Announcing the AI Agent Identity and Access Workshop

Today we are adding a fifth workshop to the Watchlight AI advisory portfolio: AI Agent Identity and Access. It is a hands-on engagement that designs and stands up identity and access for your AI agents, with the outcome enforced at runtime under Agent Runtime Governance.

Bottom line for CISOs. AI agents are becoming the fastest-growing population of identities in your enterprise, and most of them hold long-lived credentials with broad, unscoped access. Knowing who an agent is does not govern what it does. This workshop leaves your team with a working reference implementation of verifiable agent identity, scoped and time-bound access, and governed delegation, all enforced at the moment of action.

Why this workshop, and why now

Every identity and security team we work with is facing the same shift. Agents now authenticate to systems, hold tokens, call APIs, and act on behalf of users and of each other. They are identities in every operational sense, and they are multiplying faster than the directory was ever designed to handle.

The instinct is to extend existing identity tooling to cover them. That instinct is correct and incomplete. An agent needs a unique, verifiable identity (Principle 1). It also needs that identity to carry meaning at the moment it acts: what authority it holds, how that authority was scoped, who it can act for, and when the grant expires (Principle 3). Issuing a credential answers the first question. It does not answer the rest. As we have written before, identity is necessary and not sufficient, and legacy IGA was not built for this population.

This workshop closes that gap. It treats agent identity as something to be issued, scoped, delegated, and revoked under runtime governance, rather than provisioned once and trusted thereafter.

What your team produces

The engagement runs two to three days and is built for identity architects, IAM teams, security engineers, and platform and AI infrastructure leads. It is design and build, not slideware.

Focus areas:

• Unique, verifiable agent identity for every autonomous actor
• Agent identity lifecycle: provisioning, trust state, rotation, and revocation
• Scoped, time-bound authority and zero standing privilege for agents
• Governed delegation: who an agent can act for, and how far authority propagates
• Verifiable credentials and attestation across trust boundaries
• Integration with your existing IAM, identity provider, and secrets infrastructure

Deliverables you keep:

• An agent identity and access architecture mapped to your IAM stack
• A working reference implementation of agent identity and scoped access in your environment
• A delegation and authority model with runtime enforcement points
• An operational rollout plan for provisioning, rotation, and revocation at scale

Where it fits in the workshop lineup

The Watchlight AI workshops cover the full arc of an agent governance program. The new workshop sits at the foundation of the technical track. Agent Identity and Access establishes who every agent is and what it is permitted to hold. The Authorization and Runtime Control Architecture workshop then designs the control plane that evaluates each action against that identity. Many teams run them in sequence, identity first.

For organizations that want a baseline before committing to design and build, the Agent Governance Readiness Assessment scores current posture against the 12 Principles and recommends where to start.

Identity names the agent. Governance decides what it does.

The market is converging on the idea that identity is the control layer for AI agents. We agree that identity is the foundation, and we built this workshop to help you lay it well. A verifiable identity, scoped access, and a clean delegation model remove a large class of risk before an agent ever takes an action.

What identity alone cannot do is evaluate the action itself: whether this specific request, under this declared intent, through this delegation chain, is authorized right now against current policy. That evaluation is Agent Runtime Governance, and it is the layer this workshop connects your identity model into. The result is identity that is not just attested, but enforced.

Get started

The AI Agent Identity and Access workshop is available now. You can download the one-pager or book a workshop to discuss scope for your environment. If you are not sure where to start, a free 30-minute consultation will map your environment and align on the strongest next step.