Leapsome MCP is a secure connector for AI clients like Claude and ChatGPT. Your agents retrieve Leapsome context and take selected actions, and they see exactly what you can see in Leapsome, nothing more.
Works with Claude, ChatGPT, Notion AI, and any MCP-compatible client
Built on your existing Leapsome permissions and visibility rules
Agents can act for you and nothing is saved without your approval
Leapsome runs deep: AI that raises your talent density, built on the deepest people data foundation in HR. Your agents get each employee's full journey.
One connection serves the whole company, from a quick absence question to a leadership report. Ask in your AI client and Leapsome answers with live, permission-checked context.
Enable MCP
An admin enables MCP in Leapsome settings and chooses the permitted scopes, including whether to allow selected write actions.
Grant access
The admin grants access per user group in the feature access matrix.
Connect AI client
Employees connect their AI client, sign in, and approve the scopes they want to grant.
Where your admin enables it, agents can act: request time off, update goals, add talking points, give praise, or start a workflow. Every write uses a confirmation step: your agent proposes, you approve, and only then does Leapsome save it.
If your AI client also connects to Slack, Notion, calendars, or project tools, your agent holds Leapsome alongside them in one conversation.
Leapsome's MCP uses the same permissions, visibility rules, and anonymity thresholds as the platform itself. Admins control scopes and write access. EU-hosted, GDPR compliant, ISO 27001 certified, and EU AI Act compliant.
Take Leapsome for a spin
Learn how Leapsome can automate, connect, and simplify your HR processes.
Book your demo
.png)
Leapsome MCP is a secure connector based on the Model Context Protocol, an open standard that links AI clients to external systems. It lets AI tools like Claude, ChatGPT custom connectors, or Notion AI retrieve Leapsome context and, where enabled, take selected actions on your behalf. Because it runs on the deepest people data foundation in HR, your agents work with each employee's full journey instead of a thin slice of admin data.
Any MCP-compatible client. That includes Claude, ChatGPT custom connectors, Notion AI, custom agents built in Microsoft Copilot, and agents your team builds itself. Clients with OAuth support connect with a standard sign-in; personal access tokens are available for clients without it.
Exactly what the connected user can already access in Leapsome, and nothing more. Depending on your modules, scopes, and permissions, that spans four areas: people and org context (employees, teams, reporting lines, knowledge base), performance and growth (goals, feedback, reviews, meetings, surveys, learning), analytics and reporting, and operations (absences, workflows, timesheets, forms). Visibility rules and anonymity thresholds apply through MCP just as they do in the platform.
Yes, where your admin enables write access. Agents can take selected actions like creating or updating goals, adding meeting talking points, giving praise, requesting time off, or starting a self-service workflow. An employee can ask "Can I take next Friday off?" and the agent checks their absence balance, then starts the request. Every write action uses a confirmation step, so nothing is saved until the user approves it.
No, and it isn't meant to be. Leapsome MCP is a focused toolset built for specific outcomes: preparing, summarizing, drafting, and taking selected actions. Your agent supports the work, and people review and decide. The toolset grows based on how customers actually use it.
The connector is becoming standard; the data behind it is the difference. Separate HR systems keep separate data models, permissions, histories, and definitions, so an agent stitching across them adds complexity and compliance risk, especially with intentionally anonymized data like engagement results. Leapsome holds the full employee journey in one governed people data foundation, so your AI works with connected, privacy-conscious context from the start.
Leapsome AI Agents work inside the platform, doing talent work like coaching, development, and people analytics on your data foundation. Leapsome MCP brings that same foundation into the AI tools you run elsewhere, so your own agents and assistants can use it too. The two work best together.
Yes. Connecting an AI client never widens access to your data: MCP uses the same login, permissions, and boundaries as the Leapsome platform. Admins define scopes and can restrict the integration to specific user groups. Leapsome is EU-hosted, GDPR compliant, ISO 27001 certified, and EU AI Act compliant.
A super admin enables MCP in Leapsome settings, selects the permitted scopes, and grants access per user group in the feature access matrix. Employees then connect their AI client and approve the scopes they want to grant. Setup takes minutes. Once connected, describe the outcome you want and let the agent find the right Leapsome tools.
Yes. Leapsome MCP is included in all plans.
From the first candidate conversation to the next promotion, base your people decisions on connected people & performance data, along the entire employee lifecycle.
Replace your crowded HR stack with one modern platform
See how Leapsome connects HRIS and talent management to elevate HR, empower teams, and deliver real impact.
Get a personalized demo
Get a personalized walkthrough and see exactly how Leapsome can work for your team. You’ll know in one call if it’s the right fit.
