Accounting /Document processing live field guide · 8 min

FBT Return Prep Assistant

Scans the year's expense, vehicle and entertainment records to surface and classify potential fringe benefits, draft taxable values, and flag missing declarations and logbooks — so the agent reviews a structured draft instead of hunting benefits across the ledger.

Open the live demo

theater/demos/accounting_fbt-return-prep.html · sandbox · read-only

Open

FIG. 1

The live demo, running on fabricated data. Open it to step through the full flow — every output is shown for a person to approve before anything happens.

How it would work

Reads a year of expense, vehicle, entertainment and declaration records, sorts candidate fringe benefits into FBT categories with draft taxable values, and lights up the missing logbooks and declarations for the registered agent to confirm before anything is lodged.

Input 01

A year of mixed records

Expense claims and card statements, vehicle logbooks and odometer readings, entertainment and catering spend, employee declarations on file, car-parking data, and the prior-year FBT workpapers — all scoped to the 1 April – 31 March FBT year.

Agent 02

Sorts and drafts by category

Surfaces candidate benefits, sorts each into an FBT category — car, expense payment, entertainment, car parking, LAFHA, loan, residual — and drafts a taxable value with the method shown (operating-cost vs statutory-formula, apportionment, the otherwise-deductible rule).

Output 03

A draft, with the gaps visible

A categorised draft workpaper with every taxable value sourced and a missing-substantiation list, handed to the registered agent to confirm each classification, settle the judgement calls, and review before lodgement. The tool never lodges.

Where it works well

It does the hunt-and-sort across a whole year, every line, and shows the method behind each value.

Done by hand it is days of reconstructing the year line by line — and the benefit most easily missed is the one nobody coded as a benefit.
Best for an employer or agent with a recurring FBT obligation across cars, expense payments and entertainment, where the benefits repeat year to year.
The recaptured hours go back into the judgement calls — apportionment, exemptions, contested arrangements — not the clerical hunt.

The slow, error-prone part of FBT is not the arithmetic — it is finding every fringe benefit scattered across a year of expense claims, vehicle records and entertainment spend, then sorting each one into the right category before you can value it.

Where it works badly

It is confidently wrong exactly where FBT turns on judgement, and a clean draft hides a thin one.

Weak on the eligible-EV exemption since the 1 April 2025 PHEV change — a plug-in hybrid is no longer a low-emissions car unless the pre-existing-use carve-out is met, and that is a call, not a lookup.
A missing logbook silently defaults to the statutory-formula method at a higher value; that should be a flag, not a quiet assumption baked into the draft.

The honest test

If you cannot say, right now, whether your entertainment records carry the attendee and purpose detail to apportion against — this tool makes your wrong taxable value faster, not safer.

Entertainment apportionment, the otherwise-deductible rule, whether the minor-benefit exemption is genuinely available, and mixed private/business use are all calls — point it at thin records and it drafts a tidy taxable value over a guess. That is the trap.

What it doesn't do — and shouldn't

It drafts the workpaper. The registered agent classifies and lodges. That boundary is deliberate.

WHAT IT DOES

Surfaces each candidate benefit with the source record it came from

Shows the valuation method and the figure it drafted against

Flags every exemption, apportionment and missing declaration for a person

WHAT IT WON’T

Decide whether an exemption or the otherwise-deductible rule applies

Apportion contested entertainment or settle private-vs-business use

Lodge the FBT return

Lodging an FBT return for a fee is restricted to agents registered with the Tax Practitioners Board under the Tax Agent Services Act 2009, and the reasonable-care duty that comes with it cannot sit with software. A wrong classification or a missed benefit lands on the agent, not the tool — so the accountable person stays on every classification call.

What your data has to look like

Complete logbooks, declarations on file, and entertainment records that carry attendee and purpose detail.

Typical readiness

across orgs we see, before the first job

Vehicle logbooks and odometer readings

Usual weak point

Employee declarations on file

Needs shaping

Entertainment records with attendee and purpose

Needs shaping

Expense and GL data coded to spot benefits

Usual weak point

Prior-year FBT workpapers

Usually ready

The real first job

Missing declarations and incomplete logbooks are almost always the weak point — collected, if at all, in a scramble at year end. Getting declaration and logbook collection consistent through the year is usually the real first job, larger and more valuable than the AI layer on top. It is more a process change than a software one — and once the records are in order, every FBT year after is faster and right by default.

Right fit if…

You have a recurring FBT obligation across cars, expense payments and entertainment

Benefits repeat year to year and a missed one has real cost

Logbooks, declarations and entertainment detail are mostly collected through the year

You want the agent's time on the judgement calls, not the hunt

Walk away if…

Your benefit arrangements are complex, contested or one-off rather than recurring

Logbooks and employee declarations are chased at year end, if at all

Entertainment spend has no attendee or purpose detail to apportion against

You need a tool that makes the classification calls and lodges for you

Open questions

The worried-buyer questions, answered straight

It can — which is exactly why it drafts and never lodges. It surfaces each candidate fringe benefit, proposes a category and a taxable value with the source record attached, and where classification or an exemption is a judgement call it flags it for review rather than asserting it. The registered agent confirms every classification before lodgement, because the reasonable-care duty under the Tax Agent Services Act 2009 sits with them, not the tool.

It works from the records you have, and makes the gaps loud rather than silent: a missing logbook, odometer reading, employee declaration or attendee-detail entry becomes a flagged item before the deadline, not a quiet omission. A car with no logbook defaults to the statutory-formula method at a higher value — the tool flags that so the agent can chase the logbook. Getting that substantiation collected through the year is usually the first piece of work, and the piece that pays off every FBT year after.

No. It removes the hunt-and-sort grind across a year of records so the agent reviews a structured draft instead of reconstructing the year by hand. The classification judgements, the exemption and apportionment calls, and the lodgement all stay with the registered agent — the capacity it frees goes back into those calls.

Current to the FBT year you are preparing. The FBT year runs 1 April to 31 March, separate from the income year. For the 2025-26 FBT year the rate is 47%, with type-1 and type-2 gross-up rates of 2.0802 and 1.8868, and the minor-benefit exemption tests notional taxable value under $300 per benefit. The 1 April 2025 PHEV change to the electric-car exemption is already in effect. Point it at last year’s rates and it drafts confidently wrong values, so the agent confirms the figures for the year in force.

Yes. It runs inside an environment you control via API, not a public chatbot, and employee and client data is never used to train third-party models. Access is role-based and every drafted value is cited to its source record. Employee declarations and salary-packaging detail are personal information under the Privacy Act — we scope where the data sits and who can see it as part of the build. The demo runs entirely on fabricated data; Meridian Holdings is not a real employer.

It scopes records to the 1 April – 31 March FBT year and surfaces the substantiation gaps early so they can be chased before the deadline. The 2026 FBT return is due 21 May, or 25 June if a registered tax agent lodges it electronically for you. The tool prepares the draft against that window; the lodgement itself stays with the registered agent.

What it takes to build

4–6 weeks · 4 phases

Reused from template~65%

Bespoke to this skin~35%

stack · Claude · GL/expense API · rules engine · review UI

What it would cost

Fixed scope, fixed price, fixed dates.

Bite-sized first piece

One contained change, low risk

Pilot build

Most builds land here

Embedded support

Scale on proof

Considering this for your FBT season?

The honest place to start is a bite-sized first piece — one contained change, low risk. Tell us where FBT prep hurts; we'll play it back, scope it, and show you what's possible before the next 31 March.

Book a call How we work