Finance
Give Bwat the bank export, the mobile money statement, the ledger or the board pack. It lines them up, shows you how it matched them, and hands back a workbook with the problems on their own sheet. Your review starts where the problems are.
Bwat helps with analysis. It isn't your accounting system, your auditor or your approver.
Mobile money statement vs sales book, March
| Date | Details | UGX | Result |
|---|---|---|---|
| 2 Mar | Order 1043, Nakato | 250,000 | Matched |
| 5 Mar | Order 1051, Okello | 180,000 | Matched |
| 9 Mar | Transfer, no reference | 95,000 | Not in sales book |
| 17 Mar | Order 1077, Achieng | 310,000 | Matched |
| 31 Mar | Order 1090, Mugisha | 90,000 | Paid 1 April |
One payment with no sale behind it (UGX 95,000), and one timing difference.
Ask why a number moved and Bwat splits the change into price, volume, timing and one-off items, naming the sheet and cell for each. Then it drafts the commentary for management, with anything it couldn't confirm marked for you to check.
Why did March come in UGX 14M under budget?
Mostly timing. UGX 9M of fuel was booked in April instead of March 1. The rest is lower sales in two branches 2. One line, office rent, is over budget because of the new lease 3.
Get the sheets, formulas, assumptions and odd values explained without clicking through every tab.
Turn your assumptions into a calculation with formulas another analyst can follow and change.
Separate price, volume, timing, mix and data problems before anyone writes the explanation.
Check a big file or folder for duplicates, gaps, strange movements and records that break the rule.
Memo, board narrative, charts or slides, built from figures you've already reviewed.
Pull payment terms, covenants, dates, fees and conditions out of the contracts behind the numbers.
Attach the files, fill in the brackets, and ask for output someone else can check.
“Reconcile the bank export and ledger I've attached for [period]. Before matching, look at both files and suggest the match keys, the date tolerance and the amount tolerance. Then give me an Excel workbook with matched items, timing differences and unexplained exceptions on separate sheets, keeping the original row IDs and the formulas you used.”
“Compare actuals with the budget and the latest forecast for [period]. Quantify the biggest favourable and adverse variances, split them into price, volume, timing and classification where the data allows, and cite the exact sheet and cells. Draft a short management commentary, and mark any explanation that needs confirming.”
“Review this financial model without changing it. Map the input, calculation and output sheets, then find hard-coded numbers inside formulas, broken references, mismatched periods, circular logic and assumptions nobody wrote down. Rank what you find by how much it could move the result, and cite each cell.”
“List every financial obligation, fee, covenant, reporting deadline, default trigger and remedy in this agreement. Put them in a table with the obligation, who owes it, the amount or formula, the due date or frequency, any condition, and the clause number. Don't fill in a value the agreement doesn't state.”
A spreadsheet can add up perfectly and still answer the wrong question. AI can also pick the wrong matching rule, misread a period, or explain a variance with a story the data doesn't prove.
That's why Bwat shows its working, keeps the source references, and labels its assumptions for someone in finance to confirm. It gets the analysis ready faster. It doesn't approve it.
You already know the answers and the exceptions, so you'll see straight away whether it finds them.
