KPMG's UK arm revealed this week that AI agents deployed during a landmark audit trial sometimes investigated financial issues more deeply — and asked better questions — than its own human audit team, according to reporting by the Financial Times and European Business Magazine. The experiment, codenamed Project Exponential, used more than 100 AI agents to examine the financial statements of a major UK-listed company.
The trial marks a significant step for artificial intelligence in professional services, moving from AI as an assistant for isolated tasks toward autonomous agents working across an entire audit workflow. It also raises an uncomfortable question for the accountancy profession: if machines can increasingly perform the analytical work of highly trained professionals, how quickly will the economics of auditing change? Track this story and the rest of the AI business landscape on AI Buzz Wire.
How Project Exponential Worked
The AI agents examined the accounts of an unnamed FTSE 250 company alongside a conventional audit team, taking on tasks across the whole audit process rather than simply automating administrative chores. According to the FT report, the agents collected information, analyzed and documented data, developed draft conclusions and observations, and suggested next steps. They also coordinated with one another, effectively dividing responsibilities across the investigation.
KPMG UK's head of audit, Cath Burnet, said the technology had in several instances gone considerably deeper into issues than the regular audit team, identifying areas requiring further investigation and asking questions that human auditors had not raised. The technology was developed over roughly four months.
Where the Machines Still Need Supervision
KPMG was careful to draw a line between the experiment and a formally approved statutory audit. Although an audit partner signed off on the exercise, its findings were not used as the basis for approving the company's accounts. Human oversight remained essential throughout: during the trial, an auditor had to intervene when one AI agent failed to correctly match regulatory requirements against the company's documentation — an episode that illustrates both the capabilities and the limits of current systems when navigating complex regulatory frameworks.
The firm plans to expand the initiative to as many as 20 larger audits, while acknowledging the technology is not yet mature enough to deploy across the thousands of engagements a Big Four firm handles annually. Data quality remains a significant obstacle, with some businesses lacking the clean, structured information AI agents need to operate effectively.
The Trial Fits an Uneven Enterprise Picture
The audit experiment lands at a complicated moment for enterprise AI agents. A KPMG survey published in August found that nearly half of large organizations — around 49% — had narrowed, delayed or paused their AI agent deployments because running costs exceeded the value the systems delivered, a reality check on a year of aggressive corporate enthusiasm for autonomous agents. Against that backdrop, Project Exponential is a notable counterexample: a professional-services firm putting agents to work on high-value analytical labor rather than back-office automation, and finding the results competitive with its own trained staff.
The findings also suggest AI's potential in professional services may extend well beyond productivity gains, according to European Business Magazine's report. The technology could change how complex financial information is examined, interpreted and challenged — work that sits at the core of what audit firms sell. Australia's Financial Review covered the same trial under a headline noting that AI asked "better questions" than KPMG accountants in what it described as a first-of-its-kind audit run.
A Challenge to the Big Four Business Model
The implications reach beyond one firm. Traditional auditing relies on large teams of junior accountants performing detailed, repetitive work under the supervision of experienced managers and partners. AI agents offer a different operating model — automating much of the data collection, analysis and documentation while senior professionals focus on complex judgments and exceptions.
For KPMG, Deloitte, PwC and EY, the attraction is obvious: faster audits, potentially lower operating costs and the ability to analyze far larger volumes of financial information. Whether those savings reach clients is an open question. KPMG has acknowledged the potential for significant time savings but has not confirmed whether AI-driven efficiencies will translate into lower audit fees, European Business Magazine reported.
Regulators are watching closely. The UK's Financial Reporting Council has warned that AI use must not undermine auditor independence, integrity or professional skepticism, and concerns persist around data security, over-reliance on automated conclusions and the possibility of AI generating inaccurate or fabricated information. Under existing rules, responsibility for the audit opinion remains with the human auditor regardless of the technology used.
The economics nonetheless point in one direction. Audit firms depend on armies of junior staff billable at scale, and the FTSE 250 trial hints at a future where much of that work is machine-performed — with humans retained for judgment, oversight and accountability. KPMG is moving deliberately: expanding to a few dozen audits first, collecting evidence on where agents help and where they fail, before any broader rollout. Rival Big Four firms are running their own agent pilots, and the UK regulator's caution signals that adoption will be governed as much by audit standards as by technology. For continuing coverage of how AI is reshaping knowledge work, follow the latest AI developments on AI Buzz Wire.
What It Means for the Profession
The trial does not establish that AI can replace accountants or independently deliver a reliable statutory audit. It does demonstrate that AI agents are beginning to challenge professionals in territory long considered dependent on human expertise — and that the question is no longer whether AI becomes part of the audit process, but how much of it the machines will eventually perform. For continuous coverage of how AI is reshaping white-collar work, follow the latest AI developments on AI Buzz Wire.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →