AI is no longer just answering questions. It is starting to act.Autonomous AI agents can now make decisions, access systems, initiate transactions, communicate with customers, analyze data, create records, and execute business processes with limited human intervention.That changes the risk landscape completely.Agentic AI Risk & Controls is a practical guide for professionals who need to understand, govern, control, and audit autonomous AI agents before those agents become embedded across the organization.Written from a risk, governance, internal control, and audit perspective, this book explains Agentic AI without requiring a technical or data-science background. It focuses on the questions that matter most to management, risk professionals, internal auditors, compliance teams, CIOs, CISOs, finance leaders, and control owners.At the center of the book is the practical 6A Agentic AI Control Framework:Agent – Authority – Access – Action – Audit Trail – AccountabilityUsing this framework, you will learn how to determine what an AI agent is allowed to do, what systems and data it may access, which decisions it can make independently, where human approval is required, how its actions should be monitored, and who remains accountable when something goes wrong.Inside the book, you will learn how to:• Identify and classify AI agents across the organization• Assess autonomy, authority, access, and execution risks• Design effective Human-in-the-Loop and Human-on-the-Loop controls• Apply segregation of duties to AI agents• Control agent identities, permissions, credentials, and system access• Establish transaction limits and approval thresholds• Prevent unauthorized or irreversible AI actions• Monitor agent behavior and detect anomalies• Build reliable audit trails for autonomous decisions• Design kill switches and emergency controls• Audit Agentic AI using practical, risk-based procedures• Evaluate agents used in Finance, Procurement, Sales, HR, IT, and Customer Service• Define ownership, governance, and board-level oversightThe book also includes practical tools that can be adapted immediately, including:AI Agent InventoryAgent Risk Classification MatrixAgent Authority MatrixAgent PassportAgentic AI Control LibraryAudit ProgramEvidence Request ListRed Flag ChecklistRisk Assessment TemplatesBoard and Audit Committee QuestionsA chatbot that gives a wrong answer may create inconvenience.An autonomous AI agent that makes a wrong decision may create a payment, change a price, expose confidential data, approve a supplier, alter a system, or execute a transaction.The question is no longer simply:“Can we trust the AI?”The more important question is:“What are we allowing the AI to do—and what controls exist before it does it?”Agentic AI Risk & Controls provides a practical roadmap for answering that question. Read more
