How Nordflow captures invoice data

When a document arrives in Nordflow, the AI data-capture engine processes it automatically and writes the extracted values onto the document.

What gets extracted

The engine captures the following fields from every invoice:

Supplier — matched against your master data; flagged as a new creditor if no match is found.
Document number — the supplier’s own invoice number.
Dates — document date, due date, and payment date.
Amounts — total, net amount, and VAT.
Currency — shown alongside the total amount.
Payment information — bank account, IBAN/SWIFT, and payment reference (FIK, KID, or giro depending on country).
VAT registration number (CVR/MVA) for the supplier.
Line items — description and amount per line (where line capture is active).

These fields populate document details when an approver opens the invoice.

Confidence scores

Every extracted field carries a confidence score. A low score means the engine was uncertain about that field. Admins and consultants can see these scores on document details; approvers see only the pre-filled values. Low-confidence fields are reviewed by the human validation tier before the document reaches approvers.

Master data cross-check

After extraction, Nordflow cross-checks the captured data against your ERP master data. If the captured bank account differs from the stored account for that supplier, a warning is raised. If the supplier does not exist at all, the document is flagged for new-creditor handling.