Monday morning triage: from Shadow AI to Governed AI in minutes | THR854
Brendan Powers and Srikumar Nair walk through a Monday-morning triage routine that starts with known AI agents and expands to Shadow AI. The Microsoft Ignite 2026 session combines Microsoft Agent 365 demonstrations with Zenity’s perspective on investigating agent behavior and enforcing controls at runtime.
A repeatable AI agent triage routine
The session’s workflow centers on assessing risk, investigating available context, and prioritizing action:
- Start with known agents: Review agent risk and security context before expanding the investigation.
- Discover Shadow AI: Use Microsoft Agent 365’s registry and discovery capabilities to investigate AI usage outside the known-agent inventory.
- Investigate context: Examine available device and user information to support risk assessment.
- Evaluate controls: Explore blocking and isolation options for AI agents.
- Consider runtime behavior: Examine intent-based governance and runtime enforcement through Zenity’s demonstration.
- Move toward governance: Bring observability and controls together into a repeatable approach to managing agent risk.
The discussion also addresses Shadow AI risks involving local tools and remote MCP servers, agent behavior defense and runtime protection, and a customer perspective from Lumen. It distinguishes the available controls from the broader roadmap from discovery to governance.
Session chapters
| Time | Topic |
|---|---|
| 00:00 | Introduction to Monday Morning Triage and AI Agent Security |
| 01:06 | Known AI Agents, Security Context, and Governance Foundations |
| 02:02 | Agent 365 Registry and Shadow AI Discovery Overview |
| 03:15 | Investigating Shadow AI Instances and Risk Assessment |
| 05:18 | Security Controls for AI Agents Including Blocking and Isolation |
| 08:01 | Zenity Demo on Intent-Based AI Agent Governance and Runtime Enforcement |
| 13:02 | Shadow AI Risks from Local Tools and Remote MCP Servers |
| 15:52 | Checkpoint Agent Behavior Defense and Runtime Protection |
| 20:01 | Unified Agent 365 Observability and Governance Controls |
| 22:07 | Customer Perspective from Lumen and Final Shadow AI Governance Playbook |
Session details and resource
- Event: Microsoft Ignite 2026
- Session: THR854 | Copilot
- Format: Theater
- Level: 200 — Intermediate
- Resource: Copilot next steps