4 Ways AI Has Transformed Tax Practice: Benefits and Challenges
Tax professionals are experiencing a fundamental shift in how they approach their work, with artificial intelligence reshaping everything from research to client delivery. This article explores four specific ways AI is changing tax practice, drawing on insights from industry experts who have implemented these technologies in real-world scenarios. The discussion covers both the measurable benefits firms are seeing and the practical challenges they face during adoption.
Challenge Positions Uncover Downstream Tax Exposures
The most concrete change has been in how I handle tax positions. When I'm working through one, I've usually already formed a view based on experience, the client's facts, and the rules I've internalized — and that view is typically correct. What I used to miss were the second- and third-order consequences: the treatment that's right on the primary issue but creates an exposure two filings down the line, or the election that's correct for this year but boxes the client in for the next three. Now I run the position through my AI tools after I've formed my view and ask it to challenge me. The judgment is still mine. The recommendation is still mine. The rigour around it has gone up materially.
The surprising part has been the client response. I expected disclosure to invite a discount conversation — "if AI is doing the work, why am I paying you?" That fear turned out to be a partner-side projection. I've started opening the relevant client meetings with a simple line: "I ran this through my AI tools, and the position looks right on the primary issue — but here's what surfaced when I asked it to challenge me." Not once has a client responded with skepticism. Every single time, the response has been appreciation for the additional rigour. Thomson Reuters' 2026 report backs this up at the market level — 87 percent of corporate tax clients say AI should be applied to their firms' work, and 67 percent have no idea whether their firm is actually doing it.
The challenge sits on the firm side, not the client side. None of that credit happens by accident. If the AI work doesn't show up in a deliverable the client sees — the agenda item, the deepened variance analysis, the planning memo with three positions instead of one — they credit you for nothing. Silence isn't neutral. It's a billing decision your clients are making for you.
Deliver BigLaw Quality Without Billable Overhead
AI made my entire practice viable. It's part of the reason I left BigLaw to launch my own firm. For decades, the economics of tax controversy work depended on leverage: partners directing associates directing paralegals through research, drafting, and document review. Now, a solo with the right tools can produce the same work product that used to require a five-person team, in a fraction of the time, at a fraction of the cost. I built a firm around flat fees for industries that get priced out of BigLaw (gamblers, content creators, nonprofits, etc.) because AI lets me deliver BigLaw-grade work without the billable hour economics.
I've been most surprised by what's left after AI strips away the administrative and repetitive tasks; clients are ultimately paying for trust and judgment. I've found that AI eliminated everything that wasn't expertise and made expertise itself the entire product. That makes for a better business model than arbitrarily billing for the seven minutes spent reading a one-page document.
Automate Reviews Raise Consistency And Control
One of the most meaningful ways AI has transformed our tax practice is in how we approach data analysis for
compliance and advisory. Historically, a significant portion of our work involved manually reviewing financial records to
identify risks, inconsistencies, and planning opportunities. That process, while necessary, was inherently time-
intensive and dependent on human capacity.
Today, we leverage AI to conduct first-level reviews of financial data at speed and scale. These tools can
systematically identify anomalies, flag high-risk transactions, and align financial information with relevant tax
treatments. This has fundamentally shifted our operating model, thus allowing us to reallocate time and resources away from manual processing toward higher-value advisory work.
The most surprising benefit has been the level of consistency AI introduces into our processes. In a discipline like tax, where accuracy and completeness are critical, AI provides a structured and repeatable layer of review that significantly reduces the risk of oversight. It strengthens our internal controls and enhances the reliability of our outputs to clients.
That said, the key challenge lies in governance rather than capability. Tax is inherently nuanced, and AI lacks the
contextual judgment required to interpret complex, real-world scenarios. As such, we have been deliberate in
embedding strong review frameworks to ensure that all AI-generated outputs are critically assessed by experienced professionals. In parallel, we place significant emphasis on data security, confidentiality, and ensuring that any AI tools we adopt meet strict regulatory and ethical standards.
Ultimately, AI is not a substitute for professional expertise. It is an enabler. It allows us to operate with greater
precision and efficiency, while sharpening our focus on strategic advisory, where human judgment, experience, and accountability remain indispensable.
Fix Chatbot Damage Leverage Pattern Detection
The biggest change has not been on our side it has been on the client side. We now spend a lot of time undoing what ChatGPT told the clients.
A client came to us spring and he was sure he could deduct his entire home mortgage as a business expense because an AI told him so. The AI was very confident. Even gave fake references. The client had already made decisions based on this information. This is a kind of work for us we call it cleaning up the mistakes made by AI.
On our side the good thing is that AI is really good at doing tasks that involve finding patterns. For example it can help us balance our books find mistakes in our records and identify missing tax forms. These are the kind of tasks that used to take up a lot of time for our staff. AI can do these tasks faster. It does not get tired.
The big challenge is not that AI makes mistakes on our side it is that clients come to us with opinions that they got from a chatbot. The problem is that the chatbot does not know the difference, between a Schedule C and a C-Corp which're tax forms that are used for different kinds of businesses.
George Dimov, CPA | Founder & CEO Dimov Tax | dimovtax.com
