Attest
SecuritySign inRe-close last month
For controllers at companies that deploy hardware

Re-close last month. Watch it tie.

Upload last month's bank export and your GL. Attest learns the rules you already follow, writes them back to you in plain English, re-derives the month from the raw bank rows, and shows you what ties and what does not. Then it runs next month without the GL.

Re-close last monthWalk the sample close

Ten minutes. No bank connection. No sign-in for the sample.

Ridgeline Fiber Co. · July 2026 · re-derived from 360 bank rows
113/ 118 entries tie to the cent
  • Calix Inc18,440.00
  • Delta Fiber Build LLC18,400.00
  • Hudson Yards Industrial LLC14,500.00
  • Calix Inc booked twice in your ledger18,440.00
  • Payroll capitalization · Jul 154,873.25
  • Depreciation · PRJ-4412, not in your ledger3,119.74
  • 3 more differences, each explained

Every entry re-derived independently. Nothing posted to anyone's ledger.

What it does

Learns the rules you already follow

Every bank row with a ledger line behind it is an example of a rule you already apply. Attest groups them, writes each one as a sentence, and shows you how many rows it will claim — before it is used on anything.

What it proves

Re-derives the month independently

The engine books the month from the bank rows and the rules you approved, and a separate pass re-derives every entry and checks it. Your ledger is read once, at the end, only to compare against.

What you keep

Next month, without the GL

The rules July taught are the rules August runs on. New vendors and possible duplicates should be the only things that need you — and they are the only things that ask.

The part that matters

A difference is a finding, not a grade.

Where the re-derived month and your ledger disagree you get both versions, the delta, and the bank row underneath — never a verdict on your books. Choosing a side writes a rule, under your name, with the reason. Nothing posts to your ledger, ever: Attest drafts, and a person decides.

On the sample company's July: 113 of 118 entries reproduced to the cent from 360 bank rows, and 6 differences, each one explained. See the claims and the tests that prove them.