Every firm owner we talk to is doing the same math. Client demand is steady or growing. The supply of people to do the work is not. AI arrived in the middle of that squeeze, and the honest question is not whether it changes the profession. It is what changes, what does not, and what a sensible owner does about it in the next twelve months.
The math has not changed. The leverage has to come from somewhere.
The talent pipeline is tight and has been for years. The AICPA's Trends data, summarized by the Journal of Accountancy, put accounting degrees at 55,152 for the 2023-24 academic year, down 6.6% from the year before, even as most firms planned to hire as many graduates as ever. Enrollment is finally ticking up, but a freshman today is a senior associate in the 2030s. Meanwhile your clients are not getting simpler. The leverage that used to come from a deep bench of staff has to come from somewhere else. For the foreseeable future, that somewhere is technology, and increasingly that means AI.
AI compresses the production layer
Think of your firm's work in two layers. The production layer is data entry, document intake, first drafts, reconciliations, and the first pass at research. The judgment layer is everything you actually sign your name to. AI is compressing the first layer fast. Models already classify documents, extract figures, draft client letters, and summarize a regulation well enough to be a useful starting point. The trade press has tracked this shift for a while; the Woodard Report draws a useful line between broad assistants and the point solutions showing up inside the software firms already own. Both are eating production work, and that trend only goes one direction.
What gains value: judgment, relationships, and the signature
When production gets cheap, the scarce things get more valuable. Three stand out. Judgment: deciding what the numbers mean for this client, in this year, with these risks. Relationships: the trust that makes a client call you before a big decision instead of after. The signature: only a licensed human can sign a return, issue an opinion, or stand behind advice with professional accountability. None of that is automatable, and all of it is what clients were really paying for all along. AI does not threaten that work. It strips away the hours that were hiding it.
The firms that win will sell outcomes, not hours
Here is the uncomfortable part. If your pricing is built on hours, and AI removes hours, your revenue model is fighting your own efficiency. The firms positioned to win are moving toward pricing the outcome: the clean close, the filed return, the plan, the advice. That shift was already underway before AI. AI just raises the cost of waiting, because a competitor who produces the same deliverable in half the time can win your client on price or pocket the margin. You do not have to reprice the whole firm this year. You do have to stop assuming the billable hour will carry the next decade.
What does not change
Professional skepticism still applies, and applies to AI output more than anything. Models state wrong things confidently, so anything factual or numerical gets verified before it touches a client file. Ethics and confidentiality do not bend either. Tax return information is protected under IRC Section 7216, with criminal penalties attached, so client data only goes into tools your firm has vetted, contracted, and written into its WISP. And accountability never transfers. If AI drafted it and you signed it, you own it. These constraints are not obstacles to adoption. They are the reason licensed professionals will still be at the center of this profession in twenty years.
A realistic 12-month posture
You do not need a moonshot. You need three moves. First, govern it. Decide which tools are approved, write that into policy, and train the team on what may and may not go into them. Second, pick narrow wins. Choose two or three repeatable workflows, like drafting client communications or summarizing documents, and pilot them with a few people on non-sensitive work. Third, measure ROI like any capital investment. Hours saved, error rates, turnaround time. If a pilot does not pay, kill it without sentiment. A firm that does this calmly for a year will be far ahead of one that bought five licenses and hoped.
Where to go from here
The future is less dramatic than the headlines and more demanding than doing nothing. Production gets automated. Judgment, trust, and the signature carry the firm. If you want the practical version of this, start with our guide to AI adoption for accounting firms and the concrete workflows in how firms are putting Claude to work. And if you want a partner in it, this is the work we do with firms every quarter: guidance and coaching first, so you know where AI actually helps and how to use it safely, then the workflow tooling to make it real.
