Tax Return Prep from Source Documents
Turn a shoebox of client receipts, bank statements and income statements into a categorised, source-linked deduction schedule the agent can review in minutes — not an evening of data entry.
It reads each receipt, statement and income statement, proposes a category and a return line for every dollar, flags duplicates and capital items, and links each figure back to the page it came from — so the agent reviews and signs off, rather than keying.
Where this works well
The slow, invisible cost in return prep is not the tax thinking — it is the hour an agent or bookkeeper loses keying a client's pile into categories before any thinking can start. This pattern earns its keep on exactly that hour: it reads each receipt photo, bank statement PDF, STP income statement, rental statement and insurance schedule, proposes an ATO category (work expenses, motor vehicle, telephone and internet, home office, rental deductions, depreciation) and a return line for every figure, and links each number back to the page it came from.
It pays off most for an agent or bookkeeper running volume individual and rental-property returns, where the documents are varied but the categories repeat. Two things it surfaces are worth more than the speed: duplicates — the demo catches the same $347 Officeworks receipt submitted twice — and capital items that need a depreciation-versus-write-off decision rather than a straight deduction. Both are easy to miss by eye at the end of a long day and expensive to get wrong.
Where it works badly
It is weakest exactly where tax gets interesting. A document that is ambiguous about whether an expense is private or work-related — a phone bill with no business-use percentage, a tool that could be personal — gets a proposed category, but the apportionment is a judgement the tool can't make and shouldn't pretend to. If you point it at a client whose affairs are genuinely complex (trust distributions, CGT events, foreign income, mixed-use assets), the categorisation is the easy 20% and the agent is still doing all the hard 80%.
It is also only as good as the documents. A bank statement missing a page, a receipt cropped so the supplier name is gone, or a "summary" the client typed in a note app instead of an actual statement — these produce a low-confidence flag at best and a wrong line at worst. The honest test: if your prep time is dominated by chasing missing documents and resolving private-versus-business questions rather than by typing what you already have, this saves you less than you'd hope.
What it doesn't do — and shouldn't
It does not decide what is deductible, and it does not lodge. It surfaces a categorised, source-linked draft; the registered agent decides whether each claim stands and signs off before anything reaches the ATO. That line is deliberate. Lodging a return for a fee is restricted to TPB-registered agents, and the Tax Agent Services Act obligation to take reasonable care in ascertaining a client's affairs — and to act in their best interests — is a professional duty that cannot sit with software.
So the refund figure the demo shows is an orientation number, not advice. Offsets, the Medicare levy, the marginal-rate calculation and the reconciliation against the ATO's own prefill are the agent's call. The tool gets every dollar onto the right line with its source attached so the agent reaches the right answer faster — it does not become the answer.
What your data has to look like for this to work
Concretely: receipts and invoices that actually show the four things the ATO wants — supplier, amount, nature of the expense, and date — because a receipt missing any of those is a substantiation gap whether a human or the tool reads it. Bank statements as complete PDFs for the full period, not a single screenshot. The STP income statement matters most: salary and PAYG-withheld figures are only reliable once the employer has marked it 'Tax ready', so feeding it a not-yet-finalised statement gives you an incomplete income line.
Most practices have some of this in good shape and some of it as a shoebox of phone photos and forwarded emails. Getting client document intake consistent — a defined way receipts and statements arrive, named and complete, ideally captured close to the expense rather than reconstructed in July — is usually the real first job, and it is more about how information is collected than about buying a tool. That intake work is what we help with, and it is generally bigger and more valuable than the extraction layer sitting on top of it.
Could this push a client to over-claim and land us in trouble with the ATO?
No — it never lodges and never decides what is deductible. It proposes a category and a return line for each document and shows the source page behind every figure, so the registered agent checks the claim against the client's actual circumstances before approving. The substantiation judgement — whether the $300 written-evidence rule is met, whether an expense is genuinely work-related, whether apportionment is right — stays with the agent, where the Tax Agent Services Act obligation to take reasonable care sits.
Our clients send us blurry photos, bank exports and the odd handwritten receipt. Will it cope?
It handles a fair amount of mess — the demo extracts a deliberately blurry Officeworks receipt via OCR and reconciles half-year bank statements. But it flags low-confidence reads rather than guessing, and a smudged total, a cropped supplier name or a statement missing a page becomes a review item, not a silent number. Genuinely unreadable documents still need the client chased for a clean copy; the tool makes that gap visible early instead of at lodgement.
Does this replace our tax agent or bookkeeper?
No. It removes the keying and sorting — reading documents, proposing categories, matching figures, catching duplicates — so the agent spends their time on judgement calls and client advice instead of data entry. Only a TPB-registered agent can lodge a return for a fee, and the reasonable-care and best-interests obligations under the Code of Professional Conduct can't be delegated to software. The capacity it frees goes back into higher-value advisory work, not into cutting the desk.
How current does the data need to be, and what about the income statement?
It works off whatever documents you feed it for the relevant financial year. The catch is the STP income statement, which only carries reliable salary and PAYG-withheld figures once the employer has marked it 'Tax ready' (employers have until 14 July). Run it on a not-yet-finalised income statement and the salary line may be incomplete — so the agent should confirm tax-ready status before treating those figures as final.
Where does the client's financial data go, and who can see it?
This is the part we scope with you before anything is built. The data stays within an arrangement you control — the document store, the extraction step and the review screen sit inside boundaries we agree, and we don't quietly route client financials through a public chatbot. Given the privacy obligations around tax file numbers and client records, where the data lives and who can reach it is a design decision we make with you, not an afterthought.
It estimated a refund in the demo. Is that the number we'd tell the client?
No — the estimate is an orientation figure, not advice. The marginal-rate tax calculation, offsets, Medicare levy and any reconciliation against the ATO's own prefill are the agent's to confirm. The tool's job is to get every dollar categorised, sourced and on the right line so the agent reaches that final number faster — not to be the final number.
Estimated build: 4–6 weeks. Most of it is template work we've already done.
Fixed scope, fixed price, fixed dates.
The cost band reflects the engagement shape, not a per-feature line item. We work on fixed scope, fixed price, fixed dates — see the services catalogue for what falls inside each band.
Considering this for your org?
The honest place to start is a bite-sized first piece — one contained change, low risk. Tell us where it hurts; we’ll play it back, scope it, and show you what’s possible.