Three-way matches purchase orders, receipts and invoices, and sends people only the exceptions that need judgment.
auto-matched
to the Scan's evidence — before any scope
to a system in production, not a pilot
modeled targets · not client results · timings are CRFT's standard engagement
A team matches documents by hand. Most matches are obvious and take a minute each; the few that are not take an afternoon. Everyone's time goes on the obvious ones, and the hard ones get resolved under deadline pressure at month end.
Where the Scan startsWith your PO, receipt and invoice formats and where the exceptions pile up.
Four lines, the same four on every system we build. Where they fall is what makes this one different from the last one.
The signals it works from. Typically connects to ERP, Accounts payable, Receiving and Supplier portal.
The calls it makes on its own, unattended, every time it runs.
Into the tools the work already lives in.
People work exceptions only. Payment release stays a human control — this is the one system in the set where an unattended mistake moves money, so the approval gate is deliberately left in place.
The same four legs as every CRFT engagement — written for this system rather than in general.
We measure the current match rate and, more usefully, categorize the exceptions. The mix tells you whether the problem is extraction, tolerance rules or upstream data — three different builds.
Scope is set by document variety and tolerance complexity, not by volume. Volume is the easy axis.
It runs in shadow against a live month and its matches are compared to the team's before anything posts.
Every human-resolved exception teaches it that pattern. The match rate climbs and the exception queue gets harder on average, which is the correct direction.
Auto-match rate, exceptions per hundred invoices, and duplicate payments caught before release.
Five inputs, a published price, a report in 48 hours. Decide with numbers — then own what gets built.