The challenge
In accounts payable, every invoice must be matched against its purchase order and the goods receipt before approval. The matching is manual: open the invoice, find the purchase order, compare quantities, prices and terms, line by line. A batch of invoices consumes hours of work and, under closing pressure, the team prioritizes — some differences only surface after payment. And when a difference appears, reconstructing the context — what was ordered, what arrived, what was invoiced — takes longer than the matching itself.
The proposed solution
An automatic three-way matching flow between invoice, purchase order and goods receipt:
- AI-powered extraction of the invoice data, regardless of the supplier’s format.
- Automatic matching against the purchase order and the goods receipt: quantities, unit prices, totals and taxes.
- Difference detection, with the reason for each discrepancy identified.
- Prioritized approval queue: what matches flows straight through; exceptions reach the analyst with the context ready.
Expected results
The figures are preliminary targets, subject to the ongoing validation:
- Reconciliation per batch: from hours to minutes.
- 100% of invoices verified against their purchase order — no sampling, no volume-driven exceptions.
- The team spends its time resolving differences, not finding them.
Status
This concept is under validation with real accounts payable operations. The published metrics are design targets; we will confirm them with measured results before moving the case to production.