Wants AI access to follow existing authorisations, not bypass them.
Platform · Role-based access
Role-based knowledge access
Policy, documentation and internal knowledge are routed by role, with end-to-end RBAC over both retrieval and inference. No one sees more through the AI than they are allowed to see.
Who it’s for
Wants certainty that sensitive documents do not end up in an answer through a back door.
Wants to reuse roles from its own IdP, not maintain a separate access model.
The problem
An AI that has read everything leaks everything.
RAG systems sitting on one big knowledge source give everyone access to everything the model has ingested, regardless of what that person is allowed to see.
Numaga applies role-based access across the whole chain: what is retrieved and what is inferred both stay within the user’s permissions.
How it works
Access that follows the role, from retrieval to answer.
RBAC is not a filter at the end; it applies at every step.
Roles from your own IdP
Access maps to the roles and groups you already manage in your IdP, Microsoft Entra ID or Google Workspace: no second authorisation model.
Retrieval within the role
The knowledge base returns only documents the user is entitled to; the rest is never retrieved.
Inference within the role
The model, too, works only with context the user is allowed to see: no leakage through a back door.
Recorded per interaction
Who saw which context sits in the audit trail, traceable at every review.
In practice
Proven where data is most sensitive.
Numaga is running a pilot with Ontdek Zorg, a home-care organisation in Gelderland. Healthcare, with special-category personal data under the GDPR, is where responsible AI is hardest. That is exactly where we put the control plane to the test.
- Home care in Gelderland
- Special-category data (GDPR Art. 9)
- Policy and redaction on every prompt
Map your roles and knowledge sources.
We show how RBAC over retrieval and inference works in your environment.
Request demo access