Professional services

What AI realistically does for a small accounting, bookkeeping or conveyancing practice

In a small professional practice, the principal usually is the admin department: the quoting, the correspondence, the file notes, the chasing. That is exactly the work AI is good at supporting — and the professional judgement clients actually pay for is exactly the work it should never touch. The craft is in drawing that line well.

By Alex CarpenterUnlocking TechnologyUpdated June 20267 min read

The one thing to take away

AI will not do your professional work, and you should be suspicious of anyone who says it will. What it reliably does is compress the work around the professional work — drafting, summarising, documenting, responding — while you keep every judgement and every signature. For the owner-operated practices we meet around Mount Barker and the Adelaide Hills, that is routinely several hours a week back in the principal’s hands, without a single client-facing risk being taken.

The pattern that holds

Across the research on small and medium firms, augmentation — AI supporting a person who stays in charge — is the mode that actually delivers. That suits regulated professions unusually well, because the line AI must not cross is one your professional standards already draw for you.

Three worries, taken seriously

The hesitations we hear from accountants, bookkeepers and conveyancers are good ones — they come from professional discipline, not technophobia. They deserve straight answers.

“AI makes things up — I can’t have that anywhere near client work.”

Correct instinct, wrong conclusion. Models do produce confident errors, which is why nothing AI-assisted should reach a client without review. But “a person checks every output” is a design rule you can actually enforce — and checking a draft is far faster than writing one. The risk is real; it is also manageable, the same way you already manage the risk of a junior’s first draft.

“Our work is too regulated for AI.”

Regulation is a reason to adopt carefully, not a reason to abstain. The regulated core — advice, sign-off, lodgement, settlement — stays human. Around it sits a large ring of unregulated work: correspondence, scheduling, file notes, internal procedures, chasing documents. That ring is where the hours go, and it is where AI safely earns its keep.

“We’re too small for this to be worth it.”

The evidence points the other way: for small firms the dependable gains are incremental — faster drafting, quicker responses, better documentation — and those gains land hardest where the principal does the admin personally. An hour a day back in a three-person practice is a different business; no enterprise transformation required.

The four moves that work in a practice

None of these require new systems, a transformation budget, or trusting a model with anything that matters. Each one is small, reversible, and provable within a few weeks.

Let AI draft, and keep the sign-off

The most dependable gains in a practice are in first drafts: client letters and emails, engagement letters from your own precedents, file notes from meeting recordings, plain-English explanations of a process for a client. The professional still reads, corrects and signs everything — that is not a limitation of the approach, it is the approach. Drafting time falls; responsibility doesn’t move.

In practicePick one recurring letter or email type. Build a reusable prompt from your best past examples and have the AI produce first drafts only.

Use AI to summarise, never to conclude

Long documents are everyday work in these professions — files, contracts, correspondence trails, legislation updates. AI is genuinely good at producing a working summary that tells you where to look. It is not decision-grade: a summary can miss the one clause or figure that matters. Treat the summary as a map of the document, then read the parts that count yourself.

In practiceAllow AI summaries as a reading aid, with a written rule: no advice, lodgement or settlement step relies on a summary alone.

Protect client confidentiality before anything else

Accountants, bookkeepers and conveyancers hold exactly the data that must never leak: financials, identity documents, contracts, settlement details. The quiet risk is staff pasting client material into free public tools with no rules in place. The fix is not banning AI — it is an approved-tools register and a short acceptable-use policy, so everyone knows which tools are safe for which data.

In practiceWrite down which AI tools are approved, what client data may and may not enter them, and who approves exceptions. One page is enough to start.

Preserve the work that is actually the job

Judgement calls, advice, anything touching professional standards or liability, and the client relationship itself should be deliberately preserved for people. This is where today’s models are weakest and where your indemnity, registration and reputation live. A practice that automates its admin and protects its judgement gets the upside of AI without gambling what the practice is built on.

In practiceList your recurring tasks and tag each one: automate, augment, or preserve. Start only with the clearly safe, repeatable ones.

Where to start, in order

Start with the confidentiality rules, because everything else depends on them. Then pick the single most repetitive piece of writing in the practice and augment that one task until it is boringly reliable. Only then widen out. A practice that adopts AI this way builds something its competitors can’t paste in from a chatbot: a tested, written-down way of working that new staff can be trained into. The general method — automate, augment or preserve, task by task — is set out separately, and it applies to a three-person conveyancing firm just as it does to a thirty-person one.

A note on honesty

The research on AI in small firms is young, largely overseas and correlational, and evidence specific to Australian professional practices is thinner still. The patterns above are the ones that hold up best — but treat them as informed judgement, not a guarantee, and expect us to tell you where they run out.

Bring the task that eats your week

Tell me the one piece of recurring desk work that costs your practice the most time — quoting, letters, file notes, chasing documents — and I'll give you a straight view on whether to automate it, augment it, or leave it alone, and how to pilot it without putting a single client file at risk.

Get in touch

Keep reading

Augmentation, not automation — the general method this article applies to professional practices — and why it beats replacement.
A practitioner’s guide to AI for Australian SMEs — the eight evidence-backed principles, with what the research does and doesn’t support.

About this piece. It applies current research on how small and medium firms realise value from AI — alongside an active PhD into how Australian SME leaders respond to it — to the specific shape of small professional practices: accountants, bookkeepers, conveyancers and their neighbours. The evidence is early, and we will always tell you where it runs out.