Graham Hosking, Senior AI Solutions Engineer at Microsoft, breaks down Agent 365, Microsoft's approach to observing, governing, and securing AI agents across multi-platform environments.

• How Agent 365 builds on existing Entra, Defender, and Purview capabilities to provide full visibility over AI agents

• Live demo of a fully autonomous LangChain agent automatically flagged for risky behaviour by Defender

• How third-party agents from Amazon, Google, Databricks, and Salesforce can be brought into a central registry

• The Agent 365 SDK and how partners like Adobe, Atlassian, and Lovable are already baking in observability

• Shadow IT monitoring for local AI agents via Defender for Endpoint

• MCP tools and plugin management for controlling end-user access

• Licensing breakdown: end-user based, one seat lights everything up, and the E5 to E7 step-up path

• Bonus demo of Morgan, a fully autonomous digital CFO with voice and avatar interaction built on LangChain

Thanks to Crestron, this episode's sponsor, for their continued support of Empowering.Cloud