AI for accountants: where it saves hours and where it bites

AI is now standard in accountancy, not a novelty. The latest survey work puts daily AI use among accountants at nearly half, most firms now using it in some form, and tax-firm use roughly doubling in a single year. It genuinely saves hours on drafting, reconciliations, document handling and research. The catch is that the same tools will state a wrong figure or a made-up tax rule with complete confidence. So the time saved is only real where a qualified human still signs off, and where client data never goes into a tool that has not been contracted to keep it private. Used with that discipline, AI is one of the clearest wins a practice can make this year. Used without it, it is a quiet liability.

How far this has already gone

The numbers have moved fast. In Intuit's 2025 QuickBooks Accountant Technology Survey of 700 US professionals, 46% of accountants reported using AI every day. The 2026 follow-up of 725 professionals found most firms now using AI across both client services and internal operations. A separate Blue J and CPA.com survey found AI adoption among tax firms had nearly doubled in a year. This is no longer early-adopter territory. The question for a practice is not whether to engage, it is how to do it without trading accuracy or confidentiality for speed.

Where AI is actually saving hours

The strongest wins are unglamorous, which is exactly why they pay. Document extraction and processing lead the way, with the large majority of firms using AI to pull figures off invoices, receipts and statements rather than keying them by hand. First drafts are the next obvious one: client emails, engagement letters, file notes and routine reports come back in seconds for a human to refine. Reconciliations and anomaly spotting suit AI well, because the tool is flagging things for a person to judge, not making the final call. Research is the fourth: a growing share of firms now use AI for tax research at least weekly, using it to find the starting point and the relevant source rather than to settle the answer.

Notice the pattern. In every one of these, AI does the legwork and a qualified person keeps the judgement. That is the line that separates the firms getting real value from the ones quietly accumulating risk.

Where it bites

The same speed that saves hours can cost you a client. Three risks matter most.

The first is invented detail. AI does not know when it is wrong. It will produce a confident figure, a plausible tax treatment or a citation that does not exist, and it will do so in the same calm tone as a correct answer. In accountancy, where a wrong number is not a typo but a liability, this is the risk that bites hardest. We have written before about how to stop AI mistakes reaching your clients, and the principle is simple: nothing leaves the building unchecked.

The second is confidentiality. The moment client-identifiable information goes into a public AI tool, you may have a problem, even if nothing ever leaks. Many consumer tools may use what you type to improve their models. Client data belongs only in tools that are contracted not to train on it and not to retain it, and ideally not in any tool without thought. A one-page AI policy that names the approved tools and the data red lines is the cheapest protection a practice can put in place.

The third is over-reliance dressed up as efficiency. A junior who lets AI draft the analysis and never learns to do it themselves is a future problem for the firm, not a present saving. AI should take the routine load so people can spend more time on judgement, not less.

The non-negotiables

Three rules keep a practice on the right side of all of this. A qualified human signs off on anything that reaches a client or a return, every time, with no exceptions for a busy week. Client-identifiable data never goes into a tool that has not been contracted to keep it private. And you keep a light record of where AI is used in the work, so that if a figure is ever questioned you can show how it was produced and who checked it. None of these slows a good practice down. They are simply the modern version of the controls you already run.

A safe way to start

If your firm is still finding its feet, resist the urge to buy ten tools. Start with the assistant built into the software you already pay for, add a document-extraction tool for the data-entry grind, and pick one genuinely repetitive task to point AI at first. Master those, measure the time they save, and add more only when a real need appears. The firms that struggle are usually the ones that bought breadth before they built habit.

The real prize is advisory time

The point of saving hours on data entry and drafting is not to do the same work faster, it is to free people for the work clients actually value. The capacity that used to go into processing can then go into judgement, planning and the conversations clients pay for, which is where a practice earns its advisory fees. This is where the economics of a practice shift. Compliance work is being compressed by AI across the whole profession, which means its price will follow over time. Advisory work, the part that depends on a person who understands the client and the numbers, is not being compressed in the same way.

So when you measure the return on an AI tool, do not stop at hours saved. Ask what those hours are being reinvested in. The firms that come out ahead are the ones using the time AI gives back to move into higher-value work, not the ones using it to take on more low-margin processing at the same rate. That is a choice about how you run the practice, and AI only makes it sharper.

The gap is widening

The most telling finding is not an adoption number, it is a divide. In Intuit's 2026 survey, only around 30% of firms have AI embedded as the default in day-to-day work, while over half use it only situationally, and the gap between the two is widening every year. The firms that move with discipline now are pulling ahead on capacity, turnaround and the time they can spend on advisory work. The ones waiting for AI to feel safe are not standing still, they are falling behind relative to peers who got the controls right and got on with it.

That is the real opportunity for a practice. Not to chase every tool, but to adopt a few of them well, with the sign-off, the data discipline and the records that let you move quickly without putting accuracy or confidentiality at risk.

If you want help choosing where to start and setting the guardrails that keep client work safe, our guide to choosing an AI adviser is a good next read, or book a session and we will map it to your practice.