Audit and AI: The constraint was never just technology

Over the past years, I have watched AI reshape customer service, legal and financial services. Different industries, different technologies and different management teams. Yet the pattern has been remarkably consistent, and it starts with the belief of every industry that it is different. The work is too specialised. Regulation is too strict. Clients will not accept it. AI cannot be trusted.

For a while, those arguments were usually right. The technology was not good enough, the risks were real, and the economics did not justify fundamental change.

Then the technology improved. Early adopters began to redesign how work gets done. Workflows changed, followed by staffing and pricing. Eventually, the economics of the industry changed. By the time the rest of the market realised it was no longer competing against better software but against a different operating model, catching up had become difficult.

‍

Why audit looked like the exception

Thirty years in professional services, more than half of them at KPMG, taught me how audit firms actually work, how quickly they can move and how partner economics shape almost every investment decision. Investing in AI companies since then taught me something else: industries tend to transform in remarkably similar ways.

Legal is probably the closest comparison to audit, as the two operate under very similar constraints. Both are risk-averse, evidence-driven, standards-oriented and highly regulated. Reputation is the core asset. Losing trust is existential. Professional judgment cannot be delegated, and accountability ultimately remains with the individual signing the opinion or giving the advice.

In the beginning, lawyers raised many of the same objections that auditors are raising today:

• "Can I trust the output?"
• "Who is responsible if it is wrong?"
• "Can I safely use client data?"
• "What is the cost-value ratio?"

These were legitimate questions, and they still are. But in legal, they gradually became implementation questions rather than arguments against using AI. Law firms learned how to validate AI-generated work, protect client information and keep professional accountability where it belongs.

Legal did not become less regulated. Lawyers did not become less accountable. The profession simply found better ways to operate and to deliver its work. Both the business of law and the practice of law have changed significantly through solutions and services from firms such as Intapp, Harvey and Legora. And the gap between the firms that moved and those that waited became structural.

For a long time, audit appeared to be different. There was also no sense of urgency, as partner income kept rising and clients remained loyal. The traditional leverage model continued to work remarkably well. AI, in its broadest sense, was not yet capable of fundamentally changing audit.

There was little reason to rethink an operating model that had delivered decades of profitable growth. You could therefore argue that the sceptics were right.

However, there came a shorter period during which the signals of change were visible but still easy to discount:

• Venture capital was investing heavily in vertical AI companies across professional services.
• Audit fees were being scrutinised.
• Skilled people were becoming more difficult and more expensive to recruit.
• Private equity began to focus on accounting as one of the last large professional services industries yet to undergo structural transformation.
• And in 2024, foundation models became genuinely useful for knowledge-intensive work.

‍

AI reaches the business and operating model

Most audit firms still treat AI as a technology decision. Software is evaluated, a pilot runs in one service line, licenses are purchased, and the results are presented to the board as evidence of progress.

Applied to an unchanged workflow, AI does make the existing process faster. However, the staffing pyramid stays intact, review works as it always has, and revenue still scales with hours worked. The gains are real, but they are marginal.

Professional services have been built on a single assumption: more revenue requires more people. AI is the first technology capable of breaking it. Which makes this an economic question, and explains why the firms creating real distance are redesigning every component in the audit process and in the management of an audit business.

The test is whether a firm can grow without adding people in proportion. Software layered onto an unchanged process never gets there.

‍

The capital structure is changing

Transforming audit takes more than good technology. It takes capital, time and a willingness to spend before the returns are visible, which is where the ownership model starts to matter.

Private equity in accounting has moved from an exception to a defining force. European transactions involving accountancy firms ran at nine times the level of four years earlier, and by early 2026 close to half of the top 30 US CPA firms had private equity investment or an alternative practice structure.

Rebuilding audit delivery takes time and consumes capital long before it returns any. An institutional investor can absorb that. A partnership rarely can, or more to the point rarely wants to, since it asks today's partners to accept lower distributions for benefits many of them will never personally see. Annual profit distribution and long-term technology investment sit badly together. That is how partnership economics work.

Outside capital does not guarantee a successful transformation. Institutional investors bring their own return expectations and time horizons. But they are usually quicker to make decisions. Which is why some PE-backed firms are already rebuilding their operating models while their competitors are still debating whether transformation is necessary.

‍

The pressure is building

Once one influential firm changes the economics of its operation and delivery, every other firm is eventually forced to respond.

I expect the first genuinely end-to-end, AI-supported audits to be delivered within the next year. These will not be autonomous audits. Professional judgment, review and sign-off will remain with auditors. But the environment around them, from planning and evidence collection through to testing, documentation and review, will look very different.

Every completed audit produces new experience. Every redesigned workflow makes the next implementation easier. Organisational learning compounds. Capability compounds. Which means that the gap between movers and laggards widens with every quarter.

And the pressure is coming from several directions:

• PE-backed firms have the capital to invest before returns are fully visible.
• New entrants do not carry the same legacy cost base, staffing structures or internal decision-making processes.
• Clients increasingly expect faster delivery, greater consistency and demonstrably high quality.

Once clients experience an AI-enabled audit that delivers clear benefits, they will inevitably begin to ask why the traditional model still requires the same number of people, the same amount of time and the same fee structure.

I believe that firms acting now still have a choice. They can determine how quickly they transform and which part of the market they want to occupy. In two years, many firms may no longer be choosing the pace of change. They will be reacting to it.

‍

The right partners matter

A transformation of this size is not something most audit firms can execute alone.

Audit firms have built outstanding methodologies, global delivery networks and systems of quality management. But they have rarely been successful software companies. The speed of AI development also makes it increasingly difficult for an audit firm to build and maintain every required capability internally.

Many firms have therefore partnered with Microsoft, OpenAI or Anthropic to gain access to frontier models. That is the right foundation. But a foundation model is not an audit system. It does not, by itself, embody a firm's audit methodology, its materiality decisions, its inspection history or its system of quality management. It does not automatically produce documentation that can withstand inspection, nor does it understand how an engagement should move from planning to evidence, review and sign-off.

General AI platforms will be part of every firm's technology stack. But they are only one part of the answer. What firms need is purpose-built audit technology that connects AI with methodology, evidence, documentation, review and professional judgment.

‍

Supporting the future of audit

This is the context in which I invested in Cortea.

Audit is the last major professional services sector where AI-powered transformation has yet to happen at scale. That is precisely what makes it worth backing now.

I have followed the enterprise software and start-up market for years. General AI platforms are enormously powerful, but they do not provide the specialised workflow, traceability and definition of quality that audit requires.

Cortea is building specifically for that environment, connecting AI with audit methodology, evidence and review, while keeping professional judgment with the auditor. Its ambition is not simply to make existing audit tasks faster, but to provide part of the quality and operating layer required for AI-supported audits.

It is encouraging to see a European AI company, backed by international capital and global audit expertise, helping to shape the future of audit.

‍