A one-page template for governing AI agents: ownership, permissions, approval and review.
The short answer
An AI agent governance policy should define for each agent: its owner, purpose, connected systems, permissions, which outputs need approval, how it is monitored and when it is reviewed.
Key takeaways
A one-page register per agent beats a 40-page policy nobody reads.
Owner, purpose, systems, permission scope, approval rules, metrics, review date.
Deciding who owns which agents is the first layer of any agent organisation—something Switch helps customers define before building.
Typically the business owner plus IT/security sign-off.
Quarterly is a sensible default.
How this guide was made
Written and reviewed by our editors against vendor documentation, public regulation texts and practitioner interviews. We update guides when products or rules change. Tools mentioned may have a commercial relationship with this site; this never decides what we recommend.
Keep reading