Microsoft 365 is no longer just a platform for users and applications, it's becoming a platform for AI agents that can read data, generate content and take actions across Teams, SharePoint and other collaboration services.
These agents are often deployed as easily as traditional SaaS apps, yet they may operate with broad permissions and increasing autonomy. In practice, this means organizations are introducing new identities into their environment – without always realizing it.
This session examines what it really means to govern AI agents in Microsoft 365. We’ll explore how agents differ from traditional apps, what happens when users consent to autonomous capabilities, and how identity, permissions, lifecycle management and centralized publishing must evolve in an agent‑first world. The focus is practical and experience‑driven: what questions admins and architects should be asking today, where existing governance models still work, and where they clearly fall short when enabling Agent 365 in enterprise environments.