Accounting and AI in 2026: what it means for UK practices
Most UK accountants now use AI, yet few say it has changed how they work. We separate what is real from marketing, and what a small practice should do next.
- Published
- Reading time
- 8 min read
Ask ten accountants about AI this year and you will hear ten different answers.
One will tell you it writes half their client emails now. Another tried it once, watched it invent a tax rule, and has not opened it since. Most sit somewhere in between. A little curious. A little tired of hearing about it.
That middle ground is where most of the profession really is. So rather than add to the noise, we went through the research published in 2026 and asked one plain question. What is actually changing inside UK practices, and what is just a good sales deck?
Almost everyone uses it. Almost nobody is transformed by it.#
The headline numbers look like they contradict each other. They don't.
Wolters Kluwer's UK research found that 91% of accountants are either using AI or plan to. That sounds like a revolution.
Then look at the State of the Nation report from AccountingWEB and Sage, published in February. Only 7% of accountants and bookkeepers said AI had transformed their work. More than half, 53%, said the benefits so far were minimal or moderate.
Both can be true at once. In 2026, "using AI" often means a chatbot open in another tab, or a feature that switched itself on inside software you already pay for. That counts as use. It rarely counts as change.
- say AI has transformed their work (AccountingWEB and Sage)
- 7%
- say the benefits so far are minimal or moderate
- 53%
- of firms asked at Accountex 2026 did not know where to start
- 65%
say AI has transformed their work (AccountingWEB and Sage)
say the benefits so far are minimal or moderate
of firms asked at Accountex 2026 did not know where to start
The last number is the one we keep coming back to. When Workiro asked firms at Accountex London this year, 65% said they did not know where to start with AI. The smaller the firm, the more lost people felt.
That matches what we hear. The tools are not really the problem. Time is. A two partner practice in January does not have a spare afternoon to experiment with anything.
Where AI is genuinely earning its keep#
Take away the demos and the useful work falls into four quite ordinary jobs.
Getting data in. Reading receipts, invoices and bank statements, then suggesting how to code them. This is the oldest use of machine learning in accounting software, and it keeps getting better. Xero, Sage and Intuit now all ship assistants inside their products aimed at reconciliation, document capture and chasing.
Writing the first draft. A client email. A summary of an engagement letter. A plain English explanation of why someone's tax bill went up. The AI writes something rough and you fix it. Nobody should send the first version.
Summarising. Long HMRC guidance, a tangled email thread, your notes from a client call. AI is good at turning a lot of words into fewer words. It is less good at knowing which words matter.
Answering "where are we?" questions. Which clients still owe us records? What is due in the next fortnight? Honestly, this is less about clever AI and more about having the information in one place to begin with. (Which is, we admit, exactly why we are building a CRM for accounting practices.)
So how much time does all this save? For its Future of Professionals report, Thomson Reuters surveyed 2,275 professionals across legal, tax, accounting and related fields. On average they expected AI to save them about five hours a week, or roughly 240 hours a year.
That is a prediction, not a stopwatch reading. Still, five hours is more than half a working day. In the weeks before 31 January, half a day matters.
The part nobody puts in the brochure#
Every useful AI tool needs a person checking its work. That is not a bug that will be fixed next year. It is simply how these systems work.
Language models are built to produce text that sounds right. Most of the time it is right. Now and then it will confidently quote a threshold from three years ago, or a relief that never existed. In most industries that is embarrassing. In accounting it can mean a wrong return with your client's name on it and your firm's name behind it.
So we would suggest two rules before anything else.
- Never paste client data into a free consumer chatbot. Find out where the data goes, who can see it and whether it is used to train the model. Your duties under UK GDPR do not pause because a tool is convenient.
- Keep a human name against every piece of advice. If you would not sign it, do not send it.
Making Tax Digital is the bigger story of 2026#
If AI is the trend everyone talks about, Making Tax Digital for Income Tax is the change that actually landed on desks this year.
Since 6 April 2026, sole traders and landlords with qualifying income over £50,000 have had to keep digital records and send HMRC quarterly updates, followed by a final declaration. Five submissions a year where there used to be one.
HMRC's own policy paper estimated around 780,000 people in this first group, and later figures put it nearer 864,000. It does not stop there either. The threshold falls to £30,000 in April 2027 and to £20,000 in April 2028.
There is a detail in ICAEW's analysis of HMRC's numbers that deserves more attention. Around 217,000 people in the April 2026 group are not represented by an agent at all. Many of them will be looking for one.
This is where software, and yes AI, really earns its place. Quarterly updates turn a once a year scramble into a steady rhythm. Practices that already capture data automatically and chase clients on a schedule will cope fine. Practices still running on a spreadsheet and memory are going to feel every single quarter.
Making Tax Digital turns one deadline a year into five. Good systems turn five deadlines into a routine.
So, will AI take accountants' jobs?#
It is the question every trainee is quietly asking.
In May, ICAEW put it to the leaders of 35 mid sized UK firms. Their answer was calmer than either the cheerleaders or the doom merchants. 68% thought AI would reduce demand for some early career roles. But 83% said that would not simply mean fewer people. They expect the mix to change instead. More school leavers alongside graduates. More offshoring, which 40% expect to grow, and more outsourcing, at 29%.
The same firms were ambitious and honest in the same breath. In ICAEW's wider findings, 86% said their technology strategy explicitly includes AI. Only 17% felt able to judge what it would do to their own workforce.
Our reading is fairly simple. The tasks at the bottom of the ladder, keying in figures and ticking boxes, are shrinking. The parts that need judgement, explanation and a real relationship with the client are not. If anything those parts are worth more when the routine work gets faster.
What a small practice should actually do next#
You do not need an AI strategy document. You need a few sensible habits.
- Use what you already pay for. Most bookkeeping and practice software now has AI features built in. Switch them on, try them with a few clients and see what sticks.
- Pick one boring job. Bank coding, document chasing or standard emails. Time it for a month before and a month after.
- Write your rules down. Which tools are allowed, what data can go into them and who checks the output. One page is plenty.
- Get your client information into one place. AI can only work with what it can see. Deadlines in one system, emails in another and notes in a paper diary will defeat any tool you buy.
- Talk to your clients about it. Plenty of small business owners are already using AI for their own books. Be the person who helps them do it safely, rather than the one who finds out at year end.
Where we stand#
DRP Digital was founded by an accountant, so we are biased in one particular way. We think the best software for accountants starts from the work, not from the technology. AI is a genuinely useful new tool in that box. It is not the box.
That is the thinking behind the CRM for accounting practices we are building for 2027. If you would like to see it before anyone else, join the early access list.
Sources#
- AI in 2026: the accounting state of the nation, AccountingWEB and Sage, February 2026
- How is AI being used by accountants in the UK, Wolters Kluwer
- What UK accountants said: the Accountex 2026 report, Workiro
- Future of Professionals report 2025, Thomson Reuters Institute
- UK accountants still in high demand despite AI jobs shift, ICAEW, May 2026
- Mid tier firms focused on AI and consolidation, ICAEW, May 2026
- Making Tax Digital for Income Tax Self Assessment for sole traders and landlords, HMRC policy paper
- Up to 3 million individuals may need to comply with MTD, ICAEW, August 2025
Frequently asked questions
How many UK accountants are using AI in 2026?
Most of them, in some form. Wolters Kluwer found 91% of UK accountants are using AI or plan to. But use is broad. In the AccountingWEB and Sage State of the Nation survey, only 7% said AI had transformed their work, and 53% described the benefits as minimal or moderate.
Will AI replace accountants?
The evidence so far says no, but the work is changing. In ICAEW's 2026 survey of 35 mid sized firms, 68% expected AI to reduce demand for some early career roles, while 83% said this would not simply mean fewer people. Judgement, explanation and client relationships remain human work.
Is it safe to put client data into an AI tool?
Only if you know where the data goes. Check whether the provider stores it, who can see it and whether it is used to train the model. Free consumer chatbots are rarely the right place for client records, and your duties under UK GDPR still apply.
What is the first AI task a small practice should try?
Pick one repetitive job, such as coding bank transactions, chasing missing documents or drafting standard client emails. Try it on a handful of clients for a month and time it before and after.