The fastest way to bring AI to accounting? Stop selling it

Accounting has discovered one of the strangest growth problems in business: more customers than it can serve.  

A global survey of roughly 500 accountancy leaders found that 73% were turning away potential clients because they lacked staff and said the talent shortage was having a severe impact on their firms. Typically, technology discussions start with the demand problem, but in this sector, it starts with distribution.

This makes accounting an obvious candidate for artificial intelligence. Much of the work limiting firms' capacity is repetitive and operational. If technology can shoulder more of that work, the same team can serve more clients.

But the buck doesn't stop there. Even solving technical problems doesn't automatically solve the industry problem. In fact, by building better AI software and selling it to accounting firms, one runs into two structural obstacles. First, AI is making software faster and cheaper to build, which makes successful products easier to imitate. Second, accounting firms are unusually reluctant to replace the systems already embedded in their businesses.

So how can better software be introduced into the industry in the face of this? In this article, I make the case for development through acquisition.

The capacity problem

It is reasonable to wonder why firms do not simply hire more accountants. At a time when anxiety about AI-driven job displacement is widespread, an industry with persistent open positions might seem well placed to attract workers. Yet accounting has struggled for years with a weak talent pipeline and an aging workforce. 

There are some signs of improvement. Accounting enrollment at four-year U.S. undergraduate programs rose 8.9% in spring 2026, marking the third consecutive annual increase, but the recovery takes time to reach employers. The Bureau of Labor Statistics still expects approximately 115,300 openings for accountants and auditors every year through 2035, many because existing workers will change occupations or leave the labor force, including through retirement. A student entering an accounting program today does not fill the vacancy at a five-person practice next tax season. 

For small firms in particular, labor remains the constraint around an otherwise healthy business. This culminates in partners taking on staff-level work and existing employees absorbing larger workloads. There simply isn't room for new clients, so they're put on waiting lists or declined altogether.

Naturally, everyone says AI

AI fits the problem unusually well because so much accounting work consists of structured, repeatable processes. Collecting documents, categorizing transactions, reconciling accounts, preparing workpapers, managing workflows and transferring information between systems all consume time without necessarily requiring the accountant's highest-value judgment. The Bureau of Labor Statistics itself expects AI and other technologies to automate some routine accounting tasks while making advisory and analytical work more prominent. 

The SaaS era taught us to solve similar problems with software. The model was to find an inefficient workflow and build a better system. Then charge a subscription and let firms adopt the product with massive gains in productivity. But AI creates a strange second-order problem for the companies building these tools. The software simply becomes faster and cheaper to build. A company might develop a better accounting product, but other companies can build competing versions in rapid succession. Code alone is no longer the denominator of success when it becomes so easy to produce. 

The advantage shifts toward distribution, i.e., who can actually get the technology into daily usage? And also, who can convince firms to embrace this new technology given their aversion to it?

But even better software does not automatically revolutionize the field.

A practice-management system is tied into workflows, historical client information, billing, permissions, tax processes, integrations and years of accumulated staff habits. Replacing it involves risks like migration, retraining and due diligence — all while the firm tries to meet deadlines. 

This is the mismatch between technological speed and adoption speed. While an AI company might improve its product every month, a firm may reconsider its core operating system only once a decade. Firms facing the greatest capacity pressure may have the least bandwidth to evaluate a new platform. Think of it like this: A short-staffed owner working through tax season does not necessarily want another software demo, even if the product could eventually solve part of the staffing problem.

What if the technology arrives with the ownership transition?

A different distribution model is starting to emerge alongside traditional software sales. Instead of selling AI tools into independent accounting firms and waiting for them to switch, companies can acquire practices and deploy the technology directly into businesses they now operate.

Acquisition changes who makes the technology decision. A software vendor has to convince the owner to reconsider an existing system, migrate the practice and accept the risks involved in implementation. An acquirer enters at a moment when the firm is already changing, when an owner is retiring or when a succession is happening. Processes, staffing and infrastructure are already being reconsidered.

The acquiring operator can standardize its systems across multiple practices rather than winning a separate software sale at each one. The technology arrives as part of an operating transition rather than as another standalone product the firm has to evaluate. Clients do not necessarily need to experience that change directly. They can continue dealing with the accountants and relationships they already know while the operational infrastructure behind the practice changes.

AI discussions in professional services frequently jump straight to headcount, but accounting's immediate problem is that it does not have enough capacity. Any solution for the current crisis has to free up current accountants' hands, so they have more capacity to deal with operations without reducing service quality.

This is where acquisition-based deployment differs from simply installing another point solution. The operator can redesign the workflow around the technology instead of asking employees to add one more tool onto an already fragmented process. 

The intended effect is to give accountants more time to speak to clients and exercise judgment to help reconciliations happen faster, to process client documentation with less manual intervention and to distribute workflow more evenly outside of tax season.

The fork in the road

There are two paths available for accounting technology.

The first is for software companies to sell tools to accounting firms and compete for adoption inside normal purchasing cycles. The second uses ownership itself as the distribution channel. Firms change hands, and the new operator introduces standardized technology across the businesses it acquires.

Neither route guarantees better technology. Acquisition does not magically make an AI product good, and software vendors will continue to build important products for the profession. The second route simply removes one major bottleneck: waiting for thousands of independent firms to make thousands of independent switching decisions. When the underlying capacity problem is moving faster than the industry's willingness to replace systems, the latter option wins by a margin. 

Accounting M&A would be happening without AI. Owners are aging, succession remains difficult, the market is fragmented, and recurring revenue has always attracted buyers. The profession's consolidation has plenty of economic reasons behind it. AI turns acquisition into a potential technology-distribution strategy. If the profession waits for every understaffed practice to independently evaluate, purchase, migrate to and adopt a new generation of AI software, the technology may develop much faster than the industry can absorb it. Ownership transitions offer another route.