AI Is Taking Over Audit Functions. Accounting Needs to Get Ready

KPMG’s rollout of advanced artificial intelligence agents is the start of something structurally significant: a “K-shaped” remodeling of professional services, in which the top accelerates, the bottom collapses, and the distance between them becomes the competitive question.

Last week’s news that KPMG is cutting roughly 10% of its US audit partners is the K-shape made literal. The average partner whose economics rested on leverage rather than origination is being removed. The pyramid is being compressed from both ends at once. Graduate hiring is being switched off at the bottom, and the average partner is being asked to leave at the top.

Thomas Mackenzie, KPMG’s audit chief technology officer, recently said that within two to three years there will be “next to no human beings” performing routine audit testing at KPMG. This is the most honest thing any Big Four leader has said about the direction of the profession this year. Vouching, transaction testing, and other tasks that for decades defined early-career life in public accounting are being absorbed by AI agents.

But what does this do to the pipeline that produced every audit partner currently signing opinions? Audit has always developed judgment through repetition—exposure to hundreds of small misstatements before forming a view on a material one. Remove this type of apprenticeship and you have removed the mechanism by which the profession reproduces itself.

Mackenzie’s own framing was telling. He said he will no longer hire a college graduate to create workpapers. That is a reasonable shift in job design and a complete rewrite of the conditions under which professional judgment has historically been developed.

K-shaped Remodeling

It’s useful to think about what replaces the pyramid as being K-shaped rather than linear. The two arms of the K diverge from a single point, which is the old partnership model of uniform growth and predictable leverage.

The upper arm accelerates. Exceptional partners do more because clients buy them rather than the firm behind them, and AI amplifies their output rather than substituting for it. The lower arm collapses due to commoditized delivery work, time-and-materials billing, standardized compliance output, and the average partner whose economics rested on leverage rather than origination. AI actively widens the gap between the two because the tools that amplify the top are the same tools that substitute for the bottom

This plays out at a firmwide level, and we are seeing it in real time. Looking at the first-quarter 2026 data, firms that have already made the shift to their operating models are clearly outgrowing those that haven’t.

This is a pricing reset. The bottom half of the market becomes cheaper, faster, and harder to differentiate. The top becomes more expensive, more concentrated, and more dependent on individuals rather than institutions.

The most useful frame I have seen for role-level exposure within this comes from global consulting firm BCG’s labor classification, which separates roles into three categories:

• Amplified, where AI materially enhances individual output but the role remains human-led
• Divergent, where AI fundamentally changes what the role is and what it produces
• Substituted, where AI replaces the role entirely

In audit, the partner is amplified, the senior manager is divergent, and the associate is substituted. Each category carries a different economic, organizational, and talent implication. The firms that are planning workforces at firm-average level rather than role-by-role are going to find themselves with the wrong people in the wrong places by 2028.

What Survives

What will survive at the top of the K?

• Deep industry experience, not rotational coverage, but 20 years in a sector
• Trust, earned through having been right in difficult situations in front of boards that remember
• Judgment, the ability to form a view when the data is incomplete and the politics are charged
• Personal brand, a reputation that exists independently of the firm’s logo
• Charisma, the presence to walk into a client’s offices and command the room

These have always been the differentiators for the top decile. In the emerging model, they are the entire product.

A realistic read of the firm’s own position on the K—and what we see repeatedly when running AI risk diligence for private equity sponsors—is what’s missing from most firm narratives. The most revealing data point is almost always the delta between external and internal sentiment.

Partners tell the outside market the firm is leading on AI. The workforce inside tells us something very different about utilization, about leadership literacy, about whether the tools being demonstrated to clients are actually being used in delivery. A firm where 95% of mid-level staff use AI weekly, but where only 41% percent of leadership do, is a governance gap waiting to become an execution problem.

One Competitive Question

The KPMG news signals a deeper shift: that the old assumption about professional services growing as a tide that lifts every firm equally is over. The firms pulling ahead share a recognizable profile.

They are private equity-backed or well-capitalized; their leadership that has translated conviction into decisions rather than working groups. They are already rebuilding workflows around AI, creating proprietary intellectual property on top of foundational models and restructuring compensation to attract the partners the emerging model requires.

The firms standing still are largely legacy partnerships in which the partners closest to retirement hold the most influence and the least incentive to act. The technology gap and the talent gap are the same gap.

The private equity community has moved past the question of whether AI disrupts professional services. It is underwriting assets on how seriously and how early management acted. If a firm is still in the monitoring phase, it has already answered the question—just not in the way leadership thinks it has.

The K-shape is what connects the dots. The bottom arm determines cost structure and pricing power today. The top arm determines who still has a product worth selling in 2035. The firms only having the productivity conversation will discover both in sequence.

The firms already having all four—productivity, regulatory exposure, talent succession, and competitive positioning—will look by the end of this cycle like a different kind of business. What sits at the top of that business is narrower, rarer, and more expensive than anything the profession has priced for before.

This article does not necessarily reflect the opinion of Bloomberg Industry Group, Inc., the publisher of Bloomberg Law, Bloomberg Tax, and Bloomberg Government, or its owners.

Author Information

James O’Dowd is founder and CEO of Patrick Morgan.

Write for Us: Author Guidelines