Perfect for notes, docs, and LLMs. Headings, tables, lists, and links are preserved automatically — and the whole conversion happens inside your browser.
or choose a file from your device
It loads directly into the page using your browser's own PDF engine — no network request involved.
Font sizes reveal headings, bullets become lists, aligned columns become tables, and link areas are matched back to their text.
Review the Markdown, tweak it if you like, then copy it into your notes or save it as a .md file.
Contracts, research drafts, internal docs — your content stays on your machine instead of passing through a third party's server.
Markdown keeps structure with far fewer tokens than raw PDF text, so models read your documents more accurately and cheaply.
Processing happens in your browser's memory, so there's no upload cap, account, or paywall for larger files.
No. The conversion runs entirely in your browser using pdf.js. Your file is never sent to a server, and it works offline once this page has loaded.
Headings (detected by font size), bold and italic text, bullet and numbered lists, simple tables, code-style monospace blocks, and clickable links. Images are not extracted.
Version 2.0 includes built-in OCR. When no selectable text is found, choose a language and press “Run OCR”. The OCR engine and language data download once from a CDN — your PDF itself is still never uploaded. OCR is slower and can make recognition mistakes, so proofread the result.
On-device OCR for scans, smart two-column reading order, headings guided by the PDF's own bookmarks, a live rendered preview, YAML front matter, an LLM token estimate, and remembered settings.
Tables with clearly aligned columns convert well. Merged cells, tables without a consistent grid, and cells that wrap onto several lines may need a quick manual fix — the output box is editable for exactly this reason.
With Smart columns on, the tool detects a clear gutter between two columns and reads the left column before the right. Three or more columns, sidebars, and text boxes can still come out in an odd order — turn Smart columns off if a page looks wrong.
No. Remove the password in your PDF viewer first (open it, then save or print a copy without protection) and convert that copy.
There's no limit set by this site. Because your device does the work, very large PDFs (several hundred pages) depend on your available memory and may take a while.
GitHub-flavored Markdown: ATX headings (#), pipe tables, fenced code blocks, and standard links. It works in Obsidian, Notion, GitHub, VS Code, and most LLM tools.