C↗ CatalogTraceWorking beta · no charges

PDF → CSV / JSON

PDF data you can
trace back.

Extract lines and columns from a PDF catalog. Every result keeps its source page and position; uncertain data stays visible for review.

01 Your document

Up to 10 pages. For longer PDFs, specify which pages to extract.

Explicit columns (advanced)

Each start position is a fraction of the page width, from 0 to 1. Without explicit columns, we try to recognize simple headers; inferred field mapping is not guaranteed.

Your PDF is processed in your browser: this website does not upload it or store it on a server. No paid AI services are used.

02 Traceable results

You have not processed a document yet.
≡

Your data will appear here

No sample results are preloaded.
Choose a real PDF to get started.

Built for agent workflows

The same engine runs in Node.js. The local API accepts a PDF and returns JSON with coordinates, original values and warnings. It does not execute instructions found in the document.

Integration and limits →

What this beta does not do

It does not include OCR, guess currencies or product references, or certify an ERP import. Headers, notes and split lines remain available for review.