No form, no gate. If we can show the working, we should not hide it behind an email capture.
Illustrative, anonymised. The method is real; the figures are worked examples, never a client's data.
Principal on the loop. The machine does the volume; an FCPA reads the detail and owns the call.
A demonstration, not a promise
The canonical example, end to end
An owner hands over the shoebox two weeks before lodgement: bank feeds, a family trust, a company, an inter-entity loan and a looming 30 June deadline.
- 1
Ingest
Pull the mess into one structured view
Bank feeds, prior returns, the trust deed and the loan agreement are read and reconciled into a single picture — the work no one has time to do by hand.
- 2
Find
Surface the exposure before it bites
A shareholder loan is drifting toward a Division 7A deemed dividend, and a trust distribution resolution is missing before year end.
- 3
Show the working
Trace every number to its source
The loan balance is traced to the exact transactions that built it, and the minimum yearly repayment rule is cited — you see the answer and the working.
- 4
Deliver
Produce the corrective plan
A complying loan agreement with a repayment schedule, and a distribution resolution calendar so the same gap cannot recur next year.
- 5
Hand to the principal
The judgement stays human
The system flags the exposure and the options; whether to refinance or repay is a commercial call Graham makes with the owner.
What it could not determine
What it could not determine: the owner’s cash position next quarter, or their appetite for drawing down. That is a conversation, not a calculation — so it is put to the principal, not guessed.
The judgement stays human
Graham weighs refinance versus repayment against the owner’s real cash flow, and signs the plan.
Because a co-owner wants out next year, this hands off to Ventures to paper the buy-sell before it matters.
The same spine, every practice
Different work, one way of working
Business advisory, bookkeeping, founders and NDIS back office look like different jobs. Underneath, each one runs the same spine: ingest, find, show the working, deliver, and hand the judgement to a principal.
A demonstration, not a promise
Watch it think — business advisory
A profitable business, three years of financials, and a nagging sense that margins are slipping even as revenue climbs.
- 1
Ingest
Read three years, not one
Financials are structured and set against industry benchmarks — reading the numbers, not just filing them.
- 2
Find
See what the P&L hides
Gross margin is slipping about two points a year, masked by rising revenue — the kind of drift a single-year glance misses.
- 3
Show the working
Decompose it by line
Margin is broken down by product line so you can see exactly which line is bleeding, and by how much.
- 4
Deliver
A one-page margin bridge
A clear margin bridge and the three levers that actually move it — pricing, mix and input cost.
- 5
Hand to the principal
Strategy is a human call
Which lever to pull, and how hard, is a strategy conversation Graham runs with the owner.
What it could not determine
What it could not determine: whether a price rise will cost you the customer. That is judgement and relationship — it belongs to the principal, not the model.
The judgement stays human
Graham sets the sequence of levers against the owner’s market and appetite for risk.
If the fix points to a restructure, the work hands across to Ventures.
A demonstration, not a promise
Watch it think — bookkeeping & BAS
The books are “done” for the month. The real question is whether they are right — and whether they would survive an ATO review.
- 1
Ingest
Read the coded transactions
A full month of coded transactions is read back against the accounts and the GST treatment applied.
- 2
Find
Catch the quiet errors
A recurring supplier is coded to the wrong GST treatment, and a capital item has been expensed — small mistakes with a real BAS impact.
- 3
Show the working
Name the transactions
The exact transactions are listed with the correct treatment and the effect on the BAS shown — not a vague “tidy-up”.
- 4
Deliver
Fix it, then stop it recurring
Coding is corrected and a rule is set so it self-corrects next month — a rule you own, not one rented back to you by a vendor.
- 5
Hand to the principal
Grey calls escalate
Anything ambiguous — is this a repair or an improvement? — is escalated to the principal, not guessed by the automation.
What it could not determine
What it could not determine: intent on a genuine capital-versus-repair judgement. It flags the item and escalates rather than assuming.
The judgement stays human
The principal makes the capital-versus-repair call and documents the reasoning.
When the books show you are outgrowing your current structure, that signal flows up to Sydney and Ventures.
A demonstration, not a promise
Watch it think — founders & structure
Two founders, aligned today, no shareholders’ agreement — or one signed years ago and never read since.
- 1
Ingest
Map the cap table and intentions
The cap table, any existing agreement and each founder’s stated intentions are read into one view.
- 2
Find
Locate the gaps that cost most
No drag-along or tag-along, no agreed valuation method, no leaver provisions — the gaps that turn a friendly exit into a dispute.
- 3
Show the working
Tie each gap to its trigger
Every gap is mapped to the exact scenario that triggers it — a founder exit, a death, a deadlock — so the risk is concrete, not abstract.
- 4
Deliver
Scope the agreement
A shareholders’ agreement scope: valuation mechanism, buy-sell, leaver classes and deadlock resolution — court-defensible, mapped to standards.
- 5
Hand to the principal
Fairness is negotiated, not computed
What is fair between founders is a human negotiation; the paper simply makes what you agree enforceable.
What it could not determine
What it could not determine: what a fair split between founders should be. It cannot decide fairness — only make what you agree enforceable and court-defensible.
The judgement stays human
Graham facilitates the founder conversation and signs off the structure.
The tax consequence of any buy-out flows across to Sydney to model before anyone signs.
A demonstration, not a promise
Watch it think — NDIS back office
An NDIS provider believes the back office is compliant. The audit will decide whether it is audit-ready or audit-hopeful.
- 1
Ingest
Read claims, rosters and records
Claims, rosters, worker records and incident logs are read into one structured compliance view.
- 2
Find
Surface the evidence gaps
Two workers’ screening clearances lapse inside the plan period, and one line is claimed above the price cap.
- 3
Show the working
Map each gap to its obligation
Every finding is tied to the exact record and the Commission obligation it maps to — evidence, not opinion.
- 4
Deliver
A remediation checklist
A prioritised remediation checklist and a keep-trading compliance gate, so gaps are closed before they are claimed against.
- 5
Hand to the principal
The accountant reviews first
The accountant reviews the findings before anything is asserted as compliant — the system flags, the principal confirms.
What it could not determine
What it could not determine: whether the provider is compliant overall. It flags the evidence gaps; it does not certify compliance — that stays with the operator and the reviewer.
The judgement stays human
The Local Knowledge principal reviews the evidence before sign-off.
The operating entity is Local Knowledge Pty Ltd; MyCareFinders is a referral source only, never a party to the engagement.
Send me your version of this problem
Describe your own mess in a few lines. A principal will read it and reply personally with what the intelligence would look at first — and what needs a human call.