How it works

1. Your browser loads the PDF with PDF.js — the file never leaves your device.
2. Extract2MD analyses layout: headings, paragraphs, lists, tables, images and links.
3. Scanned pages can be read with optional OCR (Tesseract.js) in the browser.
4. You get clean Markdown, plus text, JSON, tables, images or ZIP when you need them.

The main product is clean PDF to Markdown. ChatGPT, Claude, RAG and Obsidian exports are secondary workflows from the same engine. We structure and clean existing content — we do not rewrite it with an LLM.