Agentic AI changes the object of governance. Traditional AI governance was built around models, prompts, outputs, approvals, and periodic reviews. Those controls still matter, but they are no longer enough when AI agents can retrieve data, invoke tools, update records, trigger workflows, delegate work, and create consequences before a human reviewer sees the full action path.
Runtime AI Governance is a practitioner’s playbook for moving from static governance artifacts to a runtime governance architecture. Through the fictional but realistic NovaCred case study, the book explains how enterprises can govern agentic AI across five dimensions: identity, intent, enforcement, oversight, and accountability. It shows how agent identity becomes the new perimeter, why permission is not the same as intent, how runtime policy gates and Governance Decision Records create evidence, and how human supervision must move up the stack as autonomous systems operate at machine speed.
The book bridges established AI governance frameworks such as ISO/IEC 42001, NIST AI RMF, and the EU AI Act with newer agentic governance lenses including Singapore’s Model AI Governance Framework for Agentic AI, OWASP Agentic Applications Top 10, and NIST’s AI Agent Standards Initiative.
It is written for AI leaders, enterprise architects, platform engineers, security teams, GRC professionals, compliance leaders, auditors, and executives who need to govern agentic AI systems without slowing innovation to a halt.
