9781972752258_

Agentic AI is no longer a future concept—it is already inside your organization, taking actions, triggering workflows, calling APIs, and making decisions at machine speed. And while these systems unlock extraordinary productivity, they also introduce a new category of risk: AI that acts without clear boundaries, without human review, and without the governance scaffolding needed to keep organizations safe. This book begins where most AI guides stop—at the moment when AI becomes an actor, not just a tool. It shows managers how to govern autonomy, constrain authority, and build identity and access controls that prevent small pilot mistakes from becoming enterprise-level incidents.

Inside this book, readers will learn how to:
n• Design governance frameworks that define agent intent, authority, and escalation thresholds
n• Implement OAuth, scopes, and token strategies that enforce least-privilege access for AI agents
n• Build identity and access architectures that prevent unauthorized actions and external exposure
n• Conduct AI-specific threat modeling and risk assessments for autonomous systems
n• Establish pre-production controls, safety reviews, and release gates that hold under pressure
n• Define human oversight roles, accountability structures, and decision rights for agentic AI
n• Monitor agent behavior in production with telemetry, anomaly detection, and kill-switch patterns
n• Build incident-response playbooks tailored to AI-driven failures and misaligned actions
n• Scale governance across teams, business units, and enterprise portfolios without slowing innovation
n• Create organizational practices that make responsible AI the default, not an afterthought

Agentic AI changes the governance equation because it changes the nature of action. Traditional AI produces outputs; agentic AI takes steps. It writes files, sends messages, triggers workflows, and adapts its behavior based on what it learns. That shift—from passive to active—means managers must govern not just model quality but model authority. This book shows how to build that authority layer: intent statements, prohibited actions, review thresholds, and access boundaries that hold even when the agent behaves in unexpected ways.

Through real-world scenarios, you’ll see how small governance gaps become major incidents: agents emailing clients without approval, accessing systems beyond their scope, or optimizing for metrics that misalign with business intent. You’ll learn how to prevent these failures with a five-layer governance stack—policy, controls, monitoring,

Serious Managers Guide to AI Identity at Scale

Comentarios