KPMG has completed a groundbreaking AI-based audit of an unnamed UK FTSE 250 company, utilising more than 100 artificial intelligence agents. This experiment, dubbed Project Exponential, saw AI agents handle tasks like data collection, analysis, and documentation, even interacting autonomously. While the AI demonstrated the ability to investigate matters more deeply and raise pertinent questions, human oversight remained crucial for addressing issues and providing final judgement, reinforcing that AI is a powerful tool to support, not replace, human auditors.
KPMG has completed its first end-to-end AI-based audit of a UK-listed company, deploying more than 100 artificial intelligence agents to examine the financial statements of an unnamed FTSE 250 company. The experiment highlights how AI could reshape the auditing profession, while reinforcing the importance of human oversight and professional judgement.
KPMG Tests AI Agents Across the Entire Audit Process
Global accounting firm KPMG has taken a significant step towards integrating artificial int
Project Exponential is KPMG's internal name for its experimental AI-based audit, which deployed over 100 AI agents to examine the financial statements of a UK FTSE 250 company.
The AI agents were responsible for collecting financial and business information, analysing financial data, documenting findings, identifying potential issues, preparing preliminary conclusions, and suggesting next steps.
The AI agents operated with a degree of autonomy and could interact with each other, but human auditors were still critical for setting objectives, providing context, reviewing work, and challenging conclusions.
No, the experimental AI audit was not used by KPMG to formally approve the company's financial statements. The final audit opinion was based on the conventional audit process conducted by human professionals.
KPMG plans to extend its AI-enabled audit approach to up to 20 larger audits and is exploring the possibility of offering AI-based audit services to clients in the future.
The experiment showed that AI agents can struggle with matching regulatory requirements to documents and that human intervention is necessary for complex judgement and accountability. Additionally, data quality and scale remain practical challenges for wider AI adoption.
Finance news reporter covering taxation, GST, income tax, business compliance, and economy updates. I simplify complex financial topics into easy-to-understand articles for professionals, taxpayers, and business owners on leading finance and tax platforms.