Below is a recap of questions the panelists answered during today's live AMA, along with associated timestamps: 

Question – Are agentic capacities opt-in, preview-only, or turned on by default? – answered at 1:49.

Question – Can we enable AI features for some users or devices but not others? – answered at 3:32.

Question – Do agents run as the user, as a service, or as a separate identity? – answered at 5:13.

Question – With the recent news on how Microsoft is pulling back certain AI integrations (Notepad, etc.), what changes will there be with managing AI on Windows? – answered at 7:45.

Question – What’s the recommended approach to managing agentic features across a mixed fleet (managed, unmanaged, and different device types)?  – answered at 10:58.

Question – How should admins think about permissions for agent connectors/workspaces—what’s the “least privilege” strategy? – answered at 13:59.

Question – Are AI and agent actions logged separately from user actions? – answered at 17:23.

Question – Do you have any updates on how MCP on Windows is evolving? (MCP = Model Context Protocol) – answered at 19:27.

Question – What’s the best way to monitor or block risky AI‑initiated behaviors? – answered at 21:13.

Question – How do you roll this out safely—pilot design, rings, and criteria for broad enablement? – answered at 24:17.