Convert scanned invoices to structured JSON & Excel

Stamps, handwriting, and dense line-item grids make invoices painful to digitise by hand. Scan Hero reads scanned PDFs, photos, and PNG/JPG invoices and returns clean header fields, an itemised table, and reconciled totals — as JSON for your ERP or as a multi-sheet Excel workbook.

Every conversion runs on Anthropic’s Claude models, so the output understands context: it ignores AP scribbles, normalises vendor and tax fields, and keeps line items aligned. Upload a file or call the REST API — no OCR pipeline to build.

Input: PDF (scan), PNG, JPG, DOCXOutput: JSON (.json), Excel (.xlsx), Markdown tables

Example: Scanned invoices → structured tables & JSON

Stamps, handwriting, and dense line-item grids become clean header fields, item tables, and totals—ready for ERP, AP workflows, or automation.

Sample scanned commercial invoice
Sample scanned commercial invoice

Invoice header

FieldValue
VendorGlobalTech Solutions LLC
Invoice numberINV-2024-07891
Invoice date2024-05-03
Due date2024-06-02
Customer IDCUST-88919
PO numberPO-55421
Bill toAcme Corporation — Shelbyville, KY
Ship toAcme Corporation — 150 Logistics Way
SalespersonMichael T.
Ship viaFedEx Ground
Tracking1234 5678 9012
Shipped date2024-05-02
CurrencyUSD

Line items

LinePart / descriptionQtyUOMUnit priceDiscountAmount
1GTX-1000 Server Rack (42U, PDU, shelf, cable mgmt)2EA1,250.000%2,500.00
2SFP-10G-SR Module (10GBase-SR, 850nm, 300m)10EA85.505%812.25
3CAT6A-PLENUM-BLUE (1000 ft box)3BX195.000%585.00
4INSTALL-SVC (on-site hourly)8HR125.000%1,000.00
5SHIP-HANDLING1EA75.000%75.00

Totals

Amount (USD)
Subtotal4,972.25
Discount-40.61
Shipping75.00
Sales tax (KY 6%)298.00
Total due5,304.64

Sample output

{
  "document_type": "invoice",
  "invoice_number": "INV-2024-07891",
  "invoice_date": "2024-05-03",
  "due_date": "2024-06-02",
  "vendor": {
    "name": "GlobalTech Solutions LLC",
    "address": "7447 Industrial Parkway, Suite 200, Plainfield, IN 46168"
  },
  "bill_to": { "name": "Acme Corporation", "city": "Shelbyville", "state": "KY" },
  "line_items": [
    { "line": 1, "sku": "GTX-1000", "qty": 2, "amount": 2500.0 },
    { "line": 2, "sku": "SFP-10G-SR", "qty": 10, "amount": 812.25 }
  ],
  "subtotal": 4972.25,
  "tax_amount": 298.0,
  "total_due": 5304.64,
  "currency": "USD"
}
Tip: Handwritten AP notes and stamps are ignored; line items and totals are normalized. Choose JSON or Excel output, or refine field names with AI.

How it works

  1. 1

    Upload your file

    Drag it into the browser or send it through the REST API. See the credit cost before you convert.

  2. 2

    Claude reads it

    Anthropic’s Claude models extract and restructure the content — context-aware, not just raw OCR text.

  3. 3

    Download clean output

    Get exactly the format you chose, and refine it with an AI prompt in seconds if needed.

Frequently asked questions

Can Scan Hero extract data from a photo of an invoice?

Yes. Upload a PNG, JPG, or HEIC photo or a scanned PDF. Claude Vision reads the layout, normalises the vendor, dates, line items, and totals, and returns structured JSON or an Excel workbook.

Does the JSON match my own schema?

You can define a Template describing the exact fields you need, then convert against it to get JSON validated to your shape — ideal for direct ERP or accounts-payable import.

How are handwriting and stamps handled?

Handwritten AP notes and stamps are ignored unless they are part of the line-item data. The structured output focuses on the printed invoice content.

More conversions

Browse all conversion workflows, read the guides, or see the examples gallery.

Start converting in seconds

100 free credits on sign-up. No code, no card, no commitment.

Get started free