Proovable AI is built for multi-tenant agent operations. Authentication, authorization, tenant isolation, audit capture, and rate controls are enforced in the application path.
Authentication uses Microsoft Entra ID. Role-based permissions determine which people can view, change, approve, run, and govern work.
Tenant-scoped records are enforced through SQL row-level security and application-level query filters. Tenant context follows each request.
HTTPS enforcement, security headers, correlation IDs, rate limits, authorization, IP allowlists, and idempotency controls operate before protected endpoints execute.
Important actions create an audit record. Policy violations, approvals, changes, and incident activity retain their history rather than overwriting it.
Production secrets are supplied through managed identity and Azure Key Vault references, not application configuration files.
Public and embedded agent surfaces validate their own access path and apply origin, authentication, rate, and tenant controls.
Policies can restrict data handling, topics, tools, access, approval requirements, models, and spend. Violations can block, redact, notify, or escalate.
Append-only audit records and attestation capabilities support reviewable evidence for governed agent activity and system changes.
Security is not a separate report. It is the set of boundaries that determine who can act, what can run, and what the organization can later verify.