AI agents are appearing across organisations faster than most IT and security teams can inventory, govern, or secure them. Every agent introduces a new identity, new permissions, new data connections, and potentially a new attack path.

In this demo-driven session, we move beyond the Microsoft Agent 365 overview and follow realistic agents from discovery to production. You will see how Microsoft Agent 365 provides a unified control plane for agents built with Microsoft, partner, and custom technologies. We will discover shadow agents, explore their relationships with users, tools, data, and other agents, assign ownership, apply lifecycle and access controls, and investigate risks such as excessive permissions, sensitive-data access, and prompt injection.

We will also examine how Microsoft Entra, Microsoft Purview, and Microsoft Defender work together to protect agents at identity, data, and runtime levels. Finally, we will connect a custom agent using the Agent 365 SDK and make its identity and telemetry visible to administrators.

You will leave with a practical model for taking agents from experimentation to governed enterprise adoption, without bringing innovation to a halt.