Your Clients Are Using AI. They Just Haven’t Told You.

There’s a conversation happening in your client base right now that you’re not part of.

It’s not happening in a meeting room or on a call. It’s happening at eleven o’clock at night, on a laptop, when a business owner has a question about their finances and can’t face waiting until Monday to ask their accountant.

They open, for instance, ChatGPT. They type the question. They get an answer.

And they don’t tell you about it. Or they do, and you spend a day undoing it.

This isn’t speculation. It’s the logical consequence of what’s happened to AI in the past two years, which is a shift so fast and so profound that most of the profession hasn’t fully processed it yet. The tools that once required expensive software, specialist expertise, and significant time to operate are now available to anyone with an internet connection and a free account.

But here’s what many accountants are still missing: the AI your clients are encountering isn’t just ChatGPT. It’s already inside the accounting software they use every day. And it’s being deployed right now, at scale, whether you’re aware of it or not.

Clients Are using AI/ Lady using a tablet which has AI on it

The AI Your Clients Are Already Seeing

Let’s be specific, because specificity is what makes this real.

Xero’s JAX  (which stands for Just Ask Xero) is Xero’s AI financial superagent, currently rolling out globally in beta. Its headline feature is automatic bank reconciliation: JAX works in the background, automatically matching and reconciling bank transactions where it has high confidence, with the stated goal of handling over 80% of bank statement lines in real time. Early users are reporting savings of four to seven hours per week on reconciliation alone. That is not a marginal efficiency gain. That is a restructuring of how a significant portion of bookkeeping time gets spent.

QuickBooks, as part of Intuit’s platform, launched a virtual team of AI agents for UK users in November 2025. These aren’t individual features, they are purpose-built agents operating across specific workflows: an Accounting Agent that categorises transactions, detects anomalies, and reconciles accounts; a Finance Agent generating forecasts; a VAT Agent; a Payments Agent. Intuit’s own data suggests users are saving up to 12 hours per month on monthly bookkeeping through these AI-powered features. QuickBooks is describing this as agentic AI, and positioning it as a fundamental shift in how the platform works, not a new tab, but a new paradigm.

Sage launched what it calls the UK’s first MTD for Income Tax AI Agent in September 2025, announced at Accountex Manchester. Built on Sage Copilot, the agent is designed to manage the quarterly MTD workflow end-to-end, segmenting clients by complexity, chasing documents, setting reminders, flagging issues, with the stated ambition of expanding into a connected network of agents across compliance, payroll, cash flow, and VAT.

Dext AI Assist launched just last week. Built into Dext’s existing bookkeeping platform and serving over 700,000 businesses and 12,000 accounting firms worldwide, AI Assist goes beyond Dext’s existing document capture and data extraction capabilities into judgment-based tasks: transaction categorisation, data structuring, reporting preferences. It learns from how each firm works, the nuances of how transactions are categorised, how description fields are used, how data is structured for specific clients, and applies those decisions consistently across future workflows without repeated manual intervention.

Take a moment to sit with what this list represents. These are not niche startups or experimental tools. These are the platforms that the vast majority of UK accounting firms, and their clients, are already using, today, deploying AI capabilities across the core workflows of bookkeeping and compliance. This is not the future of accounting software. It is the present state of it.

The Forty-Second Accountant

Beyond the platforms, there is a separate and equally significant shift happening at the client level.

A business owner, whether a sole trader, small limited company director, a landlord, can today open any of the major AI platforms and do the following: upload three months of bank statements, ask the AI to categorise every transaction, request a summary of income versus expenditure, ask it to flag anomalies, and ask for a rough picture of their likely tax position.

The whole process takes under a minute. The output is imperfect, the AI doesn’t know their specific circumstances, their allowable expenses, their prior year position. But it is directionally credible. Good enough to answer the question they actually had. Good enough that they don’t need to pick up the phone.

Now consider what that same task would have cost in accountant time three years ago. And consider what it costs today.

That gap between then and now, between the accountant’s hourly rate and free, is not going to close. It is going to widen.

Information Asymmetry: The Protection That's Eroding

For most of the profession’s history, the accountant’s value rested, in part, on a fundamental information asymmetry.

You knew things your clients didn’t. Not because you were hiding anything, but because the knowledge required to navigate tax law, interpret financial data, and apply professional judgment took years to acquire and expensive tools to act on. Clients needed you because the alternative was not just impractical but genuinely inaccessible.

That inaccessibility is gone.

Not entirely, the judgment, the experience, the relationship, the responsibility, none of that has gone. But the raw informational advantage? The part where you were the only person in the room who could interpret the numbers? That is eroding, steadily and irreversibly, with every improvement in the AI tools freely available to your clients, and every update to the platforms they already pay for.

The major software vendors understand this. That is precisely why Xero, QuickBooks, Sage and Dext are racing to embed AI directly into their platforms, to remain indispensable to clients who are increasingly discovering that basic financial tasks don’t require the friction they once did. The question for accountants is: what does that mean for you?

What Clients Are Actually Doing

The data makes uncomfortable reading if you’re not paying attention to it.

Over a third of UK SMEs were actively using AI in 2025, up from just 25% the year before, and more recent research puts that figure at 54% of UK firms using AI in some form. That growth trajectory, near doubling in under two years, is not a blip. It is a behaviour change taking hold across the small business market that makes up the core client base of most UK accounting practices.

These business owners are not using AI to replace their accountant. Not yet, and not deliberately. They are using it for the everyday questions, the ones they feel slightly embarrassed to ask, the ones that don’t seem worth a phone call, the ones where they just want a quick answer. Not just the ones the “bloke down the pub” used to freely provide.

But habits formed around quick answers have a way of expanding. The client who used AI to check whether a particular expense looks claimable is the same client who, six months later, might wonder whether they actually need someone to do their quarterly bookkeeping. The client who uploaded their bank statements for a rough tax estimate is the same client who, next year, might ask whether they need their accountant to prepare their year-end at all.

This is not inevitable. But it becomes more likely with every interaction a client has with AI that doesn’t involve their accountant, and every interaction that goes unacknowledged by the firm that should be leading this conversation.

The Quiet Client

Here is the dynamic that concerns me most.

When a client is unhappy, they usually tell you. When a client is finding alternatives, they usually don’t.

The clients most at risk of quietly drifting toward AI self-service are not the ones who complain about your fees or question your value. They are often the opposite, the low-maintenance clients, the ones who never seem to need much, the ones you don’t hear from between filing seasons.

What looks like a satisfied, low-touch client relationship may, in some cases, be a client who has already started to answer their own questions elsewhere. Not because they distrust you. But because the option is there, it’s instant, and asking their accountant feels disproportionate to the question.

The accountant doesn’t know this is happening. There is no alert, no notification, no trigger. The relationship just slowly, quietly, gets shallower.

By the time it becomes visible, a client who doesn’t renew, a client who asks whether they still need certain services, the habit is already formed.

The Right Response Is Not Alarm. It's Positioning.

Before I make that case, it’s worth pausing on something that landed in the Financial Times on 19th March 2026.

Paul Griggs, the US CEO of PwC, told the FT that partners who resist the advance of AI have “no place at the firm”. He left no room for ambiguity: “I don’t think anyone gets a free pass here. Anyone”. Anyone who believed they had the “opportunity to opt out” of AI is “not going to be here that long”.

PwC is simultaneously launching PwC One, an AI platform delivering automated services to clients in some cases “without a PwC person in the loop” and shifting away from hourly billing toward subscription-based pricing.

This is not a technology startup talking about disrupting the profession. This is the profession itself, at the very top, drawing a line in the sand.

The point is not that small and mid-size accounting firms should operate like PwC, they shouldn’t, and they can’t. The point is that when the Big Four start publicly telling their most senior partners that AI resistance is career-ending, the signal has well and truly crossed from “technology trend” to “business reality”.

I want to be clear about what I’m not saying.

I’m not saying AI is going to replace accountants. I don’t believe it will, not for the work that actually requires professional judgment, contextual knowledge, and trusted relationships. That work is real, it’s valuable, and AI is a long way from being able to replicate it.

What I am saying is that the category of work that used to sit below that threshold, the information-gathering, the data processing, the routine questions, the first-pass analysis, is moving. Not all at once, not uniformly, but directionally and irreversibly.

And the firms that are going to thrive are not the ones who resist this. They’re the ones who get ahead of it, by using AI themselves to automate the routine, freeing up the capacity to do the high-value work that clients genuinely cannot get from a chatbot or a software platform.

Notably, the AI tools being deployed by Xero, QuickBooks, Sage and Dext are, for the most part, designed to support accountants and bookkeepers, not replace them (feel free to agree or disagree with this statement reader). JAX keeps the accountant in control. Dext AI Assist applies the firm’s own judgment at scale, not a generic algorithm. These platforms understand that the accountant relationship is still load-bearing. They are not trying to disintermediate the profession. They are trying to make the profession more efficient.

But efficiency that isn’t harnessed becomes redundancy. If AI tools free up ten hours a week in your firm and you don’t redirect that capacity into higher-value work, you haven’t become more efficient. You’ve just given AI the work without replacing the revenue it was generating.

The democratisation of financial intelligence is not the accountant’s enemy. It is, handled correctly, the accountant’s greatest opportunity to finally shed the commodity work and become what every good accountant already knows they could be: a genuine strategic partner to their clients’ businesses.

But that repositioning doesn’t happen automatically. It requires a decision. And it requires that decision to be made before the client has already moved on.

The Question Worth Asking This Week

Before you close this article, I want to leave you with one question, and I’d encourage you to answer it honestly rather than optimistically.

When did one of your clients last use AI to answer a question about their finances?

If your answer is “I don’t know”, that’s the answer. And it’s probably worth finding out.

Daniel Lawrence is the Founder of bots for that and creator of the beanieverse AI-native automated accounting platform. He writes and speaks on AI transformation in accounting from the perspective of someone who spent over a decade deploying enterprise automation and AI in regulated industries before returning his attention to the profession where his career began.