As AI agents enter the workplace, identity becomes more than “who signed in.”
Agents act on behalf of users, access data across systems, and trigger workflows at scale — which raises a new question: how do we define trust, boundaries, and accountability when work is delegated to agents?

This session introduces a practical framework for designing agent identity, security boundaries, and governance in Microsoft 365 environments. We’ll explore common failure modes (over-permissioned agents, unclear ownership, untraceable actions), and how to build guardrails that enable adoption rather than blocking innovation.