Exact
Identical keys clear as a pair. Use this when references are clean.
Advanced matching
Not every settlement is one invoice and one payment. Advanced matching covers exact, tolerance, fuzzy, partial, one-to-many, many-to-one, and many-to-many cases without turning the result into a black box.
Simple exact matches miss real-world files: a payment that clears two invoices, a settlement that covers three lines, or a vendor name that is spelled two ways. Those cases need explicit rules, not a hidden model deciding the books.
How it works
Step 1
Enable only the rule families this file pair needs.
Step 2
Define which columns identify a row and how far amount or date may differ.
Step 3
The engine applies the configured rules in a deterministic order.
Step 4
Open a 1:N or N:1 group and see which rows were combined and what remainder remains.
Step 5
Reviewers can see the rule type. AI does not silently rewrite the match.
What this workflow is designed to do with uploaded files.
Identical keys clear as a pair. Use this when references are clean.
Allow a configured amount or date difference when the books are close but not identical.
Catch name or reference spelling differences without rewriting the source file.
Record a remainder instead of forcing a false full match.
One invoice can clear against several payments that add up to the invoice amount.
Several invoice lines can clear against one settlement.
Related groups can match when a simple pair is not enough.
The rule stays visible. A person can open the group and see the arithmetic that produced the match.
Example 1
Invoice 1000. Payments 600 and 400. The group clears as one invoice to two payments.
Example 2
Three invoice lines settle in one payment. The group clears as many lines to one settlement.
Sample preview
Sample groups. Amounts are fictional.
One open invoice. Two cash lines that add to the invoice total.
1:N match. Both payments attach to invoice 1000. No remainder.
Line A 250, line B 250, line C 500. One receipt of 1,000.
N:1 match. The three lines attach to the settlement.
Invoice 2,400. Receipt 1,800.
Partial match. Remainder 600 stays open for review.
Complex matching is useful only if a reviewer can reconstruct it. Deterministic rules remain visible and auditable. The engine does not invent a match that the configured rules would not produce.
Continue with the workflows that usually sit next to this one.
No. Matching is a configured, deterministic engine. AI may later explain a group; it does not force the match.
Yes. Enable only the rule families the file pair needs. Unused types stay off.
Partial matches keep the leftover amount visible so it can be reviewed instead of being forced to zero.
Start with exact keys, then add grouped matching only where the files need it.