Core Invoice

Invoice OCR, end to end

Optical character recognition turns a scanned invoice into structured data. The hard part isn't reading the characters — it's knowing which characters mean total, which mean line price, and which mean VAT. Core Invoice is built around that distinction.

What "invoice OCR" actually does

Traditional OCR converts pixels to text. Invoice OCR goes further: it identifies the fields on an invoice — supplier name, invoice number, date, currency, line items, taxes, totals — and outputs them as structured records you can post to your ERP.

Core Invoice extracts:

  • Supplier identity (name, tax ID, address)
  • Invoice metadata (number, date, due date, currency)
  • Line items (description, quantity, unit price, line total)
  • Taxes (rate, base, amount — per line and per invoice)
  • Totals (subtotal, total tax, total payable)
  • Payment terms and remittance details

From scan to posted entry

  1. Capture. Upload PDFs, drop them into a watched folder, snap a photo on mobile, or pipe them in via the API.
  2. Extract. Our OCR model reads the document and emits a structured JSON record.
  3. Review. Confidence scores flag anything the model isn't sure about. Your team approves or corrects.
  4. Post. The record lands in JDE, SAP, NetSuite, or wherever you keep your AP ledger.
  5. Reconcile. When the payment clears, Core Invoice matches it to the invoice automatically.

Why not just use generic OCR?

Tesseract or a cloud OCR API will turn pixels into text. They won't tell you which number is the total and which is the line-item subtotal. That's the work — and that's where the time savings live for AP teams processing thousands of invoices a month.

Specific to your ERP

If you already know which ERP you're posting into, jump to the integration deep-dive:

Ready to try it?

Sign in and upload your first batch