How Accounting Firms Should Explain AI-Assisted Bookkeeping Categorisation to Clients Without Sounding Like a Robot Did Their Taxes
The moment a client hears "we use AI for bookkeeping," half of them picture a robot lodging their BAS unsupervised. The fix isn't to hide that you use the tools — it's to say exactly what the tool does and exactly who checks it, in one short, calm sentence, before a client's imagination fills in the gap for you.
What most accounting firms get wrong
- Burying the disclosure in a privacy policy nobody reads, instead of saying it plainly where the client will actually see it.
- Using the word "AI" once, for everything, so clients can't tell whether their invoices are being categorised by software or their newsletter was drafted by a chatbot.
- Overselling the automation ("fully automated bookkeeping!") in sales material, then having to explain the human review step later when something goes wrong.
- Not naming who actually checks the AI's work, which leaves clients assuming nobody does.
- Treating the disclosure as a one-off instead of refreshing it whenever the tools or process change.
A disclosure script you can paste straight in
Here's the paragraph we give accounting clients to adapt and drop into their engagement letter, welcome pack or FAQ page:
Where this disclosure actually belongs
- Engagement letter or terms of engagement — one paragraph, plain English, no jargon.
- Welcome pack or onboarding email — the same paragraph, with a little more context for a new client.
- A short FAQ page on the website — for prospects researching before they sign.
- Not just the privacy policy — that's necessary but not sufficient; nobody reads it as "here's how my bookkeeping actually works."
Mistakes to avoid
- Saying "AI does our bookkeeping" when a person reviews every entry — that's not what's happening, and it reads worse if a client ever queries an error.
- Letting a new team member explain this ad hoc over the phone with no agreed wording — inconsistent explanations erode trust faster than the tool ever would.
- Assuming compliance-savvy clients (other accountants, lawyers) don't need the disclosure because "they'd get it" — they're often the ones with the most pointed follow-up questions.
- Forgetting to update the disclosure when you switch tools or add a new automated step.
Frequently asked questions
Do we legally have to disclose AI-assisted categorisation to clients?
Requirements are evolving and vary by professional body, so check current ATO, TPB and your professional body's guidance on AI use and client disclosure before finalising your wording — but even where it's not strictly mandated, disclosing it proactively heads off a much harder conversation later.
What if a client asks exactly which transactions the AI touched?
Be honest about what your software can and can't show. Most platforms can surface which entries were auto-suggested versus manually coded; if yours can't break that down cleanly, say so rather than guessing.
Should we downplay AI in our marketing if we don't want clients worrying?
You can choose where the emphasis sits without being misleading — lead with "reviewed by a qualified team member" rather than "powered by AI," since that's the part clients actually care about. Just don't claim there's no automation at all if there is.
Does this disclosure protect us if the AI miscategorises something that affects a client's return?
Honestly, no — a disclosure sets expectations, it doesn't replace your professional review obligations or shift liability for an error that review should have caught. Treat it as a trust-building tool, not a shield.
Keep reading 🤍
I help Gold Coast and Brisbane businesses grow with branding, websites and marketing that actually works.
Work with me ✦