Every allocation.
A checkable explanation.
Match partial payments across invoices. See when the data permits more than one answer, and export a certificate that can be verified independently.
1. Define the batch
Candidate pairs share a normalized counterparty and currency, within your date window. Amounts are exact decimal strings.
Import CSV files instead
Required headers: id,counterparty,currency,amount,date. Quote fields containing commas. Imports replace that side of the batch. Only use fictional or authorized records.
2. Review the allocation
unallocated cash and an ambiguity witness.
| PAYMENT | INVOICE | AMOUNT | DATE GAP |
|---|
Inspect the proof certificate
Cut capacity equals allocated flow, certifying maximum allocation. All residual reduced costs are nonnegative under the supplied node potentials, certifying minimum date-gap cost. The separate verifier rebuilds the graph from the original input.
3. Verify an exported audit
The verifier accepts an audit file, rebuilds constraints, and checks the certificate without invoking the optimizer. Changing an allocation or the objective invalidates the proof.
A precise model, with clear limits
Maximize allocated minor units, then minimize amount-weighted absolute date gaps, independently per currency. Partial allocations are allowed. Identity matching is exact after whitespace/case normalization.
An optimal allocation is not proof that a payment truly belongs to an invoice. Dates do not establish business intent. No automatic posting, banking connection, fraud detection, FX, refund or legal determination occurs.
Product and monetization plan
Designed for bookkeepers investigating ambiguous small-business batches. The proposed model is a free bounded sandbox, then subscription access to larger batches, saved workspaces and accounting-system export. Billing, organizations, integrations and paid plans are not implemented. There are no customer, revenue or adoption claims.