Most enterprises already have AI agents running in their environment, and most security teams cannot say how many. As Claude Code, Codex, Cursor and a wave of vendor-shipped agents arrive inside everyday software, the risk shifts from user-driven to agent-driven: agents can reach enterprise systems and data with far more privilege than anyone intended and exfiltrate proprietary IP to public models without anyone watching.

Detection and monitoring alone won't close the gap. What's needed is declared intent enforcement: defining what an agent is permitted to do up front and enforcing it at a layer the agent cannot bypass.

Zentera's Chief Evangelist Nathanael Iversen led a practical session on how to gain visibility into the agents already running, set the controls that keep them on task and govern agentic AI without slowing the innovation your business is counting on.

What attendees learned:

• How to inventory the AI agents already running across your environment
• What declared intent enforcement is, and where controls must sit to work
• How to keep agents on task with project-level boundaries (enclaves)
• Where AI meets IT and OT systems and how to govern that access