Partner review capacity is maxed out
Junior staff produce reconciliations and expense postings that take significant time to review per file. Not because the work is wrong, but because coding choices, vendor classifications and format inconsistencies make you check every line against how your firm handled identical transactions before.
Every file arrives already annotated
You open a bank reconciliation or expense batch and see green annotations on transactions matching your firm's historical rules, yellow flags on the three that need judgment, and red on the two with no precedent.
Review time drops substantially. You accept more clients without hiring.
| Matched precedent exactly | 12 |
| Needs judgment | 3 |
| No precedent | 2 |
From past files to a firm standard
- 1
Upload past files
Upload 10 to 15 past reconciliations or expense coding spreadsheets from one repetitive client. The system extracts your firm's classification rules: Vendor X goes to GL code Y, transaction pattern Z gets flagged.
- 2
Upload next month
Add next month's bank statement or expense report. You get back an annotated spreadsheet: exact matches in green, close matches needing judgment in yellow, no precedent in red.
- 3
Read the deviation report
One page, before you open the file: how many transactions matched precedent exactly, how many need judgment, and estimated review time.
- 4
Lock a firm standard
Lock any classification, such as AWS charges to IT Expense, sub-account Cloud, and it applies across all future files for every client.
Three reasons to hesitate, answered
"AI makes accounting mistakes that I then have to catch, and now I'm liable for whatever it posts to the ledger."
The AI does not post transactions or make accounting judgments; it returns an annotated spreadsheet, and you review and post. It catches mechanical deviations, such as a vendor coded differently than last month, missing documentation or format inconsistencies, that need zero professional judgment.
"We tried automation before and it took three months to onboard, required IT involvement, and never worked with our actual client file formats."
You configure one repetitive client as a pilot, we build the pattern library from the sample files you upload, and you measure results. No IT, and you add complexity only after proving ROI on the first client.
"This only helps if all our clients have identical workflows, but every client is different. Half our review time is context the AI can't learn."
You start with your top 3 repetitive clients, where the same vendors and transaction types appear monthly. Clients with high uniqueness stay in your current workflow: this is triage for volume that doesn't need your expertise, not autopilot for nuanced work.
€99/month at launch. Joining the list is free.
Before you join the list
What happens to my email after I sign up?
You get one email when this launches, with early access instructions. No spam, no sharing with third parties, and you can unsubscribe anytime.
Do I pay anything now?
No. The price is €99/month when it launches, and nothing is charged for joining the list.
When does this launch?
People on the list hear first and get early access before the public launch.
Does the AI post transactions to QuickBooks or Xero, or do I review first?
You review first. The AI returns an annotated spreadsheet showing which transactions match your firm's historical patterns, and you decide what to post.
What if my clients all use different formats or accounting systems?
You configure one repetitive client as a pilot and upload sample files in whatever format you work in: Excel, CSV or PDF. We build the pattern library from those samples, and you expand to more clients only after proving ROI on the first.