Most organisations start with the CoE Starter Kit and finally get visibility: what exists, who built it, and where the risks might be. Then adoption kicks in… and governance becomes reactive. The default environment turns into a junk drawer, ownership gets blurry, and the “process” is basically chasing people after something breaks.

This session is the upgrade path. It’s not a feature comparison and it’s not “CoE vs Managed Platform” as a debate. It’s a practical operating model that scales—especially now that agents make it easier than ever for solutions to spread fast.

I’ll walk through a zoned governance approach (Green / Yellow / Red) and show how to map it to the platform: use environment routing to give makers a personal “OneDrive for innovation,” apply environment groups and rules to enforce guardrails consistently (sharing limits, AI controls, connector boundaries), and use pipelines/approvals so good ideas can move from dev to test to production without chaos. We’ll also cover what changes when the asset is an agent: blast radius, monitoring/alerts, test/evals, and why “it worked once” isn’t a release strategy.

You’ll leave with a clear division of responsibilities (what CoE should keep doing vs what the Managed Platform should enforce), plus a rollout plan you can start small with and scale across the tenant.