What is the right architecture for multi-tenant agent governance?
Answer
Tenant isolation by org_id in every query — ideally enforced by the database as well (row-level security), so a query that forgets the filter returns nothing rather than another tenant’s rows. Feature flags per org enable/disable capabilities by plan tier. Encrypt sensitive data; Rends derives a per-org key for data exports, and other encrypted fields use a platform key. Shared infrastructure with logical isolation. API keys scoped to a single org and hashed with SHA-256 (never stored in plaintext). Session tokens carry the user’s organization membership and role, and the gateway checks them against the organization’s record before any request reaches an engine.
Tags
- multi-tenant
- architecture
Put governance into production
See how teams inventory agents, enforce policies, and ship audit-ready evidence on one platform.