Overview
AI agents are already in production, accessing your systems and data with little human oversight. Built by developers, spun up by business teams, and embedded in enterprise platforms, agents are everywhere, and most organizations operate without governance models designed for autonomous, tool-connected systems.
In AI Agent Governance, enterprise architect Kesha Williams gives AI engineers and platform engineers practical patterns to close that gap. Through a realistic scenario, you'll follow a Model Context Protocol (MCP)-connected agent from its first deployment through its evolution into a multi-agent system coordinated through Agent-to-Agent (A2A). Governance challenges emerge at every stage: behavioral drift, unauthorized tool access, unsafe delegation, and cascading failures. You'll design governance architectures, enforce boundaries, protect sensitive systems and data, and recover from failures, with the operational patterns to run agents safely in production. A GitHub repository and the Agent Governance Console sandbox let you build alongside every chapter.
- Design governance architectures for agents in production
- Monitor behavior across single-agent and multi-agent systems
- Enforce execution constraints and access control
- Secure MCP tool connections, A2A communication, and sensitive systems
- Interrupt unsafe behavior and recover safely from failures
Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.
Read now
Unlock full access