Session
Wild ideas, safe agents
Building an agent in Copilot Studio is getting easier. Making sure it only accesses the information it should, performs the right actions and behaves as expected takes a little more effort.
Especially when someone decides the agent should also update records, send messages and connect to another five systems.
What could possibly go wrong?
In this session, we'll build and review a Copilot Studio agent from a security and governance perspective.
We'll look at authentication, knowledge sources, permissions, actions and the controls organisations should consider before publishing agents to users.
We'll also explore where Microsoft Entra Agent ID and Agent 365 fit into agent management, including identity, access and lifecycle considerations.
Along the way, we'll test what happens when an agent encounters information or actions it shouldn't have access to.
The goal is to build useful agents without giving them more access or responsibility than they need.
What you'll learn
1. How to plan permissions and knowledge access for Copilot Studio agents.
2. How authentication and authorisation affect agent behaviour.
3. How to review actions and connected systems before deployment.
4. How Agent 365 and Entra Agent ID fit into enterprise agent governance.
5. How to test, monitor and maintain agents after publishing.
Gunnar Reyr Sigurðsson
Senior M365 Consultant - Microsoft AI and Copilot Lead | SharePoint
Reykjavík, Iceland
Links
Please note that Sessionize is not responsible for the accuracy or validity of the data provided by speakers. If you suspect this profile to be fake or spam, please let us know.
Jump to top