The RBAC console that distributes the right document to the right people — humans and AI agents.
The K-AI admin console: groups, document instances, business links. K-AI MCP Retrieval inherits the configured RBAC automatically — no ACL leak possible on the agent side.
The DKP, in practice.
K-AI implements the DKP across three surfaces — Audit, Platform, MCP — that operate the five layers.
K-AI Platform — the RBAC admin console
The web platform where admins compose the access model: groups, users, K-AI instances (knowledge bases), and the business links between them. K-AI MCP Retrieval fully inherits this policy — zero config on the agent side.
Groups & permissions
A group binds a set of users to a set of K-AI instances. The admin decides who sees what.
- HSE — Europe Operatorshse-europesafety-iso
- Legal — Contractslegal-contractscompliance-eu
- R&D — Platformsrd-platformsrd-patentsrd-archive
- Industrial Leadershiphse-europerd-platformslegal-contracts
Business links between K-AI instances
The admin declares which bases complement each other. K-AI MCP Retrieval uses these links to route a query to the right instances — rather than querying everything.
- hse-europe→safety-isonormative reference
- legal-contracts→compliance-euregulatory frame
- rd-platforms→rd-patentsrelated IP
- rd-platforms→rd-archiveproject history
Who is it for?
Operated by the Document Authority (CDO or CIO). Document Owners configure their scopes. Document Consumers (teams, AI agents) consume via MCP — without ever seeing the console.
See the relevant roles →Documentation
RBAC model, MCP inheritance, inter-instance links — schemas and API documented on GitBook.
GitBook documentation →