What is agentic AI in audit?
Agentic AI can carry out a sequence of connected tasks towards a defined objective, rather than responding to one prompt at a time. In audit, that could mean gathering evidence, identifying relevant information, applying a methodology, preparing a control assessment and routing exceptions to an auditor for review.
That makes agentic AI potentially more valuable than a standalone chatbot, but it also increases the need for control. A system that performs several steps can compound an error if its scope, evidence or decision rules are weak. Firms therefore need clear boundaries: what the agent is allowed to do, what evidence it can use, when it must stop, what the auditor must review and how the full sequence is documented.
The right model is controlled delegation. The agent completes repetitive, evidence-heavy work inside a configured workflow. The auditor reviews the evidence and rationale, challenges anything unexpected and remains accountable for the conclusion.
For audit firms, the commercial opportunity is significant because the value comes from completing work, not merely suggesting how to do it. The governance standard must rise with the capability.
References: FRC generative and agentic AI guidance · IAASB technology resources