Tax Pros Are Using AI. Now the Billable Hour Has a Problem.

Tax pros are moving past the “let’s test AI” phase. A new report from Blue J found that 60% of tax practitioners have already adopted AI tools, up from 33% last year. Another 32% plan to adopt them in the next few years. Meanwhile, the share of practitioners with no current AI plans dropped from 30% to 7%.

That is a big change in a short time. For the last few years, AI in tax research was something firms watched from a safe distance. Some ran pilots. Some let a few curious staffers experiment. Some waited for the tools to get better. That period is fading. The new question is not whether AI belongs in tax research. It is how firms should manage it, train people on it, and price the work once it starts saving real time.

The survey included 1,000 U.S. accountants working at tax and accounting firms of different sizes. Each respondent said they do tax research at least a few times a year. According to Blue J, common research topics include deductions, corporate tax questions, tax rates, individual tax, tax procedures, compliance, and partnership taxation.

AI is becoming a normal research tool

It is not hard to see why tax pros are using AI for research. Tax research is slow work. A simple client question can turn into a pile of tabs, code sections, memos, and old guidance. AI does not remove the need for judgment, but it can make the first pass faster.

That is what respondents are seeing. In the report, 84% said AI saves time. The same share said it helps simplify complex tax rules. Half said they use that saved time to improve client response and project delivery. Others said it helps with work-life balance or helps them deliver better advice.

That is the better way to understand AI in tax. It is not a magic replacement for a good tax professional. It is more like a junior researcher who can dig through the first layer quickly, while the human still checks the work, applies judgment, and owns the answer.

Most firms are using general AI tools

The odd part is that most practitioners are not using tax-specific AI tools. The report found that 90% use public AI models such as ChatGPT, Gemini, Perplexity, or Claude. About 67% use legacy tax research platforms with research or AI features, including tools from Thomson Reuters, Bloomberg, and Wolters Kluwer. Another 64% still use search engines.

Tax-specialized AI tools sit lower on the list. Only 46% said they use AI tools built specifically for tax research. Among those tools, 25% reported access to TaxGPT, 22% to Blue J, and 17% to Accordance.

That creates a real risk question for firms. Public AI tools are easy to try. They are fast and familiar. But tax research is not the same as drafting a casual email. The answer needs support. The sources matter. Client facts matter. And a confident wrong answer can become expensive fast.

So adoption is only part of the story. The next layer is governance. Firms need to decide which tools are allowed, what client information can be entered, how output gets reviewed, and who signs off before anything reaches a client.

The billing issue is getting harder to ignore

The more interesting business problem is pricing. If AI can turn hours of research into minutes, hourly billing starts to look strange. The client still wants the answer, and the answer may still be valuable. But the time needed to produce it may shrink.

That is why 69% of AI adopters in the report are considering alternatives to traditional hourly billing. Value-based billing is the most common option, with 37% considering it. Another 30% are looking at hybrid models, while only 2% are considering fixed-fee billing. Even firms that plan to keep hourly billing are adjusting. About 28% said they intend to maintain hourly billing but increase realization. Just 3% said they are not considering any billing changes.

This is the part that should make firm leaders pay attention. If a firm sells time, then saving time can look like a margin problem. But if a firm sells judgment, speed, and better answers, AI can become a margin tool.

Clients have never really paid only for the minutes on the clock. They pay for the experience behind the answer. They pay for the person who knows which issue matters, which rule applies, and which risk is worth flagging. AI makes that truth harder to avoid.

The shift will not be easy

Changing billing models is not as simple as updating an engagement letter. Hourly billing is built deep into the accounting firm model. It affects staffing, realization, partner reviews, client expectations, and how firms explain their value.

That is why many firms will probably move slowly. A full jump to value pricing may be too much for some practices. Hybrid billing may be the easier bridge. Routine work can stay more standardized. Higher judgment work can be priced around value. Some hourly work may remain where it still makes sense.

But the direction is clear. AI is making tax research faster, and firms need a pricing model that does not punish them for being efficient.

The firms that get this right will not just be the ones with the best tools. They will be the ones with the best system around the tools. They will know when AI can help, when a human needs to slow down, and how to charge for the value of the final answer.

The firms that wait may still adopt AI. But they may end up with a modern research process stuck inside an old billing model. That is where the real mess starts.