← back
CIAM for AI: Authn/Authz for Agents — Michael Grinich, CEO of WorkOS
Takeaway
Agents need a new identity class — neither human nor service-account — with patterns like shadow personas, delegation chains, and capability tokens to keep them safe in enterprise systems.
Summary
- AI is WorkOS's fastest-growing segment because every agentic product needs identity; agents are a new identity class — neither users nor machine-to-machine.
- Hard problems: headless sign-in, long-lived credential storage, scoped vs broad permissions for nondeterministic systems, compliance/audit trail tied to a real person, and the scale at which agents can mistake-loop.
- Four emerging architecture patterns: persona shadowing (agent acts as scoped-down version of human user), delegation chains (JWT-style tokens passed system-to-system), capability tokens (e.g., 'agent X may read Bob's calendar for 60 min'), and escalation to humans (HITL).
- Persona shadowing is the most common in practice today; delegation chains can extend OIDC.
- Calls for industry standards collaboration to define agent identity primitives.
agent-identityauthnworkos
Original description
AI agents are changing the way modern SaaS products operate. Whether automating workflows, integrating with APIs, or acting on behalf of users, AI-driven assistants and autonomous systems are becoming core product features. But securing these agents presents a fundamental challenge: How do you authenticate AI agents? How do you control what they can access? How do you ensure they act within the right permissions? This talk will explore these concepts and more while highlighting current research and best practices. ---related links--- https://x.com/grinich/ https://www.linkedin.com/in/grinich/ https://workos.com/guides https://workos.com/