KPMG UK’s Project Exponential used more than 100 AI agents to examine the accounts of an unnamed FTSE 250 company alongside a human audit team. Across the workflow, agents gathered information, analysed and documented data, drafted conclusions, and recommended follow-up work, coordinating tasks among themselves. KPMG’s head of audit said the agents sometimes investigated issues more deeply than the regular team and raised questions the auditors had not considered. The trial suggests AI could do more than speed up routine administration: it may help auditors scrutinise larger volumes of financial information and identify issues that warrant closer examination.
The exercise was not a statutory audit, and its results did not support formal approval of the company’s accounts. Human oversight remained necessary, including when an agent failed to match regulatory requirements correctly to company documents. KPMG plans to test the approach on up to 20 larger audits, but wider deployment depends partly on better-quality, structured business data. If the technology develops, it could reduce the extensive junior staff work that underpins traditional audits, while shifting human accountants toward complex judgements and exceptions. Potentially faster, cheaper audits could reshape the Big Four’s operating model, but lower fees are not guaranteed. Regulators and firms must also address independence, professional scepticism, data security, and inaccurate AI outputs; responsibility for the audit opinion remains with a human auditor.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.