What happens when AI raises EVERYONE’s game?
For the last two years, the conversation around artificial intelligence in accountancy has been framed in familiar, almost tired terms.
Will AI replace accountants?
Will compliance work disappear?
Will firms that don’t adopt it be left behind?
All of those questions are valid – but perhaps a more useful question – and one that is starting to play out in real time – is this:
What happens when AI raises the capability of both the adviser and the client, at the same time?
The first shift: better advisers
The immediate impact of AI inside accounting firms is straightforward - it removes friction.
Manual processing, data extraction, first‑draft reports, variance analysis – all of the work that historically absorbed time without adding much real insight – is being compressed or automated. The result is not just efficiency – it’s also more capacity.
Because once the compliance work is reduced, as is already happening, the firm has a choice: reduce fees, increase volume, or – more interestingly – increase the quality of what it delivers.
Increasingly, firms are choosing the third option.
That is not speculation. It is already happening. AI is allowing accountants to spend less time on processing and more time on advisory and client relationships.
The practical effect is subtle but powerful: for a given level of resource, advice can be better prepared; analysis can go deeper; scenarios can be tested more thoroughly; outputs are clearer and more tailored.
In short, the baseline standard of advice rises.
The second shift: better clients
At the same time, something equally important is happening on the other side of the table.
Clients are no longer passive recipients of advice. They are becoming AI‑literate at speed.
They are:
- Running their own financial scenarios
- Stress‑testing assumptions
- Using AI tools to interpret accounts and forecasts
- Asking more precise, informed questions
And in many cases, they are doing this before picking up the phone to their accountant.
This is not hypothetical. Many firms are already facing a situation where clients are experimenting with AI tools for analysis and planning, and expectations are shifting accordingly. We see it daily here at Scholes.
The dynamic is changing, with the conversation moving from “Can you tell me what this means?” to “I think this is what it means – do you agree, and what have I missed?”
That can lead to a fundamentally different engagement.
The collision: when both sides improve
Individually, each of these trends is significant. Together, they point towards something really interesting.
When the accountant is using AI to support high‑quality insight; and the client is using AI to interrogate and challenge that insight, it has the potential to create some rather powerful feedback loops.
Better preparation on both sides leads to sharper conversations, faster iteration, more refined thinking, and fewer unchallenged assumptions.
What emerges can be something closer to a collaborative decision‑making process than a traditional advisory relationship.
The quality of output improves not because one party is smarter, but because both are better equipped.
The uncomfortable middle
There is, however, a transition phase – and it is not entirely comfortable.
There is evidence that advisers can react negatively when clients “double‑check” advice using AI, interpreting it as a lack of trust. That reaction is perhaps understandable. For decades, expertise has been asymmetric. The professional knew more than the client. AI can erode that asymmetry.
Such discomfort may, however, be temporary. A majority of clients are unlikely, in my view, to replace the adviser. There will always be a demand for judgement, reassurance, empathy, and all the other things only humans can provide. Clients are more likely simply to raise the level at which the conversation starts.
And that ultimately benefits the adviser who is prepared to engage with it.
The third shift: the nature of value
This is where the real change lies. AI forces a separation between: answers (increasingly commoditised), and judgement (increasingly valuable)
AI can process data, identify patterns, and generate options; but it cannot (and may never be able to) understand context in a fully human sense; assess risk in a nuanced, commercial way; or take responsibility for a decision.
Professional value moves decisively into that second category.
Leading bodies in the profession are already pointing in this direction: as AI automates routine work, the role of the accountant expands toward strategic decision‑making and advisory judgement.
So the question becomes not “what is the answer?”, but “what should we do?” And that remains a human question.
The end state: a higher bar
Put all of this together, and the likely outcome is not the one many initially feared.
It is not a race to the bottom. It is a raising of the bar. Poor advice becomes easier to spot; superficial analysis becomes harder to hide; and generic outputs become less valuable.
At the same time, clear thinking, commercial judgement, structured insight, and the ability to explain and defend a position become more valuable, not less.
In other words, AI does not flatten the profession; it widens the gap between average and excellent.
A different way of seeing it
There is a tendency to think of AI as a substitute for professional capability.
A better way to think about it is as a multiplier, enabling firms to become more efficient, more insightful and, in some cases, genuinely strategic.
This could drive more engagement from clients; but also lead to more demanding requirements; with, overall, more of the “partnership” and “advisory” approach that much of the profession has talked about (but not necessarily executed perfectly), for many years now.
Conclusion
If AI only improved accountants, the result would be incremental; if it only improved clients, the result would be disruptive. However it is actually improving both, simultaneously, and the result in my view has the potential to become something quite interesting - a step‑change in the quality of thinking. AI seems to be forcing both sides of the relationship to operate at a higher level.
A final note
I wish to add, need it be said, that it is not my intention to downplay all the well-known concerns at this point in time around the quality of AI driven outputs, nor the data security aspects. But those (often very valid) concerns are not the subject of this particular article; and it seems likely that they will be resolved as time, and progress, marches on. If only we could all see round corners!
