DocsGuides and reference for @Human
Build, govern and run AI systems with @Human.
We're publishing the full documentation during early access. Start with the demo below, see which guides are on the way, or book a walkthrough and we'll take you through @Human live, built around your own use case.
See @Human in action
A replay of a real MCP session with a live @Human instance: an AI assistant lists projects, reads runtime guardrails, checks monitoring rules and reads the fundamental rights framework, all within the user's own permissions.
- Connect any MCP assistantEvery @Human instance includes an MCP server. The assistant works as you, within your permissions.
- Ask and exploreRead projects, guardrails, monitoring rules and frameworks in plain language.
- Changes land as draftsAnything the assistant writes goes into a draft version and triggers an impact review.
- Only people approveApproval gates can be looked up, never signed, by a tool. Everything is logged.
What the documentation will cover
Six areas, published as they're ready. Want one sooner? Tell us in a walkthrough and we'll prioritise it.
Getting started
Install @Human, create your first project and invite your team.
Building applications
Data models, forms, workflows and agents, with or without AI.
Governance and compliance
Policies, risk scenarios, the data register, approval gates and EU AI Act evidence.
Runtime guardrails
Monitoring rules, incident tracking and real-time alerting for systems in production.
MCP and integrations
Connect any MCP assistant, plus APIs, workflows and OAuth2 clients.
Self-hosting and deployment
Run @Human in the cloud you choose or on-premises, inside your own controls.