For chartered accountants
A practice does not lose time writing reports. It loses time reconciling a bank statement in a spreadsheet, chasing the ledger that arrived truncated, retyping figures into a schedule, and then answering a reviewer who wants to know where a number came from four weeks after the assistant who prepared it moved to another client. That is what this is built around.
Six problems
Each of these is something that goes wrong in a practice for reasons that have nothing to do with drafting, and each is answered by a rule rather than by a feature.
An item cleared with no explanation is indistinguishable from one somebody deleted because it was inconvenient, and it is the first thing a peer reviewer asks about. Here the narration is a schema requirement rather than a comment column — a cleared item cannot be stored without one — so the reason an item was written off sits on the record beside the amount.
It does when the workbook is rebuilt from scratch each period, which is the usual arrangement. Here an item still outstanding at period end carries into the next one pointing at its predecessor, keeping the period it was first raised in. A cheque uncleared since January reads as three months old in March, which is the only version of an ageing that a reviewer can use.
A small difference is the dangerous one — too small to be a missing entry and too small to chase. The arithmetic gate cross-foots every table before anything is drafted, and a set of books that does not tie ends the job naming the account and the amount. It also runs the three checks anybody runs by hand: an account carrying the difference, one carrying half of it, and the divisible-by-nine test for a transposition. There is no bypass anywhere in the software.
Because it produces a figure of exactly the right shape and the wrong amount, which review does not catch. Rates are effective-dated rows keyed by assessment year from one reviewed, checksummed source per year, an unseeded year is refused rather than approximated, and every computation records the checksum of the set that produced it — so if a set is later found wrong, the figures it produced can be listed rather than guessed at.
Usually because somebody rounded twice, or because a figure was typed into a paragraph rather than computed. Both are closed off: the arithmetic is exact decimal throughout and rounds once at the boundary, and a number that appears in rendered prose without a computation line behind it blocks the render. The same 5,000-row reconciliation that ties to exactly zero here leaves a residual in floating point.
Preparation and signature are different acts and the software treats them that way. An assistant prepares and cannot sign; a manager reviews and can approve an adjustment, but never one they proposed themselves — a constraint in the database rather than a rule in a screen; a deliverable with an open review note cannot be signed at all. Until somebody signs, every export carries a draft watermark, and that watermark fails closed.
An engagement, end to end
Client, kind of work and assessment year recorded — none of them defaulted, because the year decides which rate set every computation under it may use.
Ledgers, statements and register exports parsed row by row, with everything that was refused shown as the file wrote it. Parsed plus rejected plus duplicate equals what you gave it.
The two sides side by side, the difference walking to zero as items are cleared with their narrations, and anything outstanding carrying forward with its original period intact.
Tax, depreciation and the schedules computed in exact decimals against the year’s rates, then every table cross-footed. If it does not tie, nothing is drafted.
Notes raised and closed, adjustments approved by somebody other than whoever proposed them, then a partner signs — and only then does the watermark come off.
The review, not a feature
Working papers are the one deliverable whose value is entirely in whether they survive being taken apart. So the software is built for the reviewer as much as for the preparer.
Said plainly
A model that writes a fluent audit observation is not hard to build and is not worth much. The work is the reconciliation, the arithmetic, the rates, the cross-foot and the trail from the figure back to the ledger row — and the drafting is only safe to add once those exist, because fluent commentary over a trial balance that does not balance is worse than no commentary at all. If your practice does not look like the six problems above, tell us where this page is wrong about it; that is worth more to us than agreement.
Which reconciliation ate the week, what a reviewer sent back, and how the ledgers actually arrive. That conversation is more useful than any demonstration we could prepare in advance.
Get in touch
No chatbot and no ticket queue. Tell us what your practice actually looks like — how many members sign, how many prepare, which of audit, tax and compliance you do most of — and someone who works on the software will reply.
info@legosphere.com