WordGPT

Convert an invoice to JSON, field by field

Turn a PDF or scanned invoice into structured JSON with supplier, line items, tax and totals. Invoice numbers and amounts are kept as printed.

or drop a PDF, Word or text file here (up to 5 pages, 2 MB)

This preview shows the first 350 words and does not keep your file.

Need a different result? Turn anything into anything: drop any file, pick what you want out of it.Try it →

An invoice mixes supplier details, a customer address, a line-item table and totals on one page, so copying it into another system by hand is slow and amounts are easy to mistype. This page turns the invoice into one JSON object, with the supplier, customer, line items and totals grouped as nested fields. Invoice numbers, dates, amounts and tax stay exactly as printed.

What you get

Nested structure

Supplier, customer, line items and totals are grouped the way the invoice groups them, using snake_case keys and a list for the line items.

Amounts and dates as printed

Invoice numbers, dates, tax and totals are copied as they appear on the page. Nothing is reformatted or recalculated.

Null instead of guesses

If the invoice does not show a value, such as a PO number or a tax ID, that field is set to null rather than filled with a guess.

How it works

  1. Step 1

    Paste the invoice text, or drop the PDF or Word file into the box.

  2. Step 2

    Read the JSON preview, which shows the first part of the result.

  3. Step 3

    A free account gives you the whole JSON and a .json download, handles scanned and longer files and batches, and lets you ask for your own field names or a schema your tool expects.

Who uses it

  • Accounts payable teams

    Get supplier invoices into a structured form that an import tool or accounting system can read, without retyping each line item.

  • Developers

    Use a real invoice as a starting point for a JSON shape, then build a parser or test data around it.

  • Bookkeepers

    Check the subtotal, tax and total in structured form against the printed invoice before posting it.

Questions

What does the JSON look like for an invoice?

It is one object with nested groups such as supplier, customer and totals, plus top-level fields like invoice_number, invoice_date and payment_terms. Line items form a list, one object per row. Any field the invoice does not show is set to null.

Are amounts and dates changed when converted?

They are kept as printed. A date written as 03/03/2025 stays 03/03/2025, and £268.08 stays as shown, with no conversion to another date format or number format. Results can contain mistakes, so check the JSON against the original invoice when the figures matter.

Can it read scanned invoices?

Scanned invoices need a free account because they require OCR, which reads text from an image. The preview reads up to 5 pages of a PDF (2 MB) or about 6,000 characters of text, and it does not keep your file.

Can I change the field names to match my system?

With a free account you can ask for your own field names or a schema your tool expects, so the output matches what your import process needs.

Related tools

See it on real examples

Three actual runs of this tool. Open one to see the input and the whole result.