Notes

AI for Accountants in Australia, the Chasing Not the Signing

How can accountants use AI?

The best use of AI for accountants is the preparation around the professional work: chasing clients for records, sorting what arrives, suggesting how transactions should be coded, drafting workpapers and keeping track of every lodgement deadline. The accountant still reviews the output, gives the advice and signs off the return.

In an Australian practice that line is not only good sense, it is the rule. The Tax Practitioners Board has said plainly that practitioners remain responsible for the services they provide, whatever tools they use. So the useful question is not whether AI can do a tax return. It is which hours between a client's first email and your final review a machine can carry.

Ask any principal where tax season goes and the answer is rarely the technical work. It goes on the missing bank statement, the third reminder about the rental property schedule, the receipts photographed sideways, and the BAS client who sends everything on the due date.

Which accounting tasks should I automate first?

Start with work that is high volume, rule driven and already checked by a person. In most practices that means one of four lanes.

All four sit inside the admin disappears, wired into the practice software and ledger tools you already use, with a person approving anything that leaves the office. Pick the lane that loses the most hours with the least judgement in it, measure it, and build that first.

What does the TPB say about using AI?

The Tax Practitioners Board published its guidance on this in July 2026: TPB(GS) 55/2026, The use of Artificial Intelligence and the Code of Professional Conduct. It is not a ban. It explains how your existing Code obligations apply when AI is involved.

The practical points for a practice are clear enough to build around:

That shapes how we build. Data stays in accounts you own, the build logs what it touched and who approved it, and nothing lodges or advises on its own. How the guidance applies to your practice, your engagement letters and your staff is a question for your own adviser or professional body, not a software supplier.

Can AI prepare workpapers and answer technical questions?

It can draft workpapers for review and help your team find what your practice already knows. It should never be the source of a tax position.

A build over your own prior year files, checklists and internal notes can pull together a first draft workpaper, list the changes from last year and point to where a similar issue was handled before. Every answer cites the document it came from, so the reviewer reads the source rather than trusting a summary. That is what agents on your data has to mean in a practice: if it cannot cite it, it does not say it. Why that rule matters is covered in AI hallucinations in business tools.

What it does not do is decide how a law applies to a client. The TPB guidance warns that AI can produce inaccurate information and cannot replace tax knowledge or experience. A confident wrong answer in a workpaper is worse than a blank one.

Bookkeepers and BAS agents: same rules, faster rhythm

The lanes are the same, the rhythm is faster. A BAS practice runs monthly or quarterly cycles across many clients, so the gains come from chasing and coding at volume rather than from tax season peaks.

Registered BAS agents sit under the same Code and the same TPB guidance as tax agents. The approval line does not move: the build prepares, a registered person reviews and lodges.

A sensible first project

One lane, one team, thirty days, with the numbers written down before you start. Records chasing is the usual first choice, because the volume is high, nobody enjoys it, and the result is easy to count.

Take the baseline first: how many clients are waiting on records, how many reminders a typical job needs, and how many staff hours go on chasing each week. Then judge it honestly at thirty days. If you run a broader practice with legal or advisory work alongside, AI for professional services firms covers the billable hour question that sits underneath all of this.

Frequently asked questions

Will AI replace accountants?

No. It takes the chasing, sorting and first drafting, which frees accountants for review, advice and client conversations. The TPB is explicit that responsibility stays with the practitioner, so the person cannot be designed out.

Can I put client data into ChatGPT?

Be very careful. Under the TPB guidance, entering client information into an AI tool may be a disclosure that needs your client's permission, and public tools may store or use what you enter. The OAIC's guidance on privacy and commercially available AI products says the same. A build that runs in accounts you control, with client consent covered in your engagement letter, is the safer path.

Does it work with Xero or MYOB?

It is built around the systems your practice already uses, reading from them and writing back only where you allow. Which systems connect, and how, is agreed in writing before any build starts.

Can AI lodge returns or BAS on its own?

No, not in our builds. Preparation can be automated. Review and lodgement stay with a registered person, every time.

See it before you commit

If you would like to see the crew work before you talk to anyone, the test bench lets you put a question to Maggie about a staff handbook, on a staged business. Ask it something awkward and watch whether it cites its source or admits it does not know. That is the behaviour a practice needs.

Keep reading