THE CRUNCH
KPMG has run a full AI-powered parallel audit of a FTSE 250 company for the first time, deploying more than 100 AI agents through every phase of the annual audit alongside the conventional human one. The official audit opinion was still based on the regular audit.
The experiment, internally called Project Exponential, had agents independently collect, analyse and document data and draft audit conclusions. Cath Burnet, KPMG's UK head of audit, said that in several cases the AI went "so much deeper" than the regular team and sometimes asked "better questions" than the firm's own auditors.
It was not a clean sweep. In one area a human had to intervene because an agent could not correctly match a set of regulations against company documents. KPMG says human involvement stayed "critical": staff set the agents' goals, supplied context, reviewed their work and challenged conclusions before the opinion was finalised.
The firm plans to expand the project to as many as 20 larger audits, but Burnet said the technology is not yet available at the scale needed for thousands of audits, and few companies currently have clean enough data for agents to work with reliably. KPMG says the system records agent interactions with documents down to individual characters and timestamps, offering "full visibility and full traceability".


