From AI Agent to Enterprise Production - Part 2
What happens after the prototype: security, governance, lifecycle, monitoring, and the decisions that decide whether an agent ships
- Live Thursday 27 August 2026, 16:00-18:00 CEST
- Two hours, with extended time for live technical questions
- Builds on Part 1 but works as a standalone deep dive
- Security, governance, lifecycle, testing, scale, monitoring
- With Mirko Peters, hosted by M365 Show
About this session
You have built the agent. Now comes the difficult part: making it ready for the enterprise. In Part 2 of this live deep dive with Mirko Peters on M365 Show, we pick up where the prototype ends and take agents built with Copilot Studio and Power Platform toward secure, governed, scalable and maintainable production solutions. We go deeper into enterprise architecture, security, permissions and identity, organizational controls, agent lifecycle management, managing integrations and enterprise data, testing and reliability, scaling, and the monitoring that tells you whether any of it is working. The through-line is the set of practical decisions that determine whether an AI agent stays a demo or becomes part of a real business process, drawn from real implementations rather than slideware. The session builds on Part 1 on 17 August, but is designed to stand on its own.
Resources
Join the session
- M365 Show→The show's home. I joined as host and commentator in August 2026.
- Register on LinkedIn - Part 2, 27 August 2026→The live event page. Slides and the recording will be added here afterwards.
- Part 1: Building Enterprise AI Agents That Actually Work→The foundations session on 17 August, on architecture, data, and actions.
Background reading
- Key concepts - Copilot Studio security and governance→The controls that exist at tenant, environment, and agent level.
- Establish an application lifecycle management strategy→Dev, test, production environments and the ALM golden rules for agents.
- Manage your Copilot Studio projects - governance guidance→The full phased guidance from capturing requirements to monitoring.
- Configure data policies for agents→DLP for knowledge sources, connectors, channels, and triggers.
From the speaker
Elliot Margot
Microsoft MVP - M365 Copilot & Copilot Studio