On being right about AI (and slightly wrong about what to do next)
Our accountant declined to help with this site.
He was very polite about it. He said he was not trying to be difficult, and hoped we would not take offence. He would not be using AI in his accountancy practice: it is his licence on the line, and he has never had any interest in having AI try to do his work for him. He was sorry he could not be of more help.
Then he mentioned the two apprentices he had just hired, straight out of school. He likes being in a position to offer someone in need of work a regular income. And the tasks that get described as AI opportunities in accountancy, he said, are exactly the tasks young people get hired to do as part of their training: entering bank statement data, processing invoices and receipts, learning the work by doing it. AI, in his view, is not transforming accountancy. It is taking the bottom rung of the ladder away.
We did not argue with him. He is not wrong about any of that. He is describing something real, and something that matters.
What he got right
The professional caution is well-founded. Accountancy carries regulatory and fiduciary responsibility that most business software does not. The idea of delegating judgement to a system that sometimes produces plausible-sounding nonsense is not, in that context, an overreaction. It is a reasonable assessment of what the technology actually does and what the consequences of error actually are.
The point about young workers is also fair. The tasks that AI and automation handle most readily are the structured, repetitive ones: data entry, document processing, invoice matching. These are not trivial tasks. They are how people learn. The efficiency case for automating them is real, and so is the cost.
Both of these things can be true at once. AI can be genuinely useful in some contexts and genuinely harmful in others. A site that tells you only the first half of that is not being straight with you.
Where it gets more complicated
The harder question, and the one our accountant's email does not quite settle, is where the line is.
His concern is with AI doing accountancy work: making judgements, interpreting figures, signing off on things that require professional accountability. That concern is legitimate. It is also not what this site is primarily about.
The three things on our accountancy sector page are not about replacing professional judgement. Two of them are the work he is describing: pulling transaction data out of bank statements and receipts, and chasing clients for the documents they have not sent. The third drafts the standard sections of year-end notes for a qualified person to check and sign. None of these carry the regulatory weight he is describing, and the page says so. Done by hand, they take up a significant amount of time that could be spent on the work that does require a qualified person. Which is also, as he pointed out, exactly the work an apprentice is given.
Whether any of that applies to his practice is genuinely a question. It might not. Some practices are already running efficiently enough that the marginal gain is not worth the change. That is a legitimate conclusion, and this site will not argue you out of it.
The apprentice point, which deserves a direct answer
The concern about young workers and entry-level jobs is not one we want to brush past.
It is a real tension, and it is not unique to accountancy. Automation has always changed what the bottom of a career ladder looks like. Sometimes that has meant those rungs disappear. Sometimes it has meant people get to start somewhere more interesting. Usually it is some of both, in proportions that are not easy to predict in advance and not evenly distributed in practice.
We do not think that tension gets resolved by ignoring the tools, and we do not think it gets resolved by pretending it is not a tension. The honest position is that it is a live question, the outcome is not settled, and anyone who tells you with confidence which way it goes is guessing.
What our accountant is doing, hiring two apprentices and giving them the work by hand, is not irrational. It is a considered choice about what he values and what kind of business he wants to run. That is entirely his to make.
What this site actually does
The scan on this site is not trying to convince anyone that AI is good or that it should be adopted. It is trying to work out what, if anything, would help a specific business, and say plainly what that is.
Sometimes that will be AI. Often it will be automation, which is a different thing: rules-based, predictable, and usually the better answer for straightforward tasks. Sometimes it will be neither, and the honest report will say so.
Our accountant's instinct that most of what gets sold under the AI label is not worth buying is a reasonable starting point. Where we would push back, gently, is on the step from that to nothing here is worth examining. There is a smaller, quieter set of things that work. The point of this site is to find which of them apply to you, and to be honest when the answer is none of them yet.
If you want the longer version of the sceptical argument and where it does and does not hold, it's here.
If you arrived via one of our LinkedIn videos and the format itself bothered you, we've addressed that separately.