Convert PDF to JSON

Turn PDFs, scans, and images into structured JSON with AI: document fields and table rows as named keys, ready for a script, a database, or an API. No templates, no code.

A PDF to JSON converter is only useful if the JSON has structure. Dumping the text of a page into a single string gives you nothing a program can work with. ScanPilot reads each page with AI, works out what the document is, and returns its data as JSON: header fields like the invoice number and date as named keys, and every table row as an object in a list.

It needs no template and no parsing rules, so a new layout works on the first try, and scanned PDFs and photos convert the same way digital files do. Upload a document above to see the JSON for its first page free, with no signup. The same extraction downloads as Excel or CSV when a person, not a program, needs the data.

How to convert a PDF to JSON

Upload the file
Drop a PDF, a scanned page, or a photo (JPG, PNG, HEIC) into the box above.
The AI reads the page
It detects the document type, the header fields, and every table, and reads them cell by cell, including on scans.
Check the preview
Page one appears as a table in seconds, so you can see what the JSON will contain before downloading.
Download the JSON
Click JSON to save a .json file: one object per page, with fields and rows as named keys.

What the PDF to JSON file looks like

The file is an array with one object per page. Header fields sit under documentMetadata, table rows under entries, and numbers stay numbers, so you can load the file without cleanup. A typical invoice page:

[
  {
    "page": 1,
    "documentMetadata": {
      "Vendor__Name": "Acme Supplies Ltd",
      "InvoiceNumber": "INV-1042",
      "InvoiceDate": "2026-03-02"
    },
    "entries": [
      { "LineItems__Description": "Office chair", "LineItems__Qty": 4, "LineItems__Amount": 596.0 },
      { "LineItems__Description": "Desk lamp", "LineItems__Qty": 2, "LineItems__Amount": 58.5 }
    ]
  }
]

Field names come from the document's own labels, and a double underscore groups related fields (Vendor__Name, LineItems__Qty). For pages that are mainly a grid, such as a register or a printed form, you can instead keep the page exactly as laid out: each page then comes back as a grid, a list of rows with one string per cell. The free preview above uses this layout view.

Scans and images to JSON

Most PDF to JSON tools read the PDF's text layer, so a scan, a fax, or a phone photo returns nothing. ScanPilot runs AI OCR and structure detection together, which makes it an image to JSON converter as well: upload a JPG or PNG of a receipt, a form, or a handwritten table and get the same structured output. Handwriting is read far more reliably than by classic OCR; see handwriting to text for a measured test.

When JSON is the right format

Feeding a script or an app: load the file with one line of code instead of writing a PDF parser
Importing into a database or a no-code tool that accepts JSON
Keeping header fields and line items together, which a flat CSV cannot do
Testing extraction on real documents before building your own pipeline

If the data is going into a spreadsheet instead, download the same extraction as Excel or open it in Google Sheets.

Encrypted and never used for training

Files are encrypted in transit, processed only to produce your result, and never used to train AI models.

Why convert with ScanPilot?

AI table extraction

The AI reads your document the way you do, so columns land where they belong instead of being guessed from character positions.

Scans included

OCR runs automatically, so scanned pages convert the same way digital PDFs do.

Try before you decide

The first page converts free with no signup, so you judge the quality on your own document, not a marketing sample.

Private and encrypted

Files are encrypted in transit, processed only for your conversion, deleted afterwards, and never used for training.

Frequently asked questions

How do I convert a PDF to JSON?

Upload the PDF above and click JSON once the preview appears. You get a .json file with one object per page, holding the document fields and table rows as named keys. The first page converts free with no signup.

Do I need to write code or set up a template?

No. The AI works out the fields and tables on each document by itself, so any layout converts without templates, rules, or code.

Can it convert scanned PDFs and images to JSON?

Yes. AI OCR reads scans, faxes, and phone photos (JPG, PNG, HEIC), including handwriting, and returns the same structured JSON as for digital PDFs.

Are numbers exported as numbers?

Yes, in the standard field view quantities and amounts are JSON numbers, not strings. In the layout view every cell is kept as text exactly as it appears on the page.

Can it create the JSON file for ITR filing?

No. The JSON used for Income Tax Return filing in India is generated by the official ITR utility in its own format. This converter extracts data from documents, it does not produce ITR JSON.

Is there an API?

Not at the moment: conversion runs in the browser, and you download the JSON file. For large monthly volumes, the Enterprise plan on the pricing page offers custom volume and invoice billing.

Is it free?

The first page of any file converts free with no signup. Full documents need a ScanPilot plan.