TSO365 Monthly Meeting - Nov 2026

5:45p - 6:00p ET Networking
6:00p - 7:00p ET Presentations

Governing, Securing and Managing Agents, Skills, etc. Across the Enterprise - Presented by Todd Furst
Customers have a wide range of AI solutions & capabilities that require an equally as capable governance model. Copilot Chat, Microsoft 365 Copilot, Agent Builder, Copilot Studio, Copilot Cowork, GitHub Copilot and Microsoft Foundry have broad capabilities to use/create agents, touch corporate data and have unique cost models. Add the third-party and local AI tools showing up on laptops, and you get agent sprawl/lifecycle/ownership, permissions concerns, sensitive data going into prompts, and costs challenges.
Starting with the solutions employees are using to create agents/skills/etc., we will work through the controls that govern each solution:

• Discover agents with Agent 365.
• Govern agents (just like users): give each agent a real identity with Microsoft Entra Agent ID. That brings Conditional Access, agent risk detection, access packages, and lifecycle workflows that retire agents nobody owns.
• Protect the data: use Microsoft Purview for AI data security posture (DSPM), DLP that blocks sensitive content in prompts, auditing of what agents do, and Insider Risk for risky AI use.
• Control the device: use Defender and Intune to find shadow AI and govern local AI on devices.
• Cost Management – Provides visibility and controls

All of this will be discussed in the context these scenarios: Personal Productivity, Departmental & Enterprise Wide solutions.

Microsoft Copilot Through a Risk Lens: Balancing Innovation and Trust - presented by Brooke Labenberg

### Summary

This session examines Microsoft 365 Copilot risks, including hallucinations and automation bias, and presents techniques for validating AI-generated outputs.

### Description

Microsoft 365 Copilot is transforming how we create content, analyze information, and make decisions. While much of the conversation focuses on productivity gains and new capabilities, successful adoption also requires understanding the risks associated with AI-generated content.

This session explores Microsoft Copilot from a risk and assurance perspective. Through practical demonstrations and real-world examples, we'll examine common challenges such as hallucinations, automation bias, incomplete context, and overreliance on AI-generated outputs. Attendees will learn how to evaluate the reliability of Copilot responses, apply appropriate validation techniques, and build confidence in using AI as part of their daily work.