As AI agents proliferate across enterprise business units, organizations risk repeating the fragmentation that plagued decentralized data analytics a decade ago. A three-layer Agent Governance Operating Model is proposed to address this: domain autonomy (business units own agent workflows), centralized governance authority (cross-functional standards for data classification, tool certification, and risk tiers), and platform-level enforcement (RBAC, dynamic masking, immutable audit logs embedded in infrastructure). The core argument is that governance embedded into the platform accelerates agent adoption rather than slowing it, by eliminating ad-hoc access decisions and giving domain teams a trusted foundation to build on. The model shifts governance from information stewardship to action stewardship, covering not just data access but what autonomous systems can do.
Table of contents
The new governance questionRecognizing the opportunity windowThe Agent Governance Operating ModelGet Sahil Kotwal’s stories in your inboxWhy governance is a scaling enablerFrom information stewardship to action stewardshipThe advantage of moving earlySort: