KPMG’s “Project Exponential” put AI agents alongside a real audit team and found the bots dug deeper on individual questions

KPMG deployed more than 100 AI agents to work alongside a human audit team during the review of a UK-listed company. Cath…

By Vane November 18, 2026 1 min read
KPMG’s “Project Exponential” put AI agents alongside a real audit team and found the bots dug deeper on individual questions

KPMG deployed more than 100 AI agents to work alongside a human audit team during the review of a UK-listed company. Cath Burnet, the UK head of audit, stated that the automated systems asked better questions than the staff in specific instances. The firms found the bots dug deeper on individual queries by analysing far more data than traditional sampling methods allow. The Financial Reporting Council views this approach as a risk because the sheer volume of data analysis could surface findings that serve as a sales pitch for lucrative consulting work. This potential conflict could undermine auditor independence if the firm uses audit insights to generate new revenue streams for its own consultancy division.

The core issue is not whether the technology works, but whether the regulatory framework can prevent the audit function from being compromised by commercial interests. The regulator fears that deep data access creates an incentive to sell additional services based on the very findings discovered during the review.

  • The experiment involved over 100 AI agents working in parallel with human staff.
  • The Financial Reporting Council has flagged the independence risk explicitly.
  • Traditional sampling limits the data scope to prevent this specific conflict.
Scroll to Top