From manual invoice handling to a governed pipeline

The challenge

NOS processes tens of thousands of supplier invoices per month. Each one required operators to extract data, match it against purchase orders in SAP, validate business rules, and handle every exception by hand. The workflow spans multiple supplier systems, invoice formats, and business entities across subsidiaries, with more than 20 fields to validate per invoice. When mismatches occurred, operators investigated manually, often without clear visibility into why an invoice had failed or how frequently that pattern was appearing across the pipeline. There was no consolidated view of blocking reasons, no systematic way to prioritise exceptions, and no way to measure whether process changes were reducing failure rates over time.

The solution

DareData deployed GenOS Supervisor to automate the invoice treatment workflow end to end. Invoices are ingested automatically, extracted, validated against business rules, and matched to purchase orders in SAP without operator involvement on the standard cases. When a match is ambiguous, an AI inference step proposes the best candidate and routes the case to human review with the full context attached. When the pipeline cannot resolve a case, it surfaces with a structured explanation of exactly why.