No. Conversion runs entirely in your browser with PDF.js. The file never leaves your device, which makes the tool safe for confidential documents.
Yes, completely free with no account, limits or watermark. An optional API and advanced features are planned for the future.
Text-based PDFs — anything exported from Word, Google Docs, LaTeX, or a web page. Scanned PDFs are images and require OCR, which is on the roadmap.
By clustering font sizes. The largest text becomes an H1, the next size an H2, and so on. You can turn this off in the options if your PDF has unusual typography.
Yes. Tables are reconstructed by detecting shared column positions across rows and emitted as GitHub-flavored Markdown tables. Very complex or nested tables may need light cleanup.
Standard, GitHub-flavored Markdown: headings, bold, italic, inline code, bullet and numbered lists, and pipe tables.
V1 inserts a placeholder where images occur so you keep document flow. Full image extraction is planned.
Markdown keeps document structure while removing layout noise, which makes prompts clearer and more token-efficient and improves RAG retrieval.
Once the page has loaded, conversion works without a connection, because all processing is local.
There is a soft 50MB guidance limit. Very large PDFs can be slow because all the work happens in your browser, but there is no server cap.