Master Agent Governance with Agent 365 | BRK345
Hyuncheol Jo, Sneha Kohli, and Mark Wahl explore the operational risks of unmanaged AI agents and demonstrate governance workflows in Microsoft Agent 365. The session combines product demos with LG U’s experience establishing enterprise agent governance.
Governing a growing agent ecosystem
As agents multiply across an organization and operate across platforms, administrators need to answer several basic questions:
- Which agents exist?
- Who owns each agent?
- What can each agent access?
- Is each agent operating as intended?
The session focuses on how Agent 365 supports centralized oversight, accountability, policy and tools management, and lifecycle governance. It addresses governance responsibilities for IT, AI, and security administrators.
Enterprise governance at LG U
LG U shares its enterprise AI agent governance journey, including a framework built around agent registration, identity, and ownership rules. These foundations help establish which agents are in use and who is accountable for them.
The discussion also presents a future vision for a unified agent control plane and automated governance, distinguishing that direction from the governance framework covered in the enterprise example.
Lifecycle governance and tool assessments
The demonstrations cover two areas of Agent 365:
- Action Center: Agent lifecycle management and governance automation.
- Tool Registry: Agent tool governance, including MCP server assessments and GitHub integration for agent tools.
These topics connect oversight of the agents themselves with assessment of the tools they use.
Agent identity, access, and data protection
The security portion examines agent identity management and least-privilege access. It then turns to operational controls, covering access controls, lifecycle policies, and data protection as part of day-to-day agent governance.
The central concern is not just identifying agents, but governing what they can access and how they are managed throughout their lifecycle.
Session chapters
- 0:00: Introduction to agent governance challenges with A365
- 1:17: Operational risks from unmanaged and cross-platform AI agents
- 2:10: Governance focus areas for IT, AI, and security admins
- 2:42: LG U’s enterprise AI agent governance journey
- 5:39: LG U’s governance framework: registration, identity, and ownership rules
- 8:36: Future vision for a unified agent control plane and automated governance
- 9:45: A365 Action Center for agent lifecycle and governance automation
- 14:37: Tool Registry, MCP server assessments, and GitHub integration for agent tools
- 19:49: Security risks, identity management, and least privilege for AI agents
- 27:13: Operationalizing agent security with access controls, lifecycle policies, and data protection
Resources
This Microsoft Ignite 2026 breakout session is BRK345, presented in English (US), with an intermediate (200) level and a Security focus.