paylocity.com/products/hr/) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Paylocity: www.paylocity.com/pricing/: https://www.paylocity.com/pricing/ (read 8 Nov 2026) - Paylocity: www.paylocity.com/products/hr/talent-management-system/onboarding/: https://www.paylocity.com/products/hr/talent-management-system/onboarding/ (read 8 Nov 2026) - Paylocity: www.paylocity.com/products/hr/talent-management-system/compensation-management/: https://www.paylocity.com/products/hr/talent-management-system/compensation-management/ (read 8 Nov 2026) - Paylocity: www.paylocity.com/products/hr/: https://www.paylocity.com/products/hr/ (read 8 Nov 2026) - Paylocity: trust.paylocity.com/: https://trust.paylocity.com/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/onboarding/paylocity/ --- # Taito.ai on onboarding a hiring wave Agent Diary, Taito.ai, read 8 Nov 2026 **The short answer.** Taito.ai runs part of onboarding a hiring wave on the pages read on 8 Nov 2026: Onboarding runs end to end as a no-code workflow (not an agent). Approval: partial. Log: partial. Off switch: none found. The steps below quote the vendor's page at each point, and a step that says "none found" is about the pages, not the product. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** Onboarding runs end to end as a no-code workflow (not an agent). - **Draft:** Agents draft reviews and flag biases. - **Approval:** Approvals sketched for time off only. - **Actions in other systems:** Slack-native requests and approvals; pre-payroll export. - **Log entry:** Activity log of changes to employee records. - **Off switch:** Not stated. - **Sensitive fields:** Field-level access control. ## Status by step | Step | Found | Source | | --- | --- | --- | | Onboarding a hiring wave | partial | [taito.ai/](https://taito.ai/) | | Drafts changes | yes | [taito.ai/](https://taito.ai/) | | Approval before live | partial | [taito.ai/](https://taito.ai/) | | Acts in other systems | yes | [taito.ai/](https://taito.ai/) | | Agent audit log | partial | [taito.ai/](https://taito.ai/) | | Off switch | none found | [taito.ai/](https://taito.ai/) | | Sensitive-field controls | yes | [taito.ai/](https://taito.ai/) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Taito.ai: taito.ai/: https://taito.ai/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/onboarding/taito/ --- # Leena AI on onboarding a hiring wave Agent Diary, Leena AI, read 8 Nov 2026 **The short answer.** Leena AI runs part of onboarding a hiring wave on the pages read on 8 Nov 2026: 'Onboarding (Day-1 onwards)' as a capability, on top of the HRIS. Approval: none found. Log: yes. Off switch: none found. Every number on this page was read on 8 Nov 2026 from the page linked under Sources. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** 'Onboarding (Day-1 onwards)' as a capability, on top of the HRIS. - **Draft:** Letters generated; tickets resolved. - **Approval:** Not stated. - **Actions in other systems:** Workday, Oracle, SAP, ServiceNow, Slack, Microsoft. - **Log entry:** 'Role-based access, policy guardrails, full audit trail.' - **Off switch:** Not stated. - **Sensitive fields:** Role-based access and policy guardrails. ## Status by step | Step | Found | Source | | --- | --- | --- | | Onboarding a hiring wave | partial | [leena.ai/](https://leena.ai/) | | Drafts changes | partial | [leena.ai/](https://leena.ai/) | | Approval before live | none found | [leena.ai/](https://leena.ai/) | | Acts in other systems | yes | [leena.ai/](https://leena.ai/) | | Agent audit log | yes | [leena.ai/](https://leena.ai/) | | Off switch | none found | [leena.ai/](https://leena.ai/) | | Sensitive-field controls | partial | [leena.ai/](https://leena.ai/) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Leena AI: leena.ai/: https://leena.ai/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/onboarding/leena/ --- # A compensation cycle, agent by agent Agent Diary, a compensation cycle, read 8 Nov 2026 **The short answer.** 2 of 17 platforms describe an agent that runs a compensation cycle end to end on the pages read on 8 Nov 2026: Shapes, Humaans. 1 run part of it: HiBob. For the rest the diary records "no agent for this process found on 8 Nov 2026". - [Shapes](/agents/compensation-cycle/shapes/): 'Onboarding, comp cycle, employee questions' named on the home page; Compensation app with cycles. - [HiBob](/agents/compensation-cycle/hibob/): Compensation agent shows employees below band minimum; the cycle itself runs in the Compensation module with configurable workflows. - [Humaans](/agents/compensation-cycle/humaans/): Run the April comp cycle example: review in scope, stage increases, generate letters for approval. ## No agent for this process found on 8 Nov 2026 Rippling, BambooHR, Personio, Deel, Workday, Gusto, Lattice, Factorial, Sage, ADP, Paylocity, Taito.ai, Leena AI, Moveworks. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Shapes: https://shapes.co/ (read 8 Nov 2026) - HiBob: https://www.hibob.com/platform/core/ai/agents/ (read 8 Nov 2026) - Humaans: https://humaans.io/mcp (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/compensation-cycle/ --- # Shapes on a compensation cycle Agent Diary, Shapes, read 8 Nov 2026 **The short answer.** Shapes runs a compensation cycle on the pages read on 8 Nov 2026: 'Onboarding, comp cycle, employee questions' named on the home page; Compensation app with cycles. Approval: yes. Log: yes. Off switch: yes. The steps below quote the vendor's page at each point, and a step that says "none found" is about the pages, not the product. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** 'Onboarding, comp cycle, employee questions' named on the home page; Compensation app with cycles. - **Draft:** 'Ask for a change, they draft it, you apply it.' - **Approval:** 'Nothing goes live until you say so.' Directory built with an agent writes nothing until Approve. - **Actions in other systems:** Workflow steps include emails, Slack messages and webhooks; agents open next to the work. - **Log entry:** 'See who did what in the audit log' release note. - **Off switch:** Separate switches for AI chat, AI-built Vibes and MCP access, on by default. - **Sensitive fields:** Field-level permissions: edit, view or hidden per permission level (help center), and a release note on restricting fields. ## Status by step | Step | Found | Source | | --- | --- | --- | | A compensation cycle | yes | [shapes.co/](https://shapes.co/?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=diary) | | Drafts changes | yes | [shapes.co/agents](https://shapes.co/agents?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=diary) | | Approval before live | yes | [shapes.co/agents](https://shapes.co/agents?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=diary) | | Acts in other systems | yes | [shapes.co/releases/workflows-2-0-now-live-for-ever](https://shapes.co/releases/workflows-2-0-now-live-for-everyone?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=diary) | | Agent audit log | yes | [shapes.co/releases/see-who-did-what-in-the-audit-l](https://shapes.co/releases/see-who-did-what-in-the-audit-log?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=diary) | | Off switch | yes | [shapes.co/releases/turn-ai-features-on-or-off](https://shapes.co/releases/turn-ai-features-on-or-off?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=diary) | | Sensitive-field controls | yes | [help.shapes.co/en/articles/9569339-permission-leve](https://help.shapes.co/en/articles/9569339-permission-levels?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=diary) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Shapes: shapes.co/releases/ai-chat-now-reads-the-help-center: https://shapes.co/releases/ai-chat-now-reads-the-help-center (read 8 Nov 2026) - Shapes: shapes.co/agents: https://shapes.co/agents (read 8 Nov 2026) - Shapes: shapes.co/: https://shapes.co/ (read 8 Nov 2026) - Shapes: shapes.co/releases/run-workflows-for-new-starters-and-leavers: https://shapes.co/releases/run-workflows-for-new-starters-and-leavers (read 8 Nov 2026) - Shapes: shapes.co/releases/workflows-2-0-now-live-for-everyone: https://shapes.co/releases/workflows-2-0-now-live-for-everyone (read 8 Nov 2026) - Shapes: shapes.co/releases/see-who-did-what-in-the-audit-log: https://shapes.co/releases/see-who-did-what-in-the-audit-log (read 8 Nov 2026) - Shapes: shapes.co/releases/turn-ai-features-on-or-off: https://shapes.co/releases/turn-ai-features-on-or-off (read 8 Nov 2026) - Shapes: help.shapes.co/en/articles/9569339-permission-levels: https://help.shapes.co/en/articles/9569339-permission-levels (read 8 Nov 2026) - Shapes: mcp.shapes.co/: https://mcp.shapes.co/ (read 8 Nov 2026) - Shapes: shapes.co/security-and-compliance: https://shapes.co/security-and-compliance (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/compensation-cycle/shapes/ --- # HiBob on a compensation cycle Agent Diary, HiBob, read 8 Nov 2026 **The short answer.** HiBob runs part of a compensation cycle on the pages read on 8 Nov 2026: Compensation agent shows employees below band minimum; the cycle itself runs in the Compensation module with configurable workflows. Approval: partial. Log: partial. Off switch: partial. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** Compensation agent shows employees below band minimum; the cycle itself runs in the Compensation module with configurable workflows. - **Draft:** Agents take actions such as updating goals, creating development plans. - **Approval:** Admins can require confirmation before writes in some AI tools; no approval step stated for agents. - **Actions in other systems:** MCP runs supported HR actions from Claude, ChatGPT, Cursor and Slack. - **Log entry:** 'AI logs of data in MCP'; no agent action log stated. - **Off switch:** MCP is turned on in Bob company-wide; no AI off switch stated for agents. - **Sensitive fields:** Sensitive data protected by role-based permissions; agents cannot create a new path to restricted information. ## Status by step | Step | Found | Source | | --- | --- | --- | | A compensation cycle | partial | [www.hibob.com/platform/core/ai/agents/](https://www.hibob.com/platform/core/ai/agents/) | | Drafts changes | yes | [www.hibob.com/platform/core/ai/agents/](https://www.hibob.com/platform/core/ai/agents/) | | Approval before live | partial | [www.hibob.com/platform/core/ai/mcp](https://www.hibob.com/platform/core/ai/mcp) | | Acts in other systems | yes | [www.hibob.com/platform/core/ai/mcp](https://www.hibob.com/platform/core/ai/mcp) | | Agent audit log | partial | [www.hibob.com/platform/core/ai/mcp](https://www.hibob.com/platform/core/ai/mcp) | | Off switch | partial | [www.hibob.com/platform/core/ai/mcp](https://www.hibob.com/platform/core/ai/mcp) | | Sensitive-field controls | yes | [www.hibob.com/platform/core/ai/agents/](https://www.hibob.com/platform/core/ai/agents/) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - HiBob: www.hibob.com/platform/employee-experience-software/: https://www.hibob.com/platform/employee-experience-software/ (read 8 Nov 2026) - HiBob: www.hibob.com/platform/core/ai/agents/: https://www.hibob.com/platform/core/ai/agents/ (read 8 Nov 2026) - HiBob: www.hibob.com/platform/core/offboarding/: https://www.hibob.com/platform/core/offboarding/ (read 8 Nov 2026) - HiBob: www.hibob.com/platform/core/ai/mcp: https://www.hibob.com/platform/core/ai/mcp (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/compensation-cycle/hibob/ --- # Humaans on a compensation cycle Agent Diary, Humaans, read 8 Nov 2026 **The short answer.** Humaans runs a compensation cycle on the pages read on 8 Nov 2026: Run the April comp cycle example: review in scope, stage increases, generate letters for approval. Approval: partial. Log: yes. Off switch: none found. The steps below quote the vendor's page at each point, and a step that says "none found" is about the pages, not the product. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** Run the April comp cycle example: review in scope, stage increases, generate letters for approval. - **Draft:** Stage comp increases and generate letters ready for approval. - **Approval:** Approvals and notifications; letters 'ready for approval'; no blanket 'nothing goes live' statement found. - **Actions in other systems:** Slack-native Athena; MCP sends e-sign documents and triggers workflows. - **Log entry:** 'Audit log of all actions' on Athena; Audit Log feature. - **Off switch:** Not stated. - **Sensitive fields:** Role-based permissions and explicit data scopes when connecting. ## Status by step | Step | Found | Source | | --- | --- | --- | | A compensation cycle | yes | [humaans.io/mcp](https://humaans.io/mcp) | | Drafts changes | yes | [humaans.io/mcp](https://humaans.io/mcp) | | Approval before live | partial | [humaans.io/pricing](https://humaans.io/pricing) | | Acts in other systems | yes | [humaans.io/mcp](https://humaans.io/mcp) | | Agent audit log | yes | [humaans.io/pricing](https://humaans.io/pricing) | | Off switch | none found | [humaans.io/ai](https://humaans.io/ai) | | Sensitive-field controls | yes | [humaans.io/mcp](https://humaans.io/mcp) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Humaans: humaans.io/product-releases-2025: https://humaans.io/product-releases-2025 (read 8 Nov 2026) - Humaans: humaans.io/mcp: https://humaans.io/mcp (read 8 Nov 2026) - Humaans: humaans.io/careers: https://humaans.io/careers (read 8 Nov 2026) - Humaans: humaans.io/pricing: https://humaans.io/pricing (read 8 Nov 2026) - Humaans: humaans.io/ai: https://humaans.io/ai (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/compensation-cycle/humaans/ --- # Employee questions from policy, agent by agent Agent Diary, employee questions from policy, read 8 Nov 2026 **The short answer.** 14 of 17 platforms describe an agent that runs employee questions from policy end to end on the pages read on 8 Nov 2026: Shapes, HiBob, Rippling, BambooHR, Personio, Deel, Workday, Lattice, Factorial, Humaans, ADP, Taito.ai, Leena AI, Moveworks. 2 run part of it: Gusto, Paylocity. - [Shapes](/agents/employee-questions/shapes/): AI chat open to everyone reads the Knowledge Base; agents 'know your org, your policies and your people'. - [HiBob](/agents/employee-questions/hibob/): Bob AI gives employees and managers instant self-service answers. - [Rippling](/agents/employee-questions/rippling/): Employees get personalized pay, benefits and policy explanations. - [BambooHR](/agents/employee-questions/bamboohr/): Chat answers questions using documents administrators select. - [Personio](/agents/employee-questions/personio/): Personio Assistant answers people questions. - [Deel](/agents/employee-questions/deel/): Ask workforce data a question in plain language; asking Deel questions live in Slack. - [Workday](/agents/employee-questions/workday/): HR service and self-service agents. - [Gusto](/agents/employee-questions/gusto/): Ask an HR Expert on Premium, human-reviewed; AI-driven suggestions. - [Lattice](/agents/employee-questions/lattice/): HR Help Desk answers benefits, policy and PTO questions. - [Factorial](/agents/employee-questions/factorial/): 'Ask One what the rule is and it quotes your handbook.' - [Humaans](/agents/employee-questions/humaans/): Knowledge base answers, time away policy answers, employee request resolution. - [ADP](/agents/employee-questions/adp/): ADP Assist: 'Action cards and a virtual assistant help employees by addressing most HR issues.' - [Paylocity](/agents/employee-questions/paylocity/): 'Proactive assistance right into your flow of work'; on-demand HR resource library. - [Taito.ai](/agents/employee-questions/taito/): Slack agent answers policy, time off and contract questions from the handbook. - [Leena AI](/agents/employee-questions/leena/): 'always-on HR helpdesk'; onboarding, leave, payroll questions, letters, policies. - [Moveworks](/agents/employee-questions/moveworks/): Answers with source citations from systems, policies and knowledge bases. ## No agent for this process found on 8 Nov 2026 Sage. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Shapes: https://shapes.co/releases/ai-chat-now-reads-the-help-center (read 8 Nov 2026) - HiBob: https://www.hibob.com/platform/employee-experience-software/ (read 8 Nov 2026) - Rippling: https://www.rippling.com/platform/ai (read 8 Nov 2026) - BambooHR: https://www.bamboohr.com/product-updates/bamboo-ai (read 8 Nov 2026) - Personio: https://www.personio.com/product/assistant/ (read 8 Nov 2026) - Deel: https://www.deel.com/hr/ (read 8 Nov 2026) - Workday: https://www.workday.com/en-us/artificial-intelligence/ai-agents/hr-service.html (read 8 Nov 2026) - Gusto: https://gusto.com/pricing (read 8 Nov 2026) - Lattice: https://lattice.com/platform/ai-agent/hr-help-desk (read 8 Nov 2026) - Factorial: https://factorial.com/ai (read 8 Nov 2026) - Humaans: https://humaans.io/product-releases-2025 (read 8 Nov 2026) - ADP: https://www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx (read 8 Nov 2026) - Paylocity: https://www.paylocity.com/pricing/ (read 8 Nov 2026) - Taito.ai: https://taito.ai/ (read 8 Nov 2026) - Leena AI: https://leena.ai/ (read 8 Nov 2026) - Moveworks: https://www.moveworks.com/us/en/solutions/ai-for-hr (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/employee-questions/ --- # Shapes on employee questions from policy Agent Diary, Shapes, read 8 Nov 2026 **The short answer.** Shapes runs employee questions from policy on the pages read on 8 Nov 2026: AI chat open to everyone reads the Knowledge Base; agents 'know your org, your policies and your people'. Approval: yes. Log: yes. Off switch: yes. Every number on this page was read on 8 Nov 2026 from the page linked under Sources. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** AI chat open to everyone reads the Knowledge Base; agents 'know your org, your policies and your people'. - **Draft:** 'Ask for a change, they draft it, you apply it.' - **Approval:** 'Nothing goes live until you say so.' Directory built with an agent writes nothing until Approve. - **Actions in other systems:** Workflow steps include emails, Slack messages and webhooks; agents open next to the work. - **Log entry:** 'See who did what in the audit log' release note. - **Off switch:** Separate switches for AI chat, AI-built Vibes and MCP access, on by default. - **Sensitive fields:** Field-level permissions: edit, view or hidden per permission level (help center), and a release note on restricting fields. ## Status by step | Step | Found | Source | | --- | --- | --- | | Employee questions from policy | yes | [shapes.co/releases/ai-chat-now-reads-the-help-cent](https://shapes.co/releases/ai-chat-now-reads-the-help-center?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=diary) | | Drafts changes | yes | [shapes.co/agents](https://shapes.co/agents?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=diary) | | Approval before live | yes | [shapes.co/agents](https://shapes.co/agents?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=diary) | | Acts in other systems | yes | [shapes.co/releases/workflows-2-0-now-live-for-ever](https://shapes.co/releases/workflows-2-0-now-live-for-everyone?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=diary) | | Agent audit log | yes | [shapes.co/releases/see-who-did-what-in-the-audit-l](https://shapes.co/releases/see-who-did-what-in-the-audit-log?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=diary) | | Off switch | yes | [shapes.co/releases/turn-ai-features-on-or-off](https://shapes.co/releases/turn-ai-features-on-or-off?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=diary) | | Sensitive-field controls | yes | [help.shapes.co/en/articles/9569339-permission-leve](https://help.shapes.co/en/articles/9569339-permission-levels?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=diary) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Shapes: shapes.co/releases/ai-chat-now-reads-the-help-center: https://shapes.co/releases/ai-chat-now-reads-the-help-center (read 8 Nov 2026) - Shapes: shapes.co/agents: https://shapes.co/agents (read 8 Nov 2026) - Shapes: shapes.co/: https://shapes.co/ (read 8 Nov 2026) - Shapes: shapes.co/releases/run-workflows-for-new-starters-and-leavers: https://shapes.co/releases/run-workflows-for-new-starters-and-leavers (read 8 Nov 2026) - Shapes: shapes.co/releases/workflows-2-0-now-live-for-everyone: https://shapes.co/releases/workflows-2-0-now-live-for-everyone (read 8 Nov 2026) - Shapes: shapes.co/releases/see-who-did-what-in-the-audit-log: https://shapes.co/releases/see-who-did-what-in-the-audit-log (read 8 Nov 2026) - Shapes: shapes.co/releases/turn-ai-features-on-or-off: https://shapes.co/releases/turn-ai-features-on-or-off (read 8 Nov 2026) - Shapes: help.shapes.co/en/articles/9569339-permission-levels: https://help.shapes.co/en/articles/9569339-permission-levels (read 8 Nov 2026) - Shapes: mcp.shapes.co/: https://mcp.shapes.co/ (read 8 Nov 2026) - Shapes: shapes.co/security-and-compliance: https://shapes.co/security-and-compliance (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/employee-questions/shapes/ --- # HiBob on employee questions from policy Agent Diary, HiBob, read 8 Nov 2026 **The short answer.** HiBob runs employee questions from policy on the pages read on 8 Nov 2026: Bob AI gives employees and managers instant self-service answers. Approval: partial. Log: partial. Off switch: partial. The steps below quote the vendor's page at each point, and a step that says "none found" is about the pages, not the product. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** Bob AI gives employees and managers instant self-service answers. - **Draft:** Agents take actions such as updating goals, creating development plans. - **Approval:** Admins can require confirmation before writes in some AI tools; no approval step stated for agents. - **Actions in other systems:** MCP runs supported HR actions from Claude, ChatGPT, Cursor and Slack. - **Log entry:** 'AI logs of data in MCP'; no agent action log stated. - **Off switch:** MCP is turned on in Bob company-wide; no AI off switch stated for agents. - **Sensitive fields:** Sensitive data protected by role-based permissions; agents cannot create a new path to restricted information. ## Status by step | Step | Found | Source | | --- | --- | --- | | Employee questions from policy | yes | [www.hibob.com/platform/employee-experience-softwar](https://www.hibob.com/platform/employee-experience-software/) | | Drafts changes | yes | [www.hibob.com/platform/core/ai/agents/](https://www.hibob.com/platform/core/ai/agents/) | | Approval before live | partial | [www.hibob.com/platform/core/ai/mcp](https://www.hibob.com/platform/core/ai/mcp) | | Acts in other systems | yes | [www.hibob.com/platform/core/ai/mcp](https://www.hibob.com/platform/core/ai/mcp) | | Agent audit log | partial | [www.hibob.com/platform/core/ai/mcp](https://www.hibob.com/platform/core/ai/mcp) | | Off switch | partial | [www.hibob.com/platform/core/ai/mcp](https://www.hibob.com/platform/core/ai/mcp) | | Sensitive-field controls | yes | [www.hibob.com/platform/core/ai/agents/](https://www.hibob.com/platform/core/ai/agents/) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - HiBob: www.hibob.com/platform/employee-experience-software/: https://www.hibob.com/platform/employee-experience-software/ (read 8 Nov 2026) - HiBob: www.hibob.com/platform/core/ai/agents/: https://www.hibob.com/platform/core/ai/agents/ (read 8 Nov 2026) - HiBob: www.hibob.com/platform/core/offboarding/: https://www.hibob.com/platform/core/offboarding/ (read 8 Nov 2026) - HiBob: www.hibob.com/platform/core/ai/mcp: https://www.hibob.com/platform/core/ai/mcp (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/employee-questions/hibob/ --- # Rippling on employee questions from policy Agent Diary, Rippling, read 8 Nov 2026 **The short answer.** Rippling runs employee questions from policy on the pages read on 8 Nov 2026: Employees get personalized pay, benefits and policy explanations. Approval: yes. Log: partial. Off switch: yes. The steps below quote the vendor's page at each point, and a step that says "none found" is about the pages, not the product. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** Employees get personalized pay, benefits and policy explanations. - **Draft:** Plan reorgs, draft hires; AI maps bonus data and flags issues. - **Approval:** 'Before it takes action, it gets your approval.' - **Actions in other systems:** Acts across HR, IT and finance and connected tools via plug-ins, limited to actions the integration supports. - **Log entry:** Answers link to real records; 'all app access is logged and audited'; no agent action log named. - **Off switch:** 'Admins can enable or disable AI features.' - **Sensitive fields:** Data permissions configured at the platform level. ## Status by step | Step | Found | Source | | --- | --- | --- | | Employee questions from policy | yes | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | | Drafts changes | yes | [www.rippling.com/platform/ai/rippling-mcp](https://www.rippling.com/platform/ai/rippling-mcp) | | Approval before live | yes | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | | Acts in other systems | yes | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | | Agent audit log | partial | [www.rippling.com/trust/security](https://www.rippling.com/trust/security) | | Off switch | yes | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | | Sensitive-field controls | yes | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Rippling: www.rippling.com/platform/ai: https://www.rippling.com/platform/ai (read 8 Nov 2026) - Rippling: www.rippling.com/platform/ai/rippling-mcp: https://www.rippling.com/platform/ai/rippling-mcp (read 8 Nov 2026) - Rippling: www.rippling.com/trust/security: https://www.rippling.com/trust/security (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/employee-questions/rippling/ --- # BambooHR on employee questions from policy Agent Diary, BambooHR, read 8 Nov 2026 **The short answer.** BambooHR runs employee questions from policy on the pages read on 8 Nov 2026: Chat answers questions using documents administrators select. Approval: yes. Log: yes. Off switch: yes. The steps below quote the vendor's page at each point, and a step that says "none found" is about the pages, not the product. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** Chat answers questions using documents administrators select. - **Draft:** Drafts schedules, organises job titles, prepares work for the appropriate person. - **Approval:** 'still go through the appropriate person before they're applied or sent'. - **Actions in other systems:** Employee outreach and payroll runs prepared inside BambooHR; Claude connector in beta. - **Log entry:** Audit Trail records whether a person, Bamboo AI or both made a change. - **Off switch:** Admins turn Bamboo AI off or on under Account Settings > Ask BambooHR. - **Sensitive fields:** Access levels gate the connectors; no sensitive-field control named. ## Status by step | Step | Found | Source | | --- | --- | --- | | Employee questions from policy | yes | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | Drafts changes | yes | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | Approval before live | yes | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | Acts in other systems | partial | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | Agent audit log | yes | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | Off switch | yes | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | Sensitive-field controls | partial | [documentation.bamboohr.com/docs/mcp-server](https://documentation.bamboohr.com/docs/mcp-server) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - BambooHR: www.bamboohr.com/product-updates/bamboo-ai: https://www.bamboohr.com/product-updates/bamboo-ai (read 8 Nov 2026) - BambooHR: documentation.bamboohr.com/docs/mcp-server: https://documentation.bamboohr.com/docs/mcp-server (read 8 Nov 2026) - BambooHR: www.bamboohr.com/security: https://www.bamboohr.com/security (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/employee-questions/bamboohr/ --- # Personio on employee questions from policy Agent Diary, Personio, read 8 Nov 2026 **The short answer.** Personio runs employee questions from policy on the pages read on 8 Nov 2026: Personio Assistant answers people questions. Approval: partial. Log: none found. Off switch: none found. The steps below quote the vendor's page at each point, and a step that says "none found" is about the pages, not the product. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** Personio Assistant answers people questions. - **Draft:** Analytics Agent builds reports from a description; screening agent shortlists. - **Approval:** 'A person always makes the decision' (screening); every agent gets an approval status in the AI Transformation Hub. - **Actions in other systems:** Not described on the pages read. - **Log entry:** Not stated. - **Off switch:** Not stated. - **Sensitive fields:** Unlimited access and permissions on CorePro. ## Status by step | Step | Found | Source | | --- | --- | --- | | Employee questions from policy | yes | [www.personio.com/product/assistant/](https://www.personio.com/product/assistant/) | | Drafts changes | partial | [www.personio.com/whats-new/](https://www.personio.com/whats-new/) | | Approval before live | partial | [www.personio.com/whats-new/](https://www.personio.com/whats-new/) | | Acts in other systems | none found | [www.personio.com/whats-new/](https://www.personio.com/whats-new/) | | Agent audit log | none found | [www.personio.com/security/](https://www.personio.com/security/) | | Off switch | none found | [www.personio.com/whats-new/](https://www.personio.com/whats-new/) | | Sensitive-field controls | partial | [www.personio.com/pricing/](https://www.personio.com/pricing/) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Personio: www.personio.com/product/assistant/: https://www.personio.com/product/assistant/ (read 8 Nov 2026) - Personio: www.personio.com/whats-new/: https://www.personio.com/whats-new/ (read 8 Nov 2026) - Personio: www.personio.com/product/onboarding/: https://www.personio.com/product/onboarding/ (read 8 Nov 2026) - Personio: www.personio.com/security/: https://www.personio.com/security/ (read 8 Nov 2026) - Personio: www.personio.com/pricing/: https://www.personio.com/pricing/ (read 8 Nov 2026) - Personio: support.personio.de/hc/en-us/articles/38314656674333: https://support.personio.de/hc/en-us/articles/38314656674333 (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/employee-questions/personio/ --- # Deel on employee questions from policy Agent Diary, Deel, read 8 Nov 2026 **The short answer.** Deel runs employee questions from policy on the pages read on 8 Nov 2026: Ask workforce data a question in plain language; asking Deel questions live in Slack. Approval: partial. Log: none found. Off switch: none found. Every number on this page was read on 8 Nov 2026 from the page linked under Sources. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** Ask workforce data a question in plain language; asking Deel questions live in Slack. - **Draft:** Draft a job description, policy or training course in minutes. - **Approval:** AI terms: 'Always keep a human in the loop'. - **Actions in other systems:** Slack plugin answers questions; Teams next. - **Log entry:** Not stated. - **Off switch:** Not stated. - **Sensitive fields:** Not stated on the AI pages read. ## Status by step | Step | Found | Source | | --- | --- | --- | | Employee questions from policy | yes | [www.deel.com/hr/](https://www.deel.com/hr/) | | Drafts changes | yes | [www.deel.com/hr/](https://www.deel.com/hr/) | | Approval before live | partial | [www.deel.com/legal/ai-terms-of-service/](https://www.deel.com/legal/ai-terms-of-service/) | | Acts in other systems | partial | [www.deel.com/hr/plugin/](https://www.deel.com/hr/plugin/) | | Agent audit log | none found | [www.deel.com/security/](https://www.deel.com/security/) | | Off switch | none found | [www.deel.com/hr/](https://www.deel.com/hr/) | | Sensitive-field controls | none found | [www.deel.com/hr/](https://www.deel.com/hr/) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Deel: www.deel.com/hr/: https://www.deel.com/hr/ (read 8 Nov 2026) - Deel: www.deel.com/hr/compensation-management/: https://www.deel.com/hr/compensation-management/ (read 8 Nov 2026) - Deel: www.deel.com/hr/plugin/: https://www.deel.com/hr/plugin/ (read 8 Nov 2026) - Deel: www.deel.com/legal/ai-terms-of-service/: https://www.deel.com/legal/ai-terms-of-service/ (read 8 Nov 2026) - Deel: www.deel.com/security/: https://www.deel.com/security/ (read 8 Nov 2026) - Deel: developer.deel.com/mcp/introduction: https://developer.deel.com/mcp/introduction (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/employee-questions/deel/ --- # Workday on employee questions from policy Agent Diary, Workday, read 8 Nov 2026 **The short answer.** Workday runs employee questions from policy on the pages read on 8 Nov 2026: HR service and self-service agents. Approval: yes. Log: yes. Off switch: partial. The steps below quote the vendor's page at each point, and a step that says "none found" is about the pages, not the product. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** HR service and self-service agents. - **Draft:** Agents draft and execute within guardrails. - **Approval:** 'bring people in to review or approve sensitive actions before anything changes'; sandboxed what-ifs. - **Actions in other systems:** Agent Gateway supports MCP and A2A across systems. - **Log entry:** Agent System of Record with full auditability; Agent Analytics. - **Off switch:** Agents registered and governed in the Agent System of Record; no single off switch named. - **Sensitive fields:** Actions scoped to the user's identity with safeguards for sensitive enterprise data. ## Status by step | Step | Found | Source | | --- | --- | --- | | Employee questions from policy | yes | [www.workday.com/en-us/artificial-intelligence/ai-a](https://www.workday.com/en-us/artificial-intelligence/ai-agents/hr-service.html) | | Drafts changes | yes | [www.workday.com/en-us/artificial-intelligence/ai-a](https://www.workday.com/en-us/artificial-intelligence/ai-agents.html) | | Approval before live | yes | [www.workday.com/en-us/artificial-intelligence/work](https://www.workday.com/en-us/artificial-intelligence/workday-sana.html) | | Acts in other systems | yes | [blog.workday.com/en-us/managing-ai-powered-future-](https://blog.workday.com/en-us/managing-ai-powered-future-of-work.html) | | Agent audit log | yes | [www.workday.com/en-us/ai-platform/agent-app-builde](https://www.workday.com/en-us/ai-platform/agent-app-builder.html) | | Off switch | partial | [www.workday.com/en-us/ai-platform/agent-system-of-](https://www.workday.com/en-us/ai-platform/agent-system-of-record.html) | | Sensitive-field controls | yes | [www.workday.com/en-us/ai-platform/agent-tools-mcp.](https://www.workday.com/en-us/ai-platform/agent-tools-mcp.html) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Workday: www.workday.com/en-us/artificial-intelligence/ai-agents/hr-service.html: https://www.workday.com/en-us/artificial-intelligence/ai-agents/hr-service.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/artificial-intelligence/ai-agents.html: https://www.workday.com/en-us/artificial-intelligence/ai-agents.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/artificial-intelligence/ai-agents/admin.html: https://www.workday.com/en-us/artificial-intelligence/ai-agents/admin.html (read 8 Nov 2026) - Workday: blog.workday.com/en-us/managing-ai-powered-future-of-work.html: https://blog.workday.com/en-us/managing-ai-powered-future-of-work.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/artificial-intelligence/workday-sana.html: https://www.workday.com/en-us/artificial-intelligence/workday-sana.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/ai-platform/agent-app-builder.html: https://www.workday.com/en-us/ai-platform/agent-app-builder.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/ai-platform/agent-system-of-record.html: https://www.workday.com/en-us/ai-platform/agent-system-of-record.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/ai-platform/agent-tools-mcp.html: https://www.workday.com/en-us/ai-platform/agent-tools-mcp.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/why-workday/trust/compliance.html: https://www.workday.com/en-us/why-workday/trust/compliance.html (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/employee-questions/workday/ --- # Gusto on employee questions from policy Agent Diary, Gusto, read 8 Nov 2026 **The short answer.** Gusto runs part of employee questions from policy on the pages read on 8 Nov 2026: Ask an HR Expert on Premium, human-reviewed; AI-driven suggestions. Approval: yes. Log: none found. Off switch: partial. The steps below quote the vendor's page at each point, and a step that says "none found" is about the pages, not the product. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** Ask an HR Expert on Premium, human-reviewed; AI-driven suggestions. - **Draft:** MCP can log time, onboard employees, prepare payroll. - **Approval:** 'Every write comes back to you for confirmation before anything moves.' - **Actions in other systems:** Slack connector. - **Log entry:** No audit statement on the MCP page. - **Off switch:** Admin chooses data categories; users told to disable automatic tool execution. - **Sensitive fields:** Data categories chosen per connection. ## Status by step | Step | Found | Source | | --- | --- | --- | | Employee questions from policy | partial | [gusto.com/pricing](https://gusto.com/pricing) | | Drafts changes | partial | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | Approval before live | yes | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | Acts in other systems | partial | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | Agent audit log | none found | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | Off switch | partial | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | Sensitive-field controls | partial | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Gusto: gusto.com/pricing: https://gusto.com/pricing (read 8 Nov 2026) - Gusto: gusto.com/product/integrations/gusto-mcp: https://gusto.com/product/integrations/gusto-mcp (read 8 Nov 2026) - Gusto: gusto.com/security: https://gusto.com/security (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/employee-questions/gusto/ --- # Lattice on employee questions from policy Agent Diary, Lattice, read 8 Nov 2026 **The short answer.** Lattice runs employee questions from policy on the pages read on 8 Nov 2026: HR Help Desk answers benefits, policy and PTO questions. Approval: none found. Log: none found. Off switch: partial. The steps below quote the vendor's page at each point, and a step that says "none found" is about the pages, not the product. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** HR Help Desk answers benefits, policy and PTO questions. - **Draft:** Drafts evidence-backed reviews; MCP drafts, edits and submits reviews. - **Approval:** Not stated for the agent. - **Actions in other systems:** Slack, ChatGPT and Codex resources; Zoom, Meet, Teams in 1:1s. - **Log entry:** Not stated. - **Off switch:** Admins activate the AI Agent in a setup experience; role access left to the AI system's admin. - **Sensitive fields:** Answers scoped to each user's Lattice access. ## Status by step | Step | Found | Source | | --- | --- | --- | | Employee questions from policy | yes | [lattice.com/platform/ai-agent/hr-help-desk](https://lattice.com/platform/ai-agent/hr-help-desk) | | Drafts changes | yes | [lattice.com/platform/mcp](https://lattice.com/platform/mcp) | | Approval before live | none found | [lattice.com/platform/ai-agent](https://lattice.com/platform/ai-agent) | | Acts in other systems | partial | [lattice.com/platform/mcp](https://lattice.com/platform/mcp) | | Agent audit log | none found | [trustcenter.lattice.com/](https://trustcenter.lattice.com/) | | Off switch | partial | [lattice.com/blog/july-2026-product-updates](https://lattice.com/blog/july-2026-product-updates) | | Sensitive-field controls | partial | [lattice.com/platform/mcp](https://lattice.com/platform/mcp) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Lattice: lattice.com/platform/ai-agent/hr-help-desk: https://lattice.com/platform/ai-agent/hr-help-desk (read 8 Nov 2026) - Lattice: lattice.com/platform/mcp: https://lattice.com/platform/mcp (read 8 Nov 2026) - Lattice: lattice.com/platform/ai-agent: https://lattice.com/platform/ai-agent (read 8 Nov 2026) - Lattice: lattice.com/platform/compensation/cycle-management: https://lattice.com/platform/compensation/cycle-management (read 8 Nov 2026) - Lattice: trustcenter.lattice.com/: https://trustcenter.lattice.com/ (read 8 Nov 2026) - Lattice: lattice.com/blog/july-2026-product-updates: https://lattice.com/blog/july-2026-product-updates (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/employee-questions/lattice/ --- # Factorial on employee questions from policy Agent Diary, Factorial, read 8 Nov 2026 **The short answer.** Factorial runs employee questions from policy on the pages read on 8 Nov 2026: 'Ask One what the rule is and it quotes your handbook.'. Approval: yes. Log: none found. Off switch: none found. The steps below quote the vendor's page at each point, and a step that says "none found" is about the pages, not the product. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** 'Ask One what the rule is and it quotes your handbook.' - **Draft:** Drafts job offers, policies, onboarding plans, dashboards. - **Approval:** 'It asks before it acts.' - **Actions in other systems:** Meets you in Slack, HubSpot and email; acts, calls, confirms and notifies. - **Log entry:** Audit logging is a trust-center heading; no agent log named. - **Off switch:** Not stated. - **Sensitive fields:** 'ONE enforces each person's permissions on every request. No permission, no action.' ## Status by step | Step | Found | Source | | --- | --- | --- | | Employee questions from policy | yes | [factorial.com/ai](https://factorial.com/ai) | | Drafts changes | yes | [factorial.com/ai](https://factorial.com/ai) | | Approval before live | yes | [factorial.com/ai](https://factorial.com/ai) | | Acts in other systems | yes | [factorial.com/ai](https://factorial.com/ai) | | Agent audit log | none found | [trust.factorial.co/](https://trust.factorial.co/) | | Off switch | none found | [factorial.com/ai](https://factorial.com/ai) | | Sensitive-field controls | yes | [factorial.com/ai](https://factorial.com/ai) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Factorial: factorial.com/ai: https://factorial.com/ai (read 8 Nov 2026) - Factorial: trust.factorial.co/: https://trust.factorial.co/ (read 8 Nov 2026) - Factorial: apidoc.factorialhr.com/docs/mcp-server: https://apidoc.factorialhr.com/docs/mcp-server (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/employee-questions/factorial/ --- # Humaans on employee questions from policy Agent Diary, Humaans, read 8 Nov 2026 **The short answer.** Humaans runs employee questions from policy on the pages read on 8 Nov 2026: Knowledge base answers, time away policy answers, employee request resolution. Approval: partial. Log: yes. Off switch: none found. The steps below quote the vendor's page at each point, and a step that says "none found" is about the pages, not the product. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** Knowledge base answers, time away policy answers, employee request resolution. - **Draft:** Stage comp increases and generate letters ready for approval. - **Approval:** Approvals and notifications; letters 'ready for approval'; no blanket 'nothing goes live' statement found. - **Actions in other systems:** Slack-native Athena; MCP sends e-sign documents and triggers workflows. - **Log entry:** 'Audit log of all actions' on Athena; Audit Log feature. - **Off switch:** Not stated. - **Sensitive fields:** Role-based permissions and explicit data scopes when connecting. ## Status by step | Step | Found | Source | | --- | --- | --- | | Employee questions from policy | yes | [humaans.io/product-releases-2025](https://humaans.io/product-releases-2025) | | Drafts changes | yes | [humaans.io/mcp](https://humaans.io/mcp) | | Approval before live | partial | [humaans.io/pricing](https://humaans.io/pricing) | | Acts in other systems | yes | [humaans.io/mcp](https://humaans.io/mcp) | | Agent audit log | yes | [humaans.io/pricing](https://humaans.io/pricing) | | Off switch | none found | [humaans.io/ai](https://humaans.io/ai) | | Sensitive-field controls | yes | [humaans.io/mcp](https://humaans.io/mcp) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Humaans: humaans.io/product-releases-2025: https://humaans.io/product-releases-2025 (read 8 Nov 2026) - Humaans: humaans.io/mcp: https://humaans.io/mcp (read 8 Nov 2026) - Humaans: humaans.io/careers: https://humaans.io/careers (read 8 Nov 2026) - Humaans: humaans.io/pricing: https://humaans.io/pricing (read 8 Nov 2026) - Humaans: humaans.io/ai: https://humaans.io/ai (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/employee-questions/humaans/ --- # ADP on employee questions from policy Agent Diary, ADP, read 8 Nov 2026 **The short answer.** ADP runs employee questions from policy on the pages read on 8 Nov 2026: ADP Assist: 'Action cards and a virtual assistant help employees by addressing most HR issues.'. Approval: none found. Log: none found. Off switch: none found. Every number on this page was read on 8 Nov 2026 from the page linked under Sources. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** ADP Assist: 'Action cards and a virtual assistant help employees by addressing most HR issues.' - **Draft:** Action cards surface tasks; no drafting described. - **Approval:** Not found. - **Actions in other systems:** Not found. - **Log entry:** Not found. - **Off switch:** Not found. - **Sensitive fields:** Role-based access in Workforce Now. ## Status by step | Step | Found | Source | | --- | --- | --- | | Employee questions from policy | yes | [www.adp.com/what-we-offer/products/adp-workforce-n](https://www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx) | | Drafts changes | partial | [www.adp.com/what-we-offer/products/adp-workforce-n](https://www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx) | | Approval before live | none found | [www.adp.com/what-we-offer/products/adp-workforce-n](https://www.adp.com/what-we-offer/products/adp-workforce-now/capabilities) | | Acts in other systems | none found | [www.adp.com/what-we-offer/products/adp-workforce-n](https://www.adp.com/what-we-offer/products/adp-workforce-now/capabilities) | | Agent audit log | none found | [www.adp.com/what-we-offer/products/adp-workforce-n](https://www.adp.com/what-we-offer/products/adp-workforce-now/capabilities) | | Off switch | none found | [www.adp.com/what-we-offer/products/adp-workforce-n](https://www.adp.com/what-we-offer/products/adp-workforce-now/capabilities) | | Sensitive-field controls | partial | [www.adp.com/what-we-offer/products/adp-workforce-n](https://www.adp.com/what-we-offer/products/adp-workforce-now/capabilities) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - ADP: www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx: https://www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx (read 8 Nov 2026) - ADP: www.adp.com/what-we-offer/products/adp-workforce-now/hr-management: https://www.adp.com/what-we-offer/products/adp-workforce-now/hr-management (read 8 Nov 2026) - ADP: www.adp.com/what-we-offer/products/adp-workforce-now/talent: https://www.adp.com/what-we-offer/products/adp-workforce-now/talent (read 8 Nov 2026) - ADP: www.adp.com/what-we-offer/products/adp-workforce-now/capabilities: https://www.adp.com/what-we-offer/products/adp-workforce-now/capabilities (read 8 Nov 2026) - ADP: www.adp.com/knowledge/payroll/small-business/solutions/security: https://www.adp.com/knowledge/payroll/small-business/solutions/security (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/employee-questions/adp/ --- # Paylocity on employee questions from policy Agent Diary, Paylocity, read 8 Nov 2026 **The short answer.** Paylocity runs part of employee questions from policy on the pages read on 8 Nov 2026: 'Proactive assistance right into your flow of work'; on-demand HR resource library. Approval: none found. Log: none found. Off switch: none found. Every number on this page was read on 8 Nov 2026 from the page linked under Sources. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** 'Proactive assistance right into your flow of work'; on-demand HR resource library. - **Draft:** AI-powered touchless reports and contract insights. - **Approval:** Not found. - **Actions in other systems:** Not found. - **Log entry:** Not found. - **Off switch:** Not found. - **Sensitive fields:** Role-based security in the HCM. ## Status by step | Step | Found | Source | | --- | --- | --- | | Employee questions from policy | partial | [www.paylocity.com/pricing/](https://www.paylocity.com/pricing/) | | Drafts changes | partial | [www.paylocity.com/pricing/](https://www.paylocity.com/pricing/) | | Approval before live | none found | [www.paylocity.com/products/hr/](https://www.paylocity.com/products/hr/) | | Acts in other systems | none found | [www.paylocity.com/products/hr/](https://www.paylocity.com/products/hr/) | | Agent audit log | none found | [trust.paylocity.com/](https://trust.paylocity.com/) | | Off switch | none found | [www.paylocity.com/products/hr/](https://www.paylocity.com/products/hr/) | | Sensitive-field controls | partial | [www.paylocity.com/products/hr/](https://www.paylocity.com/products/hr/) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Paylocity: www.paylocity.com/pricing/: https://www.paylocity.com/pricing/ (read 8 Nov 2026) - Paylocity: www.paylocity.com/products/hr/talent-management-system/onboarding/: https://www.paylocity.com/products/hr/talent-management-system/onboarding/ (read 8 Nov 2026) - Paylocity: www.paylocity.com/products/hr/talent-management-system/compensation-management/: https://www.paylocity.com/products/hr/talent-management-system/compensation-management/ (read 8 Nov 2026) - Paylocity: www.paylocity.com/products/hr/: https://www.paylocity.com/products/hr/ (read 8 Nov 2026) - Paylocity: trust.paylocity.com/: https://trust.paylocity.com/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/employee-questions/paylocity/ --- # Taito.ai on employee questions from policy Agent Diary, Taito.ai, read 8 Nov 2026 **The short answer.** Taito.ai runs employee questions from policy on the pages read on 8 Nov 2026: Slack agent answers policy, time off and contract questions from the handbook. Approval: partial. Log: partial. Off switch: none found. The steps below quote the vendor's page at each point, and a step that says "none found" is about the pages, not the product. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** Slack agent answers policy, time off and contract questions from the handbook. - **Draft:** Agents draft reviews and flag biases. - **Approval:** Approvals sketched for time off only. - **Actions in other systems:** Slack-native requests and approvals; pre-payroll export. - **Log entry:** Activity log of changes to employee records. - **Off switch:** Not stated. - **Sensitive fields:** Field-level access control. ## Status by step | Step | Found | Source | | --- | --- | --- | | Employee questions from policy | yes | [taito.ai/](https://taito.ai/) | | Drafts changes | yes | [taito.ai/](https://taito.ai/) | | Approval before live | partial | [taito.ai/](https://taito.ai/) | | Acts in other systems | yes | [taito.ai/](https://taito.ai/) | | Agent audit log | partial | [taito.ai/](https://taito.ai/) | | Off switch | none found | [taito.ai/](https://taito.ai/) | | Sensitive-field controls | yes | [taito.ai/](https://taito.ai/) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Taito.ai: taito.ai/: https://taito.ai/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/employee-questions/taito/ --- # Leena AI on employee questions from policy Agent Diary, Leena AI, read 8 Nov 2026 **The short answer.** Leena AI runs employee questions from policy on the pages read on 8 Nov 2026: 'always-on HR helpdesk'; onboarding, leave, payroll questions, letters, policies. Approval: none found. Log: yes. Off switch: none found. The steps below quote the vendor's page at each point, and a step that says "none found" is about the pages, not the product. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** 'always-on HR helpdesk'; onboarding, leave, payroll questions, letters, policies. - **Draft:** Letters generated; tickets resolved. - **Approval:** Not stated. - **Actions in other systems:** Workday, Oracle, SAP, ServiceNow, Slack, Microsoft. - **Log entry:** 'Role-based access, policy guardrails, full audit trail.' - **Off switch:** Not stated. - **Sensitive fields:** Role-based access and policy guardrails. ## Status by step | Step | Found | Source | | --- | --- | --- | | Employee questions from policy | yes | [leena.ai/](https://leena.ai/) | | Drafts changes | partial | [leena.ai/](https://leena.ai/) | | Approval before live | none found | [leena.ai/](https://leena.ai/) | | Acts in other systems | yes | [leena.ai/](https://leena.ai/) | | Agent audit log | yes | [leena.ai/](https://leena.ai/) | | Off switch | none found | [leena.ai/](https://leena.ai/) | | Sensitive-field controls | partial | [leena.ai/](https://leena.ai/) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Leena AI: leena.ai/: https://leena.ai/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/employee-questions/leena/ --- # Moveworks on employee questions from policy Agent Diary, Moveworks, read 8 Nov 2026 **The short answer.** Moveworks runs employee questions from policy on the pages read on 8 Nov 2026: Answers with source citations from systems, policies and knowledge bases. Approval: none found. Log: none found. Off switch: none found. The steps below quote the vendor's page at each point, and a step that says "none found" is about the pages, not the product. Trigger / Draft / Approval / Actions / Log / Off switch - **Trigger:** Answers with source citations from systems, policies and knowledge bases. - **Draft:** Plugins such as workday-update-address and adp-update-legal-name. - **Approval:** Not stated. - **Actions in other systems:** Workday, ADP and ServiceNow plugins; end-to-end task automation. - **Log entry:** Not stated. - **Off switch:** Not stated. - **Sensitive fields:** PII masking, content filtering, zero-day data retention. ## Status by step | Step | Found | Source | | --- | --- | --- | | Employee questions from policy | yes | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | | Drafts changes | partial | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | | Approval before live | none found | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | | Acts in other systems | yes | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | | Agent audit log | none found | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | | Off switch | none found | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | | Sensitive-field controls | yes | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Moveworks: www.moveworks.com/us/en/solutions/ai-for-hr: https://www.moveworks.com/us/en/solutions/ai-for-hr (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/agents/employee-questions/moveworks/ --- # Twelve capabilities, who has each, with the evidence By capability, read 8 Nov 2026 **The short answer.** Twelve capability pages, one per matrix column, each listing who documents it, who documents a narrower form and for whom nothing was found on 8 Nov 2026, with the vendor's sentence and the page for every row. Answering questions is nearly universal; running a comp cycle is documented by two platforms and an off switch by three. - [Answers employee questions from policy](/capabilities/answers_policy/): yes: 14; partial: 2 - [Drafts changes](/capabilities/drafts_changes/): yes: 10; partial: 6 - [Runs onboarding end to end](/capabilities/runs_onboarding/): yes: 2; partial: 7 - [Runs a compensation cycle](/capabilities/runs_comp_cycle/): yes: 2; partial: 1 - [Runs offboarding](/capabilities/runs_offboarding/): yes: 0; partial: 5 - [Acts in other systems (Slack, Teams, email, ATS, payroll)](/capabilities/acts_other_systems/): yes: 9; partial: 4 - [Approval before anything goes live](/capabilities/approval_before_live/): yes: 6; partial: 5 - [Audit log of agent actions](/capabilities/agent_audit_log/): yes: 5; partial: 3 - [Off switch for AI features](/capabilities/off_switch/): yes: 3; partial: 4 - [Sensitive-field controls](/capabilities/sensitive_controls/): yes: 8; partial: 8 - [Official MCP server or connector, with access level](/capabilities/mcp_official/): yes: 10; partial: 2 - [AI governance certification held or in progress (ISO 42001)](/capabilities/ai_governance_cert/): yes: 2; partial: 1 **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/capabilities/ --- # Answers employee questions from policy Capability, read 8 Nov 2026 **The short answer.** 14 of 17 platforms document "answers employee questions from policy" on the pages read on 8 Nov 2026: Shapes, HiBob, Rippling, BambooHR, Personio, Deel, Workday, Lattice, Factorial, Humaans, ADP, Taito.ai, Leena AI, Moveworks. 2 document a narrower form: Gusto, Paylocity. | Platform | Found | Evidence | Source | | --- | --- | --- | --- | | [Shapes](/platforms/shapes/) | yes | AI chat open to everyone reads the Knowledge Base; agents 'know your org, your policies and your people'. | [shapes.co/releases/ai-chat-now-reads-the-help-ce](https://shapes.co/releases/ai-chat-now-reads-the-help-center?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=capability) | | [HiBob](/platforms/hibob/) | yes | Bob AI gives employees and managers instant self-service answers. | [www.hibob.com/platform/employee-experience-softw](https://www.hibob.com/platform/employee-experience-software/) | | [Rippling](/platforms/rippling/) | yes | Employees get personalized pay, benefits and policy explanations. | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | | [BambooHR](/platforms/bamboohr/) | yes | Chat answers questions using documents administrators select. | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | [Personio](/platforms/personio/) | yes | Personio Assistant answers people questions. | [www.personio.com/product/assistant/](https://www.personio.com/product/assistant/) | | [Deel](/platforms/deel/) | yes | Ask workforce data a question in plain language; asking Deel questions live in Slack. | [www.deel.com/hr/](https://www.deel.com/hr/) | | [Workday](/platforms/workday/) | yes | HR service and self-service agents. | [www.workday.com/en-us/artificial-intelligence/ai](https://www.workday.com/en-us/artificial-intelligence/ai-agents/hr-service.html) | | [Lattice](/platforms/lattice/) | yes | HR Help Desk answers benefits, policy and PTO questions. | [lattice.com/platform/ai-agent/hr-help-desk](https://lattice.com/platform/ai-agent/hr-help-desk) | | [Factorial](/platforms/factorial/) | yes | 'Ask One what the rule is and it quotes your handbook.' | [factorial.com/ai](https://factorial.com/ai) | | [Humaans](/platforms/humaans/) | yes | Knowledge base answers, time away policy answers, employee request resolution. | [humaans.io/product-releases-2025](https://humaans.io/product-releases-2025) | | [ADP](/platforms/adp/) | yes | ADP Assist: 'Action cards and a virtual assistant help employees by addressing most HR issues.' | [www.adp.com/what-we-offer/products/adp-workforce](https://www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx) | | [Taito.ai](/platforms/taito/) | yes | Slack agent answers policy, time off and contract questions from the handbook. | [taito.ai/](https://taito.ai/) | | [Leena AI](/platforms/leena/) | yes | 'always-on HR helpdesk'; onboarding, leave, payroll questions, letters, policies. | [leena.ai/](https://leena.ai/) | | [Moveworks](/platforms/moveworks/) | yes | Answers with source citations from systems, policies and knowledge bases. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | | [Gusto](/platforms/gusto/) | partial | Ask an HR Expert on Premium, human-reviewed; AI-driven suggestions. | [gusto.com/pricing](https://gusto.com/pricing) | | [Paylocity](/platforms/paylocity/) | partial | 'Proactive assistance right into your flow of work'; on-demand HR resource library. | [www.paylocity.com/pricing/](https://www.paylocity.com/pricing/) | | [Sage](/platforms/sage/) | none found | Not found on the Sage HR pages read on 2026-11-08. | [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) | Ordered yes, partial, none found. Read 8 Nov 2026. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Shapes: https://shapes.co/releases/ai-chat-now-reads-the-help-center (read 8 Nov 2026) - HiBob: https://www.hibob.com/platform/employee-experience-software/ (read 8 Nov 2026) - Rippling: https://www.rippling.com/platform/ai (read 8 Nov 2026) - BambooHR: https://www.bamboohr.com/product-updates/bamboo-ai (read 8 Nov 2026) - Personio: https://www.personio.com/product/assistant/ (read 8 Nov 2026) - Deel: https://www.deel.com/hr/ (read 8 Nov 2026) - Workday: https://www.workday.com/en-us/artificial-intelligence/ai-agents/hr-service.html (read 8 Nov 2026) - Lattice: https://lattice.com/platform/ai-agent/hr-help-desk (read 8 Nov 2026) - Factorial: https://factorial.com/ai (read 8 Nov 2026) - Humaans: https://humaans.io/product-releases-2025 (read 8 Nov 2026) - ADP: https://www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx (read 8 Nov 2026) - Taito.ai: https://taito.ai/ (read 8 Nov 2026) - Leena AI: https://leena.ai/ (read 8 Nov 2026) - Moveworks: https://www.moveworks.com/us/en/solutions/ai-for-hr (read 8 Nov 2026) - Gusto: https://gusto.com/pricing (read 8 Nov 2026) - Paylocity: https://www.paylocity.com/pricing/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/capabilities/answers_policy/ --- # Drafts changes Capability, read 8 Nov 2026 **The short answer.** 10 of 17 platforms document "drafts changes" on the pages read on 8 Nov 2026: Shapes, HiBob, Rippling, BambooHR, Deel, Workday, Lattice, Factorial, Humaans, Taito.ai. 6 document a narrower form: Personio, Gusto, ADP, Paylocity, Leena AI, Moveworks. For 1 nothing was found on the pages listed. | Platform | Found | Evidence | Source | | --- | --- | --- | --- | | [Shapes](/platforms/shapes/) | yes | 'Ask for a change, they draft it, you apply it.' | [shapes.co/agents](https://shapes.co/agents?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=capability) | | [HiBob](/platforms/hibob/) | yes | Agents take actions such as updating goals, creating development plans. | [www.hibob.com/platform/core/ai/agents/](https://www.hibob.com/platform/core/ai/agents/) | | [Rippling](/platforms/rippling/) | yes | Plan reorgs, draft hires; AI maps bonus data and flags issues. | [www.rippling.com/platform/ai/rippling-mcp](https://www.rippling.com/platform/ai/rippling-mcp) | | [BambooHR](/platforms/bamboohr/) | yes | Drafts schedules, organises job titles, prepares work for the appropriate person. | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | [Deel](/platforms/deel/) | yes | Draft a job description, policy or training course in minutes. | [www.deel.com/hr/](https://www.deel.com/hr/) | | [Workday](/platforms/workday/) | yes | Agents draft and execute within guardrails. | [www.workday.com/en-us/artificial-intelligence/ai](https://www.workday.com/en-us/artificial-intelligence/ai-agents.html) | | [Lattice](/platforms/lattice/) | yes | Drafts evidence-backed reviews; MCP drafts, edits and submits reviews. | [lattice.com/platform/mcp](https://lattice.com/platform/mcp) | | [Factorial](/platforms/factorial/) | yes | Drafts job offers, policies, onboarding plans, dashboards. | [factorial.com/ai](https://factorial.com/ai) | | [Humaans](/platforms/humaans/) | yes | Stage comp increases and generate letters ready for approval. | [humaans.io/mcp](https://humaans.io/mcp) | | [Taito.ai](/platforms/taito/) | yes | Agents draft reviews and flag biases. | [taito.ai/](https://taito.ai/) | | [Personio](/platforms/personio/) | partial | Analytics Agent builds reports from a description; screening agent shortlists. | [www.personio.com/whats-new/](https://www.personio.com/whats-new/) | | [Gusto](/platforms/gusto/) | partial | MCP can log time, onboard employees, prepare payroll. | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | [ADP](/platforms/adp/) | partial | Action cards surface tasks; no drafting described. | [www.adp.com/what-we-offer/products/adp-workforce](https://www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx) | | [Paylocity](/platforms/paylocity/) | partial | AI-powered touchless reports and contract insights. | [www.paylocity.com/pricing/](https://www.paylocity.com/pricing/) | | [Leena AI](/platforms/leena/) | partial | Letters generated; tickets resolved. | [leena.ai/](https://leena.ai/) | | [Moveworks](/platforms/moveworks/) | partial | Plugins such as workday-update-address and adp-update-legal-name. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | | [Sage](/platforms/sage/) | none found | Not found. | [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) | Ordered yes, partial, none found. Read 8 Nov 2026. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Shapes: https://shapes.co/agents (read 8 Nov 2026) - HiBob: https://www.hibob.com/platform/core/ai/agents/ (read 8 Nov 2026) - Rippling: https://www.rippling.com/platform/ai/rippling-mcp (read 8 Nov 2026) - BambooHR: https://www.bamboohr.com/product-updates/bamboo-ai (read 8 Nov 2026) - Deel: https://www.deel.com/hr/ (read 8 Nov 2026) - Workday: https://www.workday.com/en-us/artificial-intelligence/ai-agents.html (read 8 Nov 2026) - Lattice: https://lattice.com/platform/mcp (read 8 Nov 2026) - Factorial: https://factorial.com/ai (read 8 Nov 2026) - Humaans: https://humaans.io/mcp (read 8 Nov 2026) - Taito.ai: https://taito.ai/ (read 8 Nov 2026) - Personio: https://www.personio.com/whats-new/ (read 8 Nov 2026) - Gusto: https://gusto.com/product/integrations/gusto-mcp (read 8 Nov 2026) - ADP: https://www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx (read 8 Nov 2026) - Paylocity: https://www.paylocity.com/pricing/ (read 8 Nov 2026) - Leena AI: https://leena.ai/ (read 8 Nov 2026) - Moveworks: https://www.moveworks.com/us/en/solutions/ai-for-hr (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/capabilities/drafts_changes/ --- # Runs onboarding end to end Capability, read 8 Nov 2026 **The short answer.** 2 of 17 platforms document "runs onboarding end to end" on the pages read on 8 Nov 2026: Shapes, Humaans. 7 document a narrower form: Rippling, Workday, Gusto, Factorial, Paylocity, Taito.ai, Leena AI. For 8 nothing was found on the pages listed. | Platform | Found | Evidence | Source | | --- | --- | --- | --- | | [Shapes](/platforms/shapes/) | yes | Onboarding named as a process agents handle; the Workflows agent drafts the flow and runs on new hires. | [shapes.co/](https://shapes.co/?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=capability) | | [Humaans](/platforms/humaans/) | yes | 'Onboarding that runs itself'; end-to-end process automation. | [humaans.io/careers](https://humaans.io/careers) | | [Rippling](/platforms/rippling/) | partial | 'Onboard new hires' listed; automations run repeatable work from triggers. | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | | [Workday](/platforms/workday/) | partial | Onboarding workflows in HCM; agents for HR service and admin. | [www.workday.com/en-us/artificial-intelligence/ai](https://www.workday.com/en-us/artificial-intelligence/ai-agents/admin.html) | | [Gusto](/platforms/gusto/) | partial | MCP onboards employees or contractors with confirmation. | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | [Factorial](/platforms/factorial/) | partial | Drafts onboarding plans; tickets routed and resolved on approval. | [factorial.com/ai](https://factorial.com/ai) | | [Paylocity](/platforms/paylocity/) | partial | Workflows self-serve automation engine; onboarding module. | [www.paylocity.com/products/hr/talent-management-](https://www.paylocity.com/products/hr/talent-management-system/onboarding/) | | [Taito.ai](/platforms/taito/) | partial | Onboarding runs end to end as a no-code workflow (not an agent). | [taito.ai/](https://taito.ai/) | | [Leena AI](/platforms/leena/) | partial | 'Onboarding (Day-1 onwards)' as a capability, on top of the HRIS. | [leena.ai/](https://leena.ai/) | | [HiBob](/platforms/hibob/) | none found | No agent described as running onboarding end to end on 2026-11-08. | [www.hibob.com/platform/core/ai/agents/](https://www.hibob.com/platform/core/ai/agents/) | | [BambooHR](/platforms/bamboohr/) | none found | Not described on the Bamboo AI update. | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | [Personio](/platforms/personio/) | none found | Onboarding workflows exist; no agent described. | [www.personio.com/product/onboarding/](https://www.personio.com/product/onboarding/) | | [Deel](/platforms/deel/) | none found | Onboarding workflows exist; no agent described. | [www.deel.com/hr/](https://www.deel.com/hr/) | | [Lattice](/platforms/lattice/) | none found | Not described. | [lattice.com/platform/ai-agent](https://lattice.com/platform/ai-agent) | | [Sage](/platforms/sage/) | none found | Onboarding checklists exist; no agent. | [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) | | [ADP](/platforms/adp/) | none found | Onboarding module; no agent. | [www.adp.com/what-we-offer/products/adp-workforce](https://www.adp.com/what-we-offer/products/adp-workforce-now/hr-management) | | [Moveworks](/platforms/moveworks/) | none found | Not described. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | Ordered yes, partial, none found. Read 8 Nov 2026. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Shapes: https://shapes.co/ (read 8 Nov 2026) - Humaans: https://humaans.io/careers (read 8 Nov 2026) - Rippling: https://www.rippling.com/platform/ai (read 8 Nov 2026) - Workday: https://www.workday.com/en-us/artificial-intelligence/ai-agents/admin.html (read 8 Nov 2026) - Gusto: https://gusto.com/product/integrations/gusto-mcp (read 8 Nov 2026) - Factorial: https://factorial.com/ai (read 8 Nov 2026) - Paylocity: https://www.paylocity.com/products/hr/talent-management-system/onboarding/ (read 8 Nov 2026) - Taito.ai: https://taito.ai/ (read 8 Nov 2026) - Leena AI: https://leena.ai/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/capabilities/runs_onboarding/ --- # Runs a compensation cycle Capability, read 8 Nov 2026 **The short answer.** 2 of 17 platforms document "runs a compensation cycle" on the pages read on 8 Nov 2026: Shapes, Humaans. 1 document a narrower form: HiBob. For 14 nothing was found on the pages listed. Each row below quotes the page. | Platform | Found | Evidence | Source | | --- | --- | --- | --- | | [Shapes](/platforms/shapes/) | yes | 'Onboarding, comp cycle, employee questions' named on the home page; Compensation app with cycles. | [shapes.co/](https://shapes.co/?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=capability) | | [Humaans](/platforms/humaans/) | yes | Run the April comp cycle example: review in scope, stage increases, generate letters for approval. | [humaans.io/mcp](https://humaans.io/mcp) | | [HiBob](/platforms/hibob/) | partial | Compensation agent shows employees below band minimum; the cycle itself runs in the Compensation module with configurable workflows. | [www.hibob.com/platform/core/ai/agents/](https://www.hibob.com/platform/core/ai/agents/) | | [Rippling](/platforms/rippling/) | none found | Compensation cycles are not mentioned on the AI pages read. | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | | [BambooHR](/platforms/bamboohr/) | none found | Not described. | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | [Personio](/platforms/personio/) | none found | Not described. | [www.personio.com/whats-new/](https://www.personio.com/whats-new/) | | [Deel](/platforms/deel/) | none found | Not described. | [www.deel.com/hr/compensation-management/](https://www.deel.com/hr/compensation-management/) | | [Workday](/platforms/workday/) | none found | Not described as an agent on the pages read. | [www.workday.com/en-us/artificial-intelligence/ai](https://www.workday.com/en-us/artificial-intelligence/ai-agents.html) | | [Gusto](/platforms/gusto/) | none found | Not described. | [gusto.com/pricing](https://gusto.com/pricing) | | [Lattice](/platforms/lattice/) | none found | Compensation cycle management exists; no agent described. | [lattice.com/platform/compensation/cycle-manageme](https://lattice.com/platform/compensation/cycle-management) | | [Factorial](/platforms/factorial/) | none found | Not described. | [factorial.com/ai](https://factorial.com/ai) | | [Sage](/platforms/sage/) | none found | Not found. | [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) | | [ADP](/platforms/adp/) | none found | Not found. | [www.adp.com/what-we-offer/products/adp-workforce](https://www.adp.com/what-we-offer/products/adp-workforce-now/talent) | | [Paylocity](/platforms/paylocity/) | none found | Compensation module; no agent. | [www.paylocity.com/products/hr/talent-management-](https://www.paylocity.com/products/hr/talent-management-system/compensation-management/) | | [Taito.ai](/platforms/taito/) | none found | Not described. | [taito.ai/](https://taito.ai/) | | [Leena AI](/platforms/leena/) | none found | Not described. | [leena.ai/](https://leena.ai/) | | [Moveworks](/platforms/moveworks/) | none found | Not described. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | Ordered yes, partial, none found. Read 8 Nov 2026. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Shapes: https://shapes.co/ (read 8 Nov 2026) - Humaans: https://humaans.io/mcp (read 8 Nov 2026) - HiBob: https://www.hibob.com/platform/core/ai/agents/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/capabilities/runs_comp_cycle/ --- # Runs offboarding Capability, read 8 Nov 2026 **The short answer.** 0 of 17 platforms document "runs offboarding" on the pages read on 8 Nov 2026: none. 5 document a narrower form: Shapes, HiBob, Rippling, Humaans, Taito.ai. For 12 nothing was found on the pages listed. Each row below quotes the page. | Platform | Found | Evidence | Source | | --- | --- | --- | --- | | [Shapes](/platforms/shapes/) | partial | Workflows run for leavers (release note); offboarding is not named as an agent process. | [shapes.co/releases/run-workflows-for-new-starter](https://shapes.co/releases/run-workflows-for-new-starters-and-leavers?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=capability) | | [HiBob](/platforms/hibob/) | partial | Offboarding tasks start when a departure is confirmed (automation, not an agent). | [www.hibob.com/platform/core/offboarding/](https://www.hibob.com/platform/core/offboarding/) | | [Rippling](/platforms/rippling/) | partial | Offboarding disables devices, apps and access through IAM and MDM (automation). | [www.rippling.com/trust/security](https://www.rippling.com/trust/security) | | [Humaans](/platforms/humaans/) | partial | Onboarding and offboarding automation on Growth; Athena end-to-end processes. | [humaans.io/pricing](https://humaans.io/pricing) | | [Taito.ai](/platforms/taito/) | partial | Offboarding as a no-code workflow. | [taito.ai/](https://taito.ai/) | | [BambooHR](/platforms/bamboohr/) | none found | Not described. | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | [Personio](/platforms/personio/) | none found | Not described. | [www.personio.com/whats-new/](https://www.personio.com/whats-new/) | | [Deel](/platforms/deel/) | none found | Not described. | [www.deel.com/hr/](https://www.deel.com/hr/) | | [Workday](/platforms/workday/) | none found | Not described as an agent. | [www.workday.com/en-us/artificial-intelligence/ai](https://www.workday.com/en-us/artificial-intelligence/ai-agents.html) | | [Gusto](/platforms/gusto/) | none found | Not described. | [gusto.com/pricing](https://gusto.com/pricing) | | [Lattice](/platforms/lattice/) | none found | Not described. | [lattice.com/platform/ai-agent](https://lattice.com/platform/ai-agent) | | [Factorial](/platforms/factorial/) | none found | Not described. | [factorial.com/ai](https://factorial.com/ai) | | [Sage](/platforms/sage/) | none found | Not found. | [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) | | [ADP](/platforms/adp/) | none found | Not found. | [www.adp.com/what-we-offer/products/adp-workforce](https://www.adp.com/what-we-offer/products/adp-workforce-now/hr-management) | | [Paylocity](/platforms/paylocity/) | none found | Not found. | [www.paylocity.com/products/hr/](https://www.paylocity.com/products/hr/) | | [Leena AI](/platforms/leena/) | none found | Not described. | [leena.ai/](https://leena.ai/) | | [Moveworks](/platforms/moveworks/) | none found | Not described. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | Ordered yes, partial, none found. Read 8 Nov 2026. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Shapes: https://shapes.co/releases/run-workflows-for-new-starters-and-leavers (read 8 Nov 2026) - HiBob: https://www.hibob.com/platform/core/offboarding/ (read 8 Nov 2026) - Rippling: https://www.rippling.com/trust/security (read 8 Nov 2026) - Humaans: https://humaans.io/pricing (read 8 Nov 2026) - Taito.ai: https://taito.ai/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/capabilities/runs_offboarding/ --- # Acts in other systems (Slack, Teams, email, ATS, payroll) Capability, read 8 Nov 2026 **The short answer.** 9 of 17 platforms document "acts in other systems (slack, teams, email, ats, payroll)" on the pages read on 8 Nov 2026: Shapes, HiBob, Rippling, Workday, Factorial, Humaans, Taito.ai, Leena AI, Moveworks. 4 document a narrower form: BambooHR, Deel, Gusto, Lattice. | Platform | Found | Evidence | Source | | --- | --- | --- | --- | | [Shapes](/platforms/shapes/) | yes | Workflow steps include emails, Slack messages and webhooks; agents open next to the work. | [shapes.co/releases/workflows-2-0-now-live-for-ev](https://shapes.co/releases/workflows-2-0-now-live-for-everyone?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=capability) | | [HiBob](/platforms/hibob/) | yes | MCP runs supported HR actions from Claude, ChatGPT, Cursor and Slack. | [www.hibob.com/platform/core/ai/mcp](https://www.hibob.com/platform/core/ai/mcp) | | [Rippling](/platforms/rippling/) | yes | Acts across HR, IT and finance and connected tools via plug-ins, limited to actions the integration supports. | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | | [Workday](/platforms/workday/) | yes | Agent Gateway supports MCP and A2A across systems. | [blog.workday.com/en-us/managing-ai-powered-futur](https://blog.workday.com/en-us/managing-ai-powered-future-of-work.html) | | [Factorial](/platforms/factorial/) | yes | Meets you in Slack, HubSpot and email; acts, calls, confirms and notifies. | [factorial.com/ai](https://factorial.com/ai) | | [Humaans](/platforms/humaans/) | yes | Slack-native Athena; MCP sends e-sign documents and triggers workflows. | [humaans.io/mcp](https://humaans.io/mcp) | | [Taito.ai](/platforms/taito/) | yes | Slack-native requests and approvals; pre-payroll export. | [taito.ai/](https://taito.ai/) | | [Leena AI](/platforms/leena/) | yes | Workday, Oracle, SAP, ServiceNow, Slack, Microsoft. | [leena.ai/](https://leena.ai/) | | [Moveworks](/platforms/moveworks/) | yes | Workday, ADP and ServiceNow plugins; end-to-end task automation. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | | [BambooHR](/platforms/bamboohr/) | partial | Employee outreach and payroll runs prepared inside BambooHR; Claude connector in beta. | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | [Deel](/platforms/deel/) | partial | Slack plugin answers questions; Teams next. | [www.deel.com/hr/plugin/](https://www.deel.com/hr/plugin/) | | [Gusto](/platforms/gusto/) | partial | Slack connector. | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | [Lattice](/platforms/lattice/) | partial | Slack, ChatGPT and Codex resources; Zoom, Meet, Teams in 1:1s. | [lattice.com/platform/mcp](https://lattice.com/platform/mcp) | | [Personio](/platforms/personio/) | none found | Not described on the pages read. | [www.personio.com/whats-new/](https://www.personio.com/whats-new/) | | [Sage](/platforms/sage/) | none found | Not found. | [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) | | [ADP](/platforms/adp/) | none found | Not found. | [www.adp.com/what-we-offer/products/adp-workforce](https://www.adp.com/what-we-offer/products/adp-workforce-now/capabilities) | | [Paylocity](/platforms/paylocity/) | none found | Not found. | [www.paylocity.com/products/hr/](https://www.paylocity.com/products/hr/) | Ordered yes, partial, none found. Read 8 Nov 2026. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Shapes: https://shapes.co/releases/workflows-2-0-now-live-for-everyone (read 8 Nov 2026) - HiBob: https://www.hibob.com/platform/core/ai/mcp (read 8 Nov 2026) - Rippling: https://www.rippling.com/platform/ai (read 8 Nov 2026) - Workday: https://blog.workday.com/en-us/managing-ai-powered-future-of-work.html (read 8 Nov 2026) - Factorial: https://factorial.com/ai (read 8 Nov 2026) - Humaans: https://humaans.io/mcp (read 8 Nov 2026) - Taito.ai: https://taito.ai/ (read 8 Nov 2026) - Leena AI: https://leena.ai/ (read 8 Nov 2026) - Moveworks: https://www.moveworks.com/us/en/solutions/ai-for-hr (read 8 Nov 2026) - BambooHR: https://www.bamboohr.com/product-updates/bamboo-ai (read 8 Nov 2026) - Deel: https://www.deel.com/hr/plugin/ (read 8 Nov 2026) - Gusto: https://gusto.com/product/integrations/gusto-mcp (read 8 Nov 2026) - Lattice: https://lattice.com/platform/mcp (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/capabilities/acts_other_systems/ --- # Approval before anything goes live Capability, read 8 Nov 2026 **The short answer.** 6 of 17 platforms document "approval before anything goes live" on the pages read on 8 Nov 2026: Shapes, Rippling, BambooHR, Workday, Gusto, Factorial. 5 document a narrower form: HiBob, Personio, Deel, Humaans, Taito.ai. For 6 nothing was found on the pages listed. | Platform | Found | Evidence | Source | | --- | --- | --- | --- | | [Shapes](/platforms/shapes/) | yes | 'Nothing goes live until you say so.' Directory built with an agent writes nothing until Approve. | [shapes.co/agents](https://shapes.co/agents?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=capability) | | [Rippling](/platforms/rippling/) | yes | 'Before it takes action, it gets your approval.' | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | | [BambooHR](/platforms/bamboohr/) | yes | 'still go through the appropriate person before they're applied or sent'. | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | [Workday](/platforms/workday/) | yes | 'bring people in to review or approve sensitive actions before anything changes'; sandboxed what-ifs. | [www.workday.com/en-us/artificial-intelligence/wo](https://www.workday.com/en-us/artificial-intelligence/workday-sana.html) | | [Gusto](/platforms/gusto/) | yes | 'Every write comes back to you for confirmation before anything moves.' | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | [Factorial](/platforms/factorial/) | yes | 'It asks before it acts.' | [factorial.com/ai](https://factorial.com/ai) | | [HiBob](/platforms/hibob/) | partial | Admins can require confirmation before writes in some AI tools; no approval step stated for agents. | [www.hibob.com/platform/core/ai/mcp](https://www.hibob.com/platform/core/ai/mcp) | | [Personio](/platforms/personio/) | partial | 'A person always makes the decision' (screening); every agent gets an approval status in the AI Transformation Hub. | [www.personio.com/whats-new/](https://www.personio.com/whats-new/) | | [Deel](/platforms/deel/) | partial | AI terms: 'Always keep a human in the loop'. | [www.deel.com/legal/ai-terms-of-service/](https://www.deel.com/legal/ai-terms-of-service/) | | [Humaans](/platforms/humaans/) | partial | Approvals and notifications; letters 'ready for approval'; no blanket 'nothing goes live' statement found. | [humaans.io/pricing](https://humaans.io/pricing) | | [Taito.ai](/platforms/taito/) | partial | Approvals sketched for time off only. | [taito.ai/](https://taito.ai/) | | [Lattice](/platforms/lattice/) | none found | Not stated for the agent. | [lattice.com/platform/ai-agent](https://lattice.com/platform/ai-agent) | | [Sage](/platforms/sage/) | none found | Not found. | [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) | | [ADP](/platforms/adp/) | none found | Not found. | [www.adp.com/what-we-offer/products/adp-workforce](https://www.adp.com/what-we-offer/products/adp-workforce-now/capabilities) | | [Paylocity](/platforms/paylocity/) | none found | Not found. | [www.paylocity.com/products/hr/](https://www.paylocity.com/products/hr/) | | [Leena AI](/platforms/leena/) | none found | Not stated. | [leena.ai/](https://leena.ai/) | | [Moveworks](/platforms/moveworks/) | none found | Not stated. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | Ordered yes, partial, none found. Read 8 Nov 2026. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Shapes: https://shapes.co/agents (read 8 Nov 2026) - Rippling: https://www.rippling.com/platform/ai (read 8 Nov 2026) - BambooHR: https://www.bamboohr.com/product-updates/bamboo-ai (read 8 Nov 2026) - Workday: https://www.workday.com/en-us/artificial-intelligence/workday-sana.html (read 8 Nov 2026) - Gusto: https://gusto.com/product/integrations/gusto-mcp (read 8 Nov 2026) - Factorial: https://factorial.com/ai (read 8 Nov 2026) - HiBob: https://www.hibob.com/platform/core/ai/mcp (read 8 Nov 2026) - Personio: https://www.personio.com/whats-new/ (read 8 Nov 2026) - Deel: https://www.deel.com/legal/ai-terms-of-service/ (read 8 Nov 2026) - Humaans: https://humaans.io/pricing (read 8 Nov 2026) - Taito.ai: https://taito.ai/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/capabilities/approval_before_live/ --- # Audit log of agent actions Capability, read 8 Nov 2026 **The short answer.** 5 of 17 platforms document "audit log of agent actions" on the pages read on 8 Nov 2026: Shapes, BambooHR, Workday, Humaans, Leena AI. 3 document a narrower form: HiBob, Rippling, Taito.ai. For 9 nothing was found on the pages listed. | Platform | Found | Evidence | Source | | --- | --- | --- | --- | | [Shapes](/platforms/shapes/) | yes | 'See who did what in the audit log' release note. | [shapes.co/releases/see-who-did-what-in-the-audit](https://shapes.co/releases/see-who-did-what-in-the-audit-log?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=capability) | | [BambooHR](/platforms/bamboohr/) | yes | Audit Trail records whether a person, Bamboo AI or both made a change. | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | [Workday](/platforms/workday/) | yes | Agent System of Record with full auditability; Agent Analytics. | [www.workday.com/en-us/ai-platform/agent-app-buil](https://www.workday.com/en-us/ai-platform/agent-app-builder.html) | | [Humaans](/platforms/humaans/) | yes | 'Audit log of all actions' on Athena; Audit Log feature. | [humaans.io/pricing](https://humaans.io/pricing) | | [Leena AI](/platforms/leena/) | yes | 'Role-based access, policy guardrails, full audit trail.' | [leena.ai/](https://leena.ai/) | | [HiBob](/platforms/hibob/) | partial | 'AI logs of data in MCP'; no agent action log stated. | [www.hibob.com/platform/core/ai/mcp](https://www.hibob.com/platform/core/ai/mcp) | | [Rippling](/platforms/rippling/) | partial | Answers link to real records; 'all app access is logged and audited'; no agent action log named. | [www.rippling.com/trust/security](https://www.rippling.com/trust/security) | | [Taito.ai](/platforms/taito/) | partial | Activity log of changes to employee records. | [taito.ai/](https://taito.ai/) | | [Personio](/platforms/personio/) | none found | Not stated. | [www.personio.com/security/](https://www.personio.com/security/) | | [Deel](/platforms/deel/) | none found | Not stated. | [www.deel.com/security/](https://www.deel.com/security/) | | [Gusto](/platforms/gusto/) | none found | No audit statement on the MCP page. | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | [Lattice](/platforms/lattice/) | none found | Not stated. | [trustcenter.lattice.com/](https://trustcenter.lattice.com/) | | [Factorial](/platforms/factorial/) | none found | Audit logging is a trust-center heading; no agent log named. | [trust.factorial.co/](https://trust.factorial.co/) | | [Sage](/platforms/sage/) | none found | Not found. | [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) | | [ADP](/platforms/adp/) | none found | Not found. | [www.adp.com/what-we-offer/products/adp-workforce](https://www.adp.com/what-we-offer/products/adp-workforce-now/capabilities) | | [Paylocity](/platforms/paylocity/) | none found | Not found. | [trust.paylocity.com/](https://trust.paylocity.com/) | | [Moveworks](/platforms/moveworks/) | none found | Not stated. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | Ordered yes, partial, none found. Read 8 Nov 2026. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Shapes: https://shapes.co/releases/see-who-did-what-in-the-audit-log (read 8 Nov 2026) - BambooHR: https://www.bamboohr.com/product-updates/bamboo-ai (read 8 Nov 2026) - Workday: https://www.workday.com/en-us/ai-platform/agent-app-builder.html (read 8 Nov 2026) - Humaans: https://humaans.io/pricing (read 8 Nov 2026) - Leena AI: https://leena.ai/ (read 8 Nov 2026) - HiBob: https://www.hibob.com/platform/core/ai/mcp (read 8 Nov 2026) - Rippling: https://www.rippling.com/trust/security (read 8 Nov 2026) - Taito.ai: https://taito.ai/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/capabilities/agent_audit_log/ --- # Off switch for AI features Capability, read 8 Nov 2026 **The short answer.** 3 of 17 platforms document "off switch for ai features" on the pages read on 8 Nov 2026: Shapes, Rippling, BambooHR. 4 document a narrower form: HiBob, Workday, Gusto, Lattice. For 10 nothing was found on the pages listed. Each row below quotes the page. | Platform | Found | Evidence | Source | | --- | --- | --- | --- | | [Shapes](/platforms/shapes/) | yes | Separate switches for AI chat, AI-built Vibes and MCP access, on by default. | [shapes.co/releases/turn-ai-features-on-or-off](https://shapes.co/releases/turn-ai-features-on-or-off?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=capability) | | [Rippling](/platforms/rippling/) | yes | 'Admins can enable or disable AI features.' | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | | [BambooHR](/platforms/bamboohr/) | yes | Admins turn Bamboo AI off or on under Account Settings > Ask BambooHR. | [www.bamboohr.com/product-updates/bamboo-ai](https://www.bamboohr.com/product-updates/bamboo-ai) | | [HiBob](/platforms/hibob/) | partial | MCP is turned on in Bob company-wide; no AI off switch stated for agents. | [www.hibob.com/platform/core/ai/mcp](https://www.hibob.com/platform/core/ai/mcp) | | [Workday](/platforms/workday/) | partial | Agents registered and governed in the Agent System of Record; no single off switch named. | [www.workday.com/en-us/ai-platform/agent-system-o](https://www.workday.com/en-us/ai-platform/agent-system-of-record.html) | | [Gusto](/platforms/gusto/) | partial | Admin chooses data categories; users told to disable automatic tool execution. | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | [Lattice](/platforms/lattice/) | partial | Admins activate the AI Agent in a setup experience; role access left to the AI system's admin. | [lattice.com/blog/july-2026-product-updates](https://lattice.com/blog/july-2026-product-updates) | | [Personio](/platforms/personio/) | none found | Not stated. | [www.personio.com/whats-new/](https://www.personio.com/whats-new/) | | [Deel](/platforms/deel/) | none found | Not stated. | [www.deel.com/hr/](https://www.deel.com/hr/) | | [Factorial](/platforms/factorial/) | none found | Not stated. | [factorial.com/ai](https://factorial.com/ai) | | [Humaans](/platforms/humaans/) | none found | Not stated. | [humaans.io/ai](https://humaans.io/ai) | | [Sage](/platforms/sage/) | none found | Not found. | [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) | | [ADP](/platforms/adp/) | none found | Not found. | [www.adp.com/what-we-offer/products/adp-workforce](https://www.adp.com/what-we-offer/products/adp-workforce-now/capabilities) | | [Paylocity](/platforms/paylocity/) | none found | Not found. | [www.paylocity.com/products/hr/](https://www.paylocity.com/products/hr/) | | [Taito.ai](/platforms/taito/) | none found | Not stated. | [taito.ai/](https://taito.ai/) | | [Leena AI](/platforms/leena/) | none found | Not stated. | [leena.ai/](https://leena.ai/) | | [Moveworks](/platforms/moveworks/) | none found | Not stated. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | Ordered yes, partial, none found. Read 8 Nov 2026. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Shapes: https://shapes.co/releases/turn-ai-features-on-or-off (read 8 Nov 2026) - Rippling: https://www.rippling.com/platform/ai (read 8 Nov 2026) - BambooHR: https://www.bamboohr.com/product-updates/bamboo-ai (read 8 Nov 2026) - HiBob: https://www.hibob.com/platform/core/ai/mcp (read 8 Nov 2026) - Workday: https://www.workday.com/en-us/ai-platform/agent-system-of-record.html (read 8 Nov 2026) - Gusto: https://gusto.com/product/integrations/gusto-mcp (read 8 Nov 2026) - Lattice: https://lattice.com/blog/july-2026-product-updates (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/capabilities/off_switch/ --- # Sensitive-field controls Capability, read 8 Nov 2026 **The short answer.** 8 of 17 platforms document "sensitive-field controls" on the pages read on 8 Nov 2026: Shapes, HiBob, Rippling, Workday, Factorial, Humaans, Taito.ai, Moveworks. 8 document a narrower form: BambooHR, Personio, Gusto, Lattice, Sage, ADP, Paylocity, Leena AI. For 1 nothing was found on the pages listed. | Platform | Found | Evidence | Source | | --- | --- | --- | --- | | [Shapes](/platforms/shapes/) | yes | Field-level permissions: edit, view or hidden per permission level (help center), and a release note on restricting fields. | [help.shapes.co/en/articles/9569339-permission-le](https://help.shapes.co/en/articles/9569339-permission-levels?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=capability) | | [HiBob](/platforms/hibob/) | yes | Sensitive data protected by role-based permissions; agents cannot create a new path to restricted information. | [www.hibob.com/platform/core/ai/agents/](https://www.hibob.com/platform/core/ai/agents/) | | [Rippling](/platforms/rippling/) | yes | Data permissions configured at the platform level. | [www.rippling.com/platform/ai](https://www.rippling.com/platform/ai) | | [Workday](/platforms/workday/) | yes | Actions scoped to the user's identity with safeguards for sensitive enterprise data. | [www.workday.com/en-us/ai-platform/agent-tools-mc](https://www.workday.com/en-us/ai-platform/agent-tools-mcp.html) | | [Factorial](/platforms/factorial/) | yes | 'ONE enforces each person's permissions on every request. No permission, no action.' | [factorial.com/ai](https://factorial.com/ai) | | [Humaans](/platforms/humaans/) | yes | Role-based permissions and explicit data scopes when connecting. | [humaans.io/mcp](https://humaans.io/mcp) | | [Taito.ai](/platforms/taito/) | yes | Field-level access control. | [taito.ai/](https://taito.ai/) | | [Moveworks](/platforms/moveworks/) | yes | PII masking, content filtering, zero-day data retention. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | | [BambooHR](/platforms/bamboohr/) | partial | Access levels gate the connectors; no sensitive-field control named. | [documentation.bamboohr.com/docs/mcp-server](https://documentation.bamboohr.com/docs/mcp-server) | | [Personio](/platforms/personio/) | partial | Unlimited access and permissions on CorePro. | [www.personio.com/pricing/](https://www.personio.com/pricing/) | | [Gusto](/platforms/gusto/) | partial | Data categories chosen per connection. | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | [Lattice](/platforms/lattice/) | partial | Answers scoped to each user's Lattice access. | [lattice.com/platform/mcp](https://lattice.com/platform/mcp) | | [Sage](/platforms/sage/) | partial | Permission levels in Sage HR. | [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) | | [ADP](/platforms/adp/) | partial | Role-based access in Workforce Now. | [www.adp.com/what-we-offer/products/adp-workforce](https://www.adp.com/what-we-offer/products/adp-workforce-now/capabilities) | | [Paylocity](/platforms/paylocity/) | partial | Role-based security in the HCM. | [www.paylocity.com/products/hr/](https://www.paylocity.com/products/hr/) | | [Leena AI](/platforms/leena/) | partial | Role-based access and policy guardrails. | [leena.ai/](https://leena.ai/) | | [Deel](/platforms/deel/) | none found | Not stated on the AI pages read. | [www.deel.com/hr/](https://www.deel.com/hr/) | Ordered yes, partial, none found. Read 8 Nov 2026. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Shapes: https://help.shapes.co/en/articles/9569339-permission-levels (read 8 Nov 2026) - HiBob: https://www.hibob.com/platform/core/ai/agents/ (read 8 Nov 2026) - Rippling: https://www.rippling.com/platform/ai (read 8 Nov 2026) - Workday: https://www.workday.com/en-us/ai-platform/agent-tools-mcp.html (read 8 Nov 2026) - Factorial: https://factorial.com/ai (read 8 Nov 2026) - Humaans: https://humaans.io/mcp (read 8 Nov 2026) - Taito.ai: https://taito.ai/ (read 8 Nov 2026) - Moveworks: https://www.moveworks.com/us/en/solutions/ai-for-hr (read 8 Nov 2026) - BambooHR: https://documentation.bamboohr.com/docs/mcp-server (read 8 Nov 2026) - Personio: https://www.personio.com/pricing/ (read 8 Nov 2026) - Gusto: https://gusto.com/product/integrations/gusto-mcp (read 8 Nov 2026) - Lattice: https://lattice.com/platform/mcp (read 8 Nov 2026) - Sage: https://www.sage.com/en-us/sage-business-cloud/hr/ (read 8 Nov 2026) - ADP: https://www.adp.com/what-we-offer/products/adp-workforce-now/capabilities (read 8 Nov 2026) - Paylocity: https://www.paylocity.com/products/hr/ (read 8 Nov 2026) - Leena AI: https://leena.ai/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/capabilities/sensitive_controls/ --- # Official MCP server or connector, with access level Capability, read 8 Nov 2026 **The short answer.** 10 of 17 platforms document "official mcp server or connector, with access level" on the pages read on 8 Nov 2026: Shapes, HiBob, Rippling, Deel, Workday, Gusto, Lattice, Factorial, Humaans, Taito.ai. 2 document a narrower form: BambooHR, Personio. For 5 nothing was found on the pages listed. | Platform | Found | Evidence | Source | | --- | --- | --- | --- | | [Shapes](/platforms/shapes/) | yes | mcp.shapes.co: 35 read-only tools, OAuth 2.0 with PKCE, official Claude connector. | [mcp.shapes.co/](https://mcp.shapes.co/?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=capability) | | [HiBob](/platforms/hibob/) | yes | HiBob MCP server, read and write where Bob already allows. | [www.hibob.com/platform/core/ai/mcp](https://www.hibob.com/platform/core/ai/mcp) | | [Rippling](/platforms/rippling/) | yes | Rippling MCP page with ask_ai (read-only) and an admin access policy; no release label. | [www.rippling.com/platform/ai/rippling-mcp](https://www.rippling.com/platform/ai/rippling-mcp) | | [Deel](/platforms/deel/) | yes | Official MCP server over HTTP with read and write tools, OAuth scopes checked per call (developer docs, read 2026-11-10). | [developer.deel.com/mcp/introduction](https://developer.deel.com/mcp/introduction) | | [Workday](/platforms/workday/) | yes | Hundreds of Agent-Ready Tools over MCP; no release label. | [www.workday.com/en-us/ai-platform/agent-tools-mc](https://www.workday.com/en-us/ai-platform/agent-tools-mcp.html) | | [Gusto](/platforms/gusto/) | yes | mcp.api.gusto.com, OAuth2, 31,000+ companies connected. | [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) | | [Lattice](/platforms/lattice/) | yes | Lattice MCP server, labelled New. | [lattice.com/platform/mcp](https://lattice.com/platform/mcp) | | [Factorial](/platforms/factorial/) | yes | Hosted MCP server at mcp.factorialhr.com with OAuth 2.0, added to AI clients by URL; not yet in the Claude or ChatGPT connector directories (API docs, read 2026-11-10). | [apidoc.factorialhr.com/docs/mcp-server](https://apidoc.factorialhr.com/docs/mcp-server) | | [Humaans](/platforms/humaans/) | yes | Included at no extra cost for all customers, OAuth 2.0 with DCR, write actions. | [humaans.io/mcp](https://humaans.io/mcp) | | [Taito.ai](/platforms/taito/) | yes | MCP interface for Claude web, Claude Code or any MCP host; status not stated. | [taito.ai/](https://taito.ai/) | | [BambooHR](/platforms/bamboohr/) | partial | MCP server and AI connectors in beta, OAuth 2.0, mostly read-only with Write and Delete tools. | [documentation.bamboohr.com/docs/mcp-server](https://documentation.bamboohr.com/docs/mcp-server) | | [Personio](/platforms/personio/) | partial | Official MCP server in open beta on Core Pro: read access, limited write that is off by default; named clients Claude, ChatGPT, VS Code with GitHub Copilot (help center, read 2026-11-10). | [support.personio.de/hc/en-us/articles/3831465667](https://support.personio.de/hc/en-us/articles/38314656674333) | | [Sage](/platforms/sage/) | none found | No MCP server found. | [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) | | [ADP](/platforms/adp/) | none found | No MCP server found; APIs via ADP Marketplace. | [www.adp.com/what-we-offer/products/adp-workforce](https://www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx) | | [Paylocity](/platforms/paylocity/) | none found | No MCP server found; API libraries. | [www.paylocity.com/pricing/](https://www.paylocity.com/pricing/) | | [Leena AI](/platforms/leena/) | none found | Not found on the home page. | [leena.ai/](https://leena.ai/) | | [Moveworks](/platforms/moveworks/) | none found | Not found on the HR page. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | Ordered yes, partial, none found. Read 8 Nov 2026. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Shapes: https://mcp.shapes.co/ (read 8 Nov 2026) - HiBob: https://www.hibob.com/platform/core/ai/mcp (read 8 Nov 2026) - Rippling: https://www.rippling.com/platform/ai/rippling-mcp (read 8 Nov 2026) - Deel: https://developer.deel.com/mcp/introduction (read 8 Nov 2026) - Workday: https://www.workday.com/en-us/ai-platform/agent-tools-mcp.html (read 8 Nov 2026) - Gusto: https://gusto.com/product/integrations/gusto-mcp (read 8 Nov 2026) - Lattice: https://lattice.com/platform/mcp (read 8 Nov 2026) - Factorial: https://apidoc.factorialhr.com/docs/mcp-server (read 8 Nov 2026) - Humaans: https://humaans.io/mcp (read 8 Nov 2026) - Taito.ai: https://taito.ai/ (read 8 Nov 2026) - BambooHR: https://documentation.bamboohr.com/docs/mcp-server (read 8 Nov 2026) - Personio: https://support.personio.de/hc/en-us/articles/38314656674333 (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/capabilities/mcp_official/ --- # AI governance certification held or in progress (ISO 42001) Capability, read 8 Nov 2026 **The short answer.** 2 of 17 platforms document "ai governance certification held or in progress (iso 42001)" on the pages read on 8 Nov 2026: Rippling, Workday. 1 document a narrower form: Shapes. For 14 nothing was found on the pages listed. Each row below quotes the page. | Platform | Found | Evidence | Source | | --- | --- | --- | --- | | [Rippling](/platforms/rippling/) | yes | ISO 42001 certified. | [www.rippling.com/trust/security](https://www.rippling.com/trust/security) | | [Workday](/platforms/workday/) | yes | ISO 42001 certificate; NIST AI RMF attestation. | [www.workday.com/en-us/why-workday/trust/complian](https://www.workday.com/en-us/why-workday/trust/compliance.html) | | [Shapes](/platforms/shapes/) | partial | ISO 42001 in progress on the security page. | [shapes.co/security-and-compliance](https://shapes.co/security-and-compliance?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=capability) | | [HiBob](/platforms/hibob/) | none found | ISO 42001 not named; GDPR, SOC 2, ISO 27001, HIPAA badges. | [www.hibob.com/platform/core/ai/mcp](https://www.hibob.com/platform/core/ai/mcp) | | [BambooHR](/platforms/bamboohr/) | none found | Not named on the security page. | [www.bamboohr.com/security](https://www.bamboohr.com/security) | | [Personio](/platforms/personio/) | none found | Responsible AI principles; ISO 42001 not named. | [www.personio.com/security/](https://www.personio.com/security/) | | [Deel](/platforms/deel/) | none found | Not named. | [www.deel.com/security/](https://www.deel.com/security/) | | [Gusto](/platforms/gusto/) | none found | Not named. | [gusto.com/security](https://gusto.com/security) | | [Lattice](/platforms/lattice/) | none found | Not named. | [trustcenter.lattice.com/](https://trustcenter.lattice.com/) | | [Factorial](/platforms/factorial/) | none found | Not named. | [trust.factorial.co/](https://trust.factorial.co/) | | [Humaans](/platforms/humaans/) | none found | SOC 2 Type II, ISO 27001; ISO 42001 not named. | [humaans.io/mcp](https://humaans.io/mcp) | | [Sage](/platforms/sage/) | none found | Not named. | [www.sage.com/en-us/trust-security/security/techn](https://www.sage.com/en-us/trust-security/security/technical/standards-compliance/) | | [ADP](/platforms/adp/) | none found | Not named. | [www.adp.com/knowledge/payroll/small-business/sol](https://www.adp.com/knowledge/payroll/small-business/solutions/security) | | [Paylocity](/platforms/paylocity/) | none found | Not named. | [trust.paylocity.com/](https://trust.paylocity.com/) | | [Taito.ai](/platforms/taito/) | none found | ISO 27001 and GDPR; ISO 42001 not named. | [taito.ai/](https://taito.ai/) | | [Leena AI](/platforms/leena/) | none found | 'Responsible AI'; no certification named. | [leena.ai/](https://leena.ai/) | | [Moveworks](/platforms/moveworks/) | none found | FedRAMP, SOC 2, HIPAA, GDPR; ISO 42001 not named. | [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) | Ordered yes, partial, none found. Read 8 Nov 2026. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Rippling: https://www.rippling.com/trust/security (read 8 Nov 2026) - Workday: https://www.workday.com/en-us/why-workday/trust/compliance.html (read 8 Nov 2026) - Shapes: https://shapes.co/security-and-compliance (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/capabilities/ai_governance_cert/ --- # MCP in HR software: status, access and what each server does, read from the vendor's own page MCP status, read 8 Nov 2026 **The short answer.** Eight of 17 platforms have an official MCP server live for all customers on 8 Nov 2026 (Shapes, HiBob, Rippling, Workday, Gusto, Lattice, Humaans, Taito.ai); BambooHR's is in beta, Personio's is announced, and the matrix shows none for the rest, though documentation read on 10 Nov 2026 describes servers for Deel and Factorial. Shapes's is read-only. An MCP server lets an outside AI client (Claude, ChatGPT, Gemini, Cursor) read, and sometimes write, the HRIS record under the connecting user's permissions. The status is read from the vendor's own page and nothing else: a directory listing, a partner's wrapper or a news story does not count, which is why Personio reads 'announced' and the wrappers around it do not make it live. The access level is the second fact: Shapes exposes 35 read-only tools; HiBob writes where Bob allows; Humaans writes for every customer; Gusto writes with a confirmation on every change; Lattice drafts and submits reviews; Workday executes actions through its Agent Gateway; Rippling's ask_ai tool is read-only with an admin-granted write policy. | Platform | Status | Access | Evidence | | --- | --- | --- | --- | | [Shapes](/mcp/shapes/) | live for all customers | read | mcp.shapes.co: 35 read-only tools, OAuth 2.0 with PKCE, official Claude connector. | | [HiBob](/mcp/hibob/) | live for all customers | read and write where Bob allows | HiBob MCP server, read and write where Bob already allows. | | [Rippling](/mcp/rippling/) | live for all customers | read, with admin-granted write policy | Rippling MCP page with ask_ai (read-only) and an admin access policy; no release label. | | [BambooHR](/mcp/bamboohr/) | beta or early access | read, write and delete tools (beta) | MCP server and AI connectors in beta, OAuth 2.0, mostly read-only with Write and Delete tools. | | [Personio](/mcp/personio/) | beta or early access | read, limited write off by default | Official MCP server in open beta on Core Pro: read access, limited write that is off by default; named clients Claude, ChatGPT, VS Code with GitHub Copilot (help center, read 2026-11-10). | | [Deel](/mcp/deel/) | live for all customers | read and write (OAuth scopes per tool) | Official MCP server over HTTP with read and write tools, OAuth scopes checked per call (developer docs, read 2026-11-10). | | [Workday](/mcp/workday/) | live for all customers | executes Workday actions (write) | Hundreds of Agent-Ready Tools over MCP; no release label. | | [Gusto](/mcp/gusto/) | live for all customers | read and write with confirmation | mcp.api.gusto.com, OAuth2, 31,000+ companies connected. | | [Lattice](/mcp/lattice/) | live for all customers | read and write (reviews, updates) | Lattice MCP server, labelled New. | | [Factorial](/mcp/factorial/) | live for all customers | read and write (OAuth 2.0) | Hosted MCP server at mcp.factorialhr.com with OAuth 2.0, added to AI clients by URL; not yet in the Claude or ChatGPT connector directories (API docs, read 2026-11-10). | | [Humaans](/mcp/humaans/) | live for all customers | read and write | Included at no extra cost for all customers, OAuth 2.0 with DCR, write actions. | | [Sage](/mcp/sage/) | none found | not stated | No MCP server found. | | [ADP](/mcp/adp/) | none found | not stated | No MCP server found; APIs via ADP Marketplace. | | [Paylocity](/mcp/paylocity/) | none found | not stated | No MCP server found; API libraries. | | [Taito.ai](/mcp/taito/) | live for all customers | not stated | MCP interface for Claude web, Claude Code or any MCP host; status not stated. | | [Leena AI](/mcp/leena/) | none found | not stated | Not found on the home page. | | [Moveworks](/mcp/moveworks/) | none found | not stated | Not found on the HR page. | Status values: live for all customers; beta or early access; announced; third-party wrapper only; none found. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Shapes: https://mcp.shapes.co/ (read 8 Nov 2026) - HiBob: https://www.hibob.com/platform/core/ai/mcp (read 8 Nov 2026) - Rippling: https://www.rippling.com/platform/ai/rippling-mcp (read 8 Nov 2026) - BambooHR: https://documentation.bamboohr.com/docs/mcp-server (read 8 Nov 2026) - Personio: https://support.personio.de/hc/en-us/articles/38314656674333 (read 8 Nov 2026) - Deel: https://developer.deel.com/mcp/introduction (read 8 Nov 2026) - Workday: https://www.workday.com/en-us/ai-platform/agent-tools-mcp.html (read 8 Nov 2026) - Gusto: https://gusto.com/product/integrations/gusto-mcp (read 8 Nov 2026) - Lattice: https://lattice.com/platform/mcp (read 8 Nov 2026) - Factorial: https://apidoc.factorialhr.com/docs/mcp-server (read 8 Nov 2026) - Humaans: https://humaans.io/mcp (read 8 Nov 2026) - Sage: https://www.sage.com/en-us/sage-business-cloud/hr/ (read 8 Nov 2026) - ADP: https://www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx (read 8 Nov 2026) - Paylocity: https://www.paylocity.com/pricing/ (read 8 Nov 2026) - Taito.ai: https://taito.ai/ (read 8 Nov 2026) - Leena AI: https://leena.ai/ (read 8 Nov 2026) - Moveworks: https://www.moveworks.com/us/en/solutions/ai-for-hr (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/mcp/ --- # Shapes MCP server MCP, Shapes, read 8 Nov 2026 **The short answer.** Shapes's MCP status on 8 Nov 2026: live for all customers, read. mcp.shapes.co: 35 read-only tools, OAuth 2.0 with PKCE, official Claude connector. The status is read from Shapes's own page, linked under Sources; a secondary source is not accepted for status. - **Status:** live for all customers - **Access:** read - **Evidence:** mcp.shapes.co: 35 read-only tools, OAuth 2.0 with PKCE, official Claude connector. - **Source:** [mcp.shapes.co/](https://mcp.shapes.co/?utm_source=bestaiagentsforhr.com&utm_medium=referral&utm_campaign=mcp) - **Off switch:** yes: Separate switches for AI chat, AI-built Vibes and MCP access, on by default. - **Sensitive fields:** yes: Field-level permissions: edit, view or hidden per permission level (help center), and a release note on restricting fields. - **AI governance:** partial: ISO 42001 in progress on the security page. ## What an outside agent can do through it Read, under the connecting user's permissions. An approval step is stated: 'Nothing goes live until you say so.' Directory built with an agent writes nothing until Approve. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Shapes: shapes.co/releases/ai-chat-now-reads-the-help-center: https://shapes.co/releases/ai-chat-now-reads-the-help-center (read 8 Nov 2026) - Shapes: shapes.co/agents: https://shapes.co/agents (read 8 Nov 2026) - Shapes: shapes.co/: https://shapes.co/ (read 8 Nov 2026) - Shapes: shapes.co/releases/run-workflows-for-new-starters-and-leavers: https://shapes.co/releases/run-workflows-for-new-starters-and-leavers (read 8 Nov 2026) - Shapes: shapes.co/releases/workflows-2-0-now-live-for-everyone: https://shapes.co/releases/workflows-2-0-now-live-for-everyone (read 8 Nov 2026) - Shapes: shapes.co/releases/see-who-did-what-in-the-audit-log: https://shapes.co/releases/see-who-did-what-in-the-audit-log (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/mcp/shapes/ --- # HiBob MCP server MCP, HiBob, read 8 Nov 2026 **The short answer.** HiBob's MCP status on 8 Nov 2026: live for all customers, read and write where Bob allows. HiBob MCP server, read and write where Bob already allows. The status is read from HiBob's own page, linked under Sources; a secondary source is not accepted for status. - **Status:** live for all customers - **Access:** read and write where Bob allows - **Evidence:** HiBob MCP server, read and write where Bob already allows. - **Source:** [www.hibob.com/platform/core/ai/mcp](https://www.hibob.com/platform/core/ai/mcp) - **Off switch:** partial: MCP is turned on in Bob company-wide; no AI off switch stated for agents. - **Sensitive fields:** yes: Sensitive data protected by role-based permissions; agents cannot create a new path to restricted information. - **AI governance:** none found: ISO 42001 not named; GDPR, SOC 2, ISO 27001, HIPAA badges. ## What an outside agent can do through it Read and write, under the connecting user's permissions. No blanket approval step is stated for writes: Admins can require confirmation before writes in some AI tools; no approval step stated for agents. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - HiBob: www.hibob.com/platform/employee-experience-software/: https://www.hibob.com/platform/employee-experience-software/ (read 8 Nov 2026) - HiBob: www.hibob.com/platform/core/ai/agents/: https://www.hibob.com/platform/core/ai/agents/ (read 8 Nov 2026) - HiBob: www.hibob.com/platform/core/offboarding/: https://www.hibob.com/platform/core/offboarding/ (read 8 Nov 2026) - HiBob: www.hibob.com/platform/core/ai/mcp: https://www.hibob.com/platform/core/ai/mcp (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/mcp/hibob/ --- # Rippling MCP server MCP, Rippling, read 8 Nov 2026 **The short answer.** Rippling's MCP status on 8 Nov 2026: live for all customers, read, with admin-granted write policy. Rippling MCP page with ask_ai (read-only) and an admin access policy; no release label. The status is read from Rippling's own page, linked under Sources; a secondary source is not accepted for status. - **Status:** live for all customers - **Access:** read, with admin-granted write policy - **Evidence:** Rippling MCP page with ask_ai (read-only) and an admin access policy; no release label. - **Source:** [www.rippling.com/platform/ai/rippling-mcp](https://www.rippling.com/platform/ai/rippling-mcp) - **Off switch:** yes: 'Admins can enable or disable AI features.' - **Sensitive fields:** yes: Data permissions configured at the platform level. - **AI governance:** yes: ISO 42001 certified. ## What an outside agent can do through it Read and write, under the connecting user's permissions. An approval step is stated: 'Before it takes action, it gets your approval.' **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Rippling: www.rippling.com/platform/ai: https://www.rippling.com/platform/ai (read 8 Nov 2026) - Rippling: www.rippling.com/platform/ai/rippling-mcp: https://www.rippling.com/platform/ai/rippling-mcp (read 8 Nov 2026) - Rippling: www.rippling.com/trust/security: https://www.rippling.com/trust/security (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/mcp/rippling/ --- # BambooHR MCP server MCP, BambooHR, read 8 Nov 2026 **The short answer.** BambooHR's MCP status on 8 Nov 2026: beta or early access, read, write and delete tools (beta). MCP server and AI connectors in beta, OAuth 2.0, mostly read-only with Write and Delete tools. The status is read from BambooHR's own page, linked under Sources; a secondary source is not accepted for status. - **Status:** beta or early access - **Access:** read, write and delete tools (beta) - **Evidence:** MCP server and AI connectors in beta, OAuth 2.0, mostly read-only with Write and Delete tools. - **Source:** [documentation.bamboohr.com/docs/mcp-server](https://documentation.bamboohr.com/docs/mcp-server) - **Off switch:** yes: Admins turn Bamboo AI off or on under Account Settings > Ask BambooHR. - **Sensitive fields:** partial: Access levels gate the connectors; no sensitive-field control named. - **AI governance:** none found: Not named on the security page. ## What an outside agent can do through it Read and write, under the connecting user's permissions. An approval step is stated: 'still go through the appropriate person before they're applied or sent'. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - BambooHR: www.bamboohr.com/product-updates/bamboo-ai: https://www.bamboohr.com/product-updates/bamboo-ai (read 8 Nov 2026) - BambooHR: documentation.bamboohr.com/docs/mcp-server: https://documentation.bamboohr.com/docs/mcp-server (read 8 Nov 2026) - BambooHR: www.bamboohr.com/security: https://www.bamboohr.com/security (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/mcp/bamboohr/ --- # Personio MCP server MCP, Personio, read 8 Nov 2026 **The short answer.** Personio's MCP status on 8 Nov 2026: beta or early access, read, limited write off by default. Official MCP server in open beta on Core Pro: read access, limited write that is off by default; named clients Claude, ChatGPT, VS Code with GitHub Copilot (help center, read 2026-11-10). - **Status:** beta or early access - **Access:** read, limited write off by default - **Evidence:** Official MCP server in open beta on Core Pro: read access, limited write that is off by default; named clients Claude, ChatGPT, VS Code with GitHub Copilot (help center, read 2026-11-10). - **Source:** [support.personio.de/hc/en-us/articles/38314656674333](https://support.personio.de/hc/en-us/articles/38314656674333) - **Off switch:** none found: Not stated. - **Sensitive fields:** partial: Unlimited access and permissions on CorePro. - **AI governance:** none found: Responsible AI principles; ISO 42001 not named. ## What an outside agent can do through it Read and write, under the connecting user's permissions. No blanket approval step is stated for writes: 'A person always makes the decision' (screening); every agent gets an approval status in the AI Transformation Hub. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Personio: www.personio.com/product/assistant/: https://www.personio.com/product/assistant/ (read 8 Nov 2026) - Personio: www.personio.com/whats-new/: https://www.personio.com/whats-new/ (read 8 Nov 2026) - Personio: www.personio.com/product/onboarding/: https://www.personio.com/product/onboarding/ (read 8 Nov 2026) - Personio: www.personio.com/security/: https://www.personio.com/security/ (read 8 Nov 2026) - Personio: www.personio.com/pricing/: https://www.personio.com/pricing/ (read 8 Nov 2026) - Personio: support.personio.de/hc/en-us/articles/38314656674333: https://support.personio.de/hc/en-us/articles/38314656674333 (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/mcp/personio/ --- # Deel MCP server MCP, Deel, read 8 Nov 2026 **The short answer.** Deel's MCP status on 8 Nov 2026: live for all customers, read and write (OAuth scopes per tool). Official MCP server over HTTP with read and write tools, OAuth scopes checked per call (developer docs, read 2026-11-10). The status is read from Deel's own page, linked under Sources; a secondary source is not accepted for status. - **Status:** live for all customers - **Access:** read and write (OAuth scopes per tool) - **Evidence:** Official MCP server over HTTP with read and write tools, OAuth scopes checked per call (developer docs, read 2026-11-10). - **Source:** [developer.deel.com/mcp/introduction](https://developer.deel.com/mcp/introduction) - **Off switch:** none found: Not stated. - **Sensitive fields:** none found: Not stated on the AI pages read. - **AI governance:** none found: Not named. ## What an outside agent can do through it Read and write, under the connecting user's permissions. No blanket approval step is stated for writes: AI terms: 'Always keep a human in the loop'. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Deel: www.deel.com/hr/: https://www.deel.com/hr/ (read 8 Nov 2026) - Deel: www.deel.com/hr/compensation-management/: https://www.deel.com/hr/compensation-management/ (read 8 Nov 2026) - Deel: www.deel.com/hr/plugin/: https://www.deel.com/hr/plugin/ (read 8 Nov 2026) - Deel: www.deel.com/legal/ai-terms-of-service/: https://www.deel.com/legal/ai-terms-of-service/ (read 8 Nov 2026) - Deel: www.deel.com/security/: https://www.deel.com/security/ (read 8 Nov 2026) - Deel: developer.deel.com/mcp/introduction: https://developer.deel.com/mcp/introduction (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/mcp/deel/ --- # Workday MCP server MCP, Workday, read 8 Nov 2026 **The short answer.** Workday's MCP status on 8 Nov 2026: live for all customers, executes Workday actions (write). Hundreds of Agent-Ready Tools over MCP; no release label. The status is read from Workday's own page, linked under Sources; a secondary source is not accepted for status. - **Status:** live for all customers - **Access:** executes Workday actions (write) - **Evidence:** Hundreds of Agent-Ready Tools over MCP; no release label. - **Source:** [www.workday.com/en-us/ai-platform/agent-tools-mcp.html](https://www.workday.com/en-us/ai-platform/agent-tools-mcp.html) - **Off switch:** partial: Agents registered and governed in the Agent System of Record; no single off switch named. - **Sensitive fields:** yes: Actions scoped to the user's identity with safeguards for sensitive enterprise data. - **AI governance:** yes: ISO 42001 certificate; NIST AI RMF attestation. ## What an outside agent can do through it Read and write, under the connecting user's permissions. An approval step is stated: 'bring people in to review or approve sensitive actions before anything changes'; sandboxed what-ifs. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Workday: www.workday.com/en-us/artificial-intelligence/ai-agents/hr-service.html: https://www.workday.com/en-us/artificial-intelligence/ai-agents/hr-service.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/artificial-intelligence/ai-agents.html: https://www.workday.com/en-us/artificial-intelligence/ai-agents.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/artificial-intelligence/ai-agents/admin.html: https://www.workday.com/en-us/artificial-intelligence/ai-agents/admin.html (read 8 Nov 2026) - Workday: blog.workday.com/en-us/managing-ai-powered-future-of-work.html: https://blog.workday.com/en-us/managing-ai-powered-future-of-work.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/artificial-intelligence/workday-sana.html: https://www.workday.com/en-us/artificial-intelligence/workday-sana.html (read 8 Nov 2026) - Workday: www.workday.com/en-us/ai-platform/agent-app-builder.html: https://www.workday.com/en-us/ai-platform/agent-app-builder.html (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/mcp/workday/ --- # Gusto MCP server MCP, Gusto, read 8 Nov 2026 **The short answer.** Gusto's MCP status on 8 Nov 2026: live for all customers, read and write with confirmation. mcp.api.gusto.com, OAuth2, 31,000+ companies connected. The status is read from Gusto's own page, linked under Sources; a secondary source is not accepted for status. - **Status:** live for all customers - **Access:** read and write with confirmation - **Evidence:** mcp.api.gusto.com, OAuth2, 31,000+ companies connected. - **Source:** [gusto.com/product/integrations/gusto-mcp](https://gusto.com/product/integrations/gusto-mcp) - **Off switch:** partial: Admin chooses data categories; users told to disable automatic tool execution. - **Sensitive fields:** partial: Data categories chosen per connection. - **AI governance:** none found: Not named. ## What an outside agent can do through it Read and write, under the connecting user's permissions. An approval step is stated: 'Every write comes back to you for confirmation before anything moves.' **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Gusto: gusto.com/pricing: https://gusto.com/pricing (read 8 Nov 2026) - Gusto: gusto.com/product/integrations/gusto-mcp: https://gusto.com/product/integrations/gusto-mcp (read 8 Nov 2026) - Gusto: gusto.com/security: https://gusto.com/security (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/mcp/gusto/ --- # Lattice MCP server MCP, Lattice, read 8 Nov 2026 **The short answer.** Lattice's MCP status on 8 Nov 2026: live for all customers, read and write (reviews, updates). Lattice MCP server, labelled New. The status is read from Lattice's own page, linked under Sources; a secondary source is not accepted for status. - **Status:** live for all customers - **Access:** read and write (reviews, updates) - **Evidence:** Lattice MCP server, labelled New. - **Source:** [lattice.com/platform/mcp](https://lattice.com/platform/mcp) - **Off switch:** partial: Admins activate the AI Agent in a setup experience; role access left to the AI system's admin. - **Sensitive fields:** partial: Answers scoped to each user's Lattice access. - **AI governance:** none found: Not named. ## What an outside agent can do through it Read and write, under the connecting user's permissions. No blanket approval step is stated for writes: Not stated for the agent. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Lattice: lattice.com/platform/ai-agent/hr-help-desk: https://lattice.com/platform/ai-agent/hr-help-desk (read 8 Nov 2026) - Lattice: lattice.com/platform/mcp: https://lattice.com/platform/mcp (read 8 Nov 2026) - Lattice: lattice.com/platform/ai-agent: https://lattice.com/platform/ai-agent (read 8 Nov 2026) - Lattice: lattice.com/platform/compensation/cycle-management: https://lattice.com/platform/compensation/cycle-management (read 8 Nov 2026) - Lattice: trustcenter.lattice.com/: https://trustcenter.lattice.com/ (read 8 Nov 2026) - Lattice: lattice.com/blog/july-2026-product-updates: https://lattice.com/blog/july-2026-product-updates (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/mcp/lattice/ --- # Factorial MCP server MCP, Factorial, read 8 Nov 2026 **The short answer.** Factorial's MCP status on 8 Nov 2026: live for all customers, read and write (OAuth 2.0). Hosted MCP server at mcp.factorialhr.com with OAuth 2.0, added to AI clients by URL; not yet in the Claude or ChatGPT connector directories (API docs, read 2026-11-10). - **Status:** live for all customers - **Access:** read and write (OAuth 2.0) - **Evidence:** Hosted MCP server at mcp.factorialhr.com with OAuth 2.0, added to AI clients by URL; not yet in the Claude or ChatGPT connector directories (API docs, read 2026-11-10). - **Source:** [apidoc.factorialhr.com/docs/mcp-server](https://apidoc.factorialhr.com/docs/mcp-server) - **Off switch:** none found: Not stated. - **Sensitive fields:** yes: 'ONE enforces each person's permissions on every request. No permission, no action.' - **AI governance:** none found: Not named. ## What an outside agent can do through it Read and write, under the connecting user's permissions. An approval step is stated: 'It asks before it acts.' **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Factorial: factorial.com/ai: https://factorial.com/ai (read 8 Nov 2026) - Factorial: trust.factorial.co/: https://trust.factorial.co/ (read 8 Nov 2026) - Factorial: apidoc.factorialhr.com/docs/mcp-server: https://apidoc.factorialhr.com/docs/mcp-server (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/mcp/factorial/ --- # Humaans MCP server MCP, Humaans, read 8 Nov 2026 **The short answer.** Humaans's MCP status on 8 Nov 2026: live for all customers, read and write. Included at no extra cost for all customers, OAuth 2.0 with DCR, write actions. The status is read from Humaans's own page, linked under Sources; a secondary source is not accepted for status. - **Status:** live for all customers - **Access:** read and write - **Evidence:** Included at no extra cost for all customers, OAuth 2.0 with DCR, write actions. - **Source:** [humaans.io/mcp](https://humaans.io/mcp) - **Off switch:** none found: Not stated. - **Sensitive fields:** yes: Role-based permissions and explicit data scopes when connecting. - **AI governance:** none found: SOC 2 Type II, ISO 27001; ISO 42001 not named. ## What an outside agent can do through it Read and write, under the connecting user's permissions. No blanket approval step is stated for writes: Approvals and notifications; letters 'ready for approval'; no blanket 'nothing goes live' statement found. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Humaans: humaans.io/product-releases-2025: https://humaans.io/product-releases-2025 (read 8 Nov 2026) - Humaans: humaans.io/mcp: https://humaans.io/mcp (read 8 Nov 2026) - Humaans: humaans.io/careers: https://humaans.io/careers (read 8 Nov 2026) - Humaans: humaans.io/pricing: https://humaans.io/pricing (read 8 Nov 2026) - Humaans: humaans.io/ai: https://humaans.io/ai (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/mcp/humaans/ --- # Sage MCP server MCP, Sage, read 8 Nov 2026 **The short answer.** Sage's MCP status on 8 Nov 2026: none found. No MCP server found. An outside AI client cannot connect to Sage through MCP on the pages read; the API, where published, is the route. The off switch and the sensitive-field control that apply to the connector are listed below. - **Status:** none found - **Access:** not stated - **Evidence:** No MCP server found. - **Source:** [www.sage.com/en-us/sage-business-cloud/hr/](https://www.sage.com/en-us/sage-business-cloud/hr/) - **Off switch:** none found: Not found. - **Sensitive fields:** partial: Permission levels in Sage HR. - **AI governance:** none found: Not named. ## What an outside agent can do through it Nothing through MCP on 8 Nov 2026. Sage's pages describe no agent actions in other systems; a company that runs its AI stack outside the HRIS reaches Sage by API or export. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Sage: www.sage.com/en-us/sage-business-cloud/hr/: https://www.sage.com/en-us/sage-business-cloud/hr/ (read 8 Nov 2026) - Sage: www.sage.com/en-us/trust-security/security/technical/standards-compliance/: https://www.sage.com/en-us/trust-security/security/technical/standards-compliance/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/mcp/sage/ --- # ADP MCP server MCP, ADP, read 8 Nov 2026 **The short answer.** ADP's MCP status on 8 Nov 2026: none found. No MCP server found; APIs via ADP Marketplace. An outside AI client cannot connect to ADP through MCP on the pages read; the API, where published, is the route. The off switch and the sensitive-field control that apply to the connector are listed below. - **Status:** none found - **Access:** not stated - **Evidence:** No MCP server found; APIs via ADP Marketplace. - **Source:** [www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx](https://www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx) - **Off switch:** none found: Not found. - **Sensitive fields:** partial: Role-based access in Workforce Now. - **AI governance:** none found: Not named. ## What an outside agent can do through it Nothing through MCP on 8 Nov 2026. ADP's pages describe no agent actions in other systems; a company that runs its AI stack outside the HRIS reaches ADP by API or export. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - ADP: www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx: https://www.adp.com/what-we-offer/products/adp-workforce-now/compare-options.aspx (read 8 Nov 2026) - ADP: www.adp.com/what-we-offer/products/adp-workforce-now/hr-management: https://www.adp.com/what-we-offer/products/adp-workforce-now/hr-management (read 8 Nov 2026) - ADP: www.adp.com/what-we-offer/products/adp-workforce-now/talent: https://www.adp.com/what-we-offer/products/adp-workforce-now/talent (read 8 Nov 2026) - ADP: www.adp.com/what-we-offer/products/adp-workforce-now/capabilities: https://www.adp.com/what-we-offer/products/adp-workforce-now/capabilities (read 8 Nov 2026) - ADP: www.adp.com/knowledge/payroll/small-business/solutions/security: https://www.adp.com/knowledge/payroll/small-business/solutions/security (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/mcp/adp/ --- # Paylocity MCP server MCP, Paylocity, read 8 Nov 2026 **The short answer.** Paylocity's MCP status on 8 Nov 2026: none found. No MCP server found; API libraries. An outside AI client cannot connect to Paylocity through MCP on the pages read; the API, where published, is the route. The off switch and the sensitive-field control that apply to the connector are listed below. - **Status:** none found - **Access:** not stated - **Evidence:** No MCP server found; API libraries. - **Source:** [www.paylocity.com/pricing/](https://www.paylocity.com/pricing/) - **Off switch:** none found: Not found. - **Sensitive fields:** partial: Role-based security in the HCM. - **AI governance:** none found: Not named. ## What an outside agent can do through it Nothing through MCP on 8 Nov 2026. Paylocity's pages describe no agent actions in other systems; a company that runs its AI stack outside the HRIS reaches Paylocity by API or export. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Paylocity: www.paylocity.com/pricing/: https://www.paylocity.com/pricing/ (read 8 Nov 2026) - Paylocity: www.paylocity.com/products/hr/talent-management-system/onboarding/: https://www.paylocity.com/products/hr/talent-management-system/onboarding/ (read 8 Nov 2026) - Paylocity: www.paylocity.com/products/hr/talent-management-system/compensation-management/: https://www.paylocity.com/products/hr/talent-management-system/compensation-management/ (read 8 Nov 2026) - Paylocity: www.paylocity.com/products/hr/: https://www.paylocity.com/products/hr/ (read 8 Nov 2026) - Paylocity: trust.paylocity.com/: https://trust.paylocity.com/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/mcp/paylocity/ --- # Taito.ai MCP server MCP, Taito.ai, read 8 Nov 2026 **The short answer.** Taito.ai's MCP status on 8 Nov 2026: live for all customers, not stated. MCP interface for Claude web, Claude Code or any MCP host; status not stated. The status is read from Taito.ai's own page, linked under Sources; a secondary source is not accepted for status. - **Status:** live for all customers - **Access:** not stated - **Evidence:** MCP interface for Claude web, Claude Code or any MCP host; status not stated. - **Source:** [taito.ai/](https://taito.ai/) - **Off switch:** none found: Not stated. - **Sensitive fields:** yes: Field-level access control. - **AI governance:** none found: ISO 27001 and GDPR; ISO 42001 not named. ## What an outside agent can do through it Access as stated above, under the connecting user's permissions. No blanket approval step is stated for writes: Approvals sketched for time off only. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Taito.ai: taito.ai/: https://taito.ai/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/mcp/taito/ --- # Leena AI MCP server MCP, Leena AI, read 8 Nov 2026 **The short answer.** Leena AI's MCP status on 8 Nov 2026: none found. Not found on the home page. An outside AI client cannot connect to Leena AI through MCP on the pages read; the API, where published, is the route. The off switch and the sensitive-field control that apply to the connector are listed below. - **Status:** none found - **Access:** not stated - **Evidence:** Not found on the home page. - **Source:** [leena.ai/](https://leena.ai/) - **Off switch:** none found: Not stated. - **Sensitive fields:** partial: Role-based access and policy guardrails. - **AI governance:** none found: 'Responsible AI'; no certification named. ## What an outside agent can do through it Nothing through MCP on 8 Nov 2026. Leena AI's pages describe Workday, Oracle, SAP, ServiceNow, Slack, Microsoft; a company that runs its AI stack outside the HRIS reaches Leena AI by API or export. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Leena AI: leena.ai/: https://leena.ai/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/mcp/leena/ --- # Moveworks MCP server MCP, Moveworks, read 8 Nov 2026 **The short answer.** Moveworks's MCP status on 8 Nov 2026: none found. Not found on the HR page. An outside AI client cannot connect to Moveworks through MCP on the pages read; the API, where published, is the route. The off switch and the sensitive-field control that apply to the connector are listed below. - **Status:** none found - **Access:** not stated - **Evidence:** Not found on the HR page. - **Source:** [www.moveworks.com/us/en/solutions/ai-for-hr](https://www.moveworks.com/us/en/solutions/ai-for-hr) - **Off switch:** none found: Not stated. - **Sensitive fields:** yes: PII masking, content filtering, zero-day data retention. - **AI governance:** none found: FedRAMP, SOC 2, HIPAA, GDPR; ISO 42001 not named. ## What an outside agent can do through it Nothing through MCP on 8 Nov 2026. Moveworks's pages describe Workday, ADP and ServiceNow plugins; end-to-end task automation; a company that runs its AI stack outside the HRIS reaches Moveworks by API or export. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Moveworks: www.moveworks.com/us/en/solutions/ai-for-hr: https://www.moveworks.com/us/en/solutions/ai-for-hr (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/mcp/moveworks/ --- # Service desk agents against HRIS-native agents, on the same twelve capabilities Service desk agents, read 8 Nov 2026 **The short answer.** Leena AI and Moveworks answer employee questions and act in other systems as well as the HRIS-native agents do, and neither holds the record: both need an HRIS to read from and write to, and neither publishes an MCP server, an approval step or an off switch on 8 Nov 2026. The table compares them with four HRIS-native agents. A service desk agent sits in Slack or Teams and answers the questions a People team answers all day, routes tickets and generates letters; Leena AI names Workday, Oracle, SAP, ServiceNow, Slack and Microsoft as the systems it acts in, and Moveworks names Workday, ADP and ServiceNow plugins. The record each acts on lives in those systems. The comparison is reported rather than ranked: a service desk agent is scored on what it does with the HRIS record through connectors, and the ranking on this site covers the fourteen Index platforms only. | Platform | Answers employee questions from policy | Drafts changes | Runs onboarding end to end | Runs a compensation cycle | Runs offboarding | Acts in other systems (Slack, Teams, email, ATS, payroll) | Approval before anything goes live | Audit log of agent actions | Off switch for AI features | Sensitive-field controls | Official MCP server or connector, with access level | AI governance certification held or in progress (ISO 42001) | Yes | MCP | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | Leena AI | yes | partial | partial | none found | none found | yes | none found | yes | none found | partial | none found | none found | 3 | none found | | Moveworks | yes | partial | none found | none found | none found | yes | none found | none found | none found | yes | none found | none found | 3 | none found | | Shapes | yes | yes | yes | yes | partial | yes | yes | yes | yes | yes | yes | partial | 10 | live for all customers | | Humaans | yes | yes | yes | yes | partial | yes | partial | yes | none found | yes | yes | none found | 8 | live for all customers | | Rippling | yes | yes | partial | none found | partial | yes | yes | partial | yes | yes | yes | yes | 8 | live for all customers | | Workday | yes | yes | partial | none found | none found | yes | yes | yes | partial | yes | yes | yes | 8 | live for all customers | | Agent | Yes | Partial | MCP | What it needs an HRIS for | | --- | --- | --- | --- | --- | | [Leena AI](/platforms/leena/) | 3 | 3 | none found | the employee record, the policies and the systems it routes tickets to, all of which live in the HRIS and the tools it connects | | [Moveworks](/platforms/moveworks/) | 3 | 1 | none found | the HRIS record its plugins update (Workday, ADP) and the ticketing system it resolves into | | [Shapes](/platforms/shapes/) | 10 | 2 | live for all customers, read | it is the HRIS | | [Humaans](/platforms/humaans/) | 8 | 2 | live for all customers, read and write | it is the HRIS | | [Rippling](/platforms/rippling/) | 8 | 3 | live for all customers, read, with admin-granted write policy | it is the HRIS | | [Workday](/platforms/workday/) | 8 | 2 | live for all customers, executes Workday actions (write) | it is the HRIS | Reported, not ranked: a service desk agent is scored on what it does with the HRIS record through connectors, and the ranking on this site covers the 14 Index platforms only. **Where Shapes is weak.** Shapes's MCP server is documented as read access on the date read; write actions run inside Shapes, not through the connector. ## Sources - Leena AI: leena.ai/: https://leena.ai/ (read 8 Nov 2026) - Moveworks: www.moveworks.com/us/en/solutions/ai-for-hr: https://www.moveworks.com/us/en/solutions/ai-for-hr (read 8 Nov 2026) - Shapes: shapes.co/releases/ai-chat-now-reads-the-help-center: https://shapes.co/releases/ai-chat-now-reads-the-help-center (read 8 Nov 2026) - Shapes: shapes.co/agents: https://shapes.co/agents (read 8 Nov 2026) - Shapes: shapes.co/: https://shapes.co/ (read 8 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://bestaiagentsforhr.com/service-desk-agents/