Upload a PDF, a Word file, or a scan and get a clean .md file. OCR reads scanned pages, and headings and tables come out as real Markdown.
Digital PDFs with selectable text give the cleanest results: structure is read directly from the file. Scanned PDFs and photos rely on OCR, which handles printed text well and stumbles on messy handwriting. Word documents (.docx) keep their heading hierarchy. Plain text and existing Markdown pass through with light cleanup.
Markdown is what note tools and AI pipelines actually want. Obsidian and Notion import .md files directly. If you are building a RAG pipeline or preparing training data for an LLM, Markdown keeps the heading structure that chunking depends on, without the layout noise PDFs carry around.
Yes. Scanned pages go through OCR, so a photographed contract or an old scanned book chapter comes out as editable Markdown text. Free accounts convert the first 25 pages of a scan.
Yes. PNG, JPG, WEBP, and TIFF images are read with OCR and returned as Markdown text. Screenshots of tables work too.
The free plan covers 3 downloads per day with files up to 10 MB and 25 OCR pages per document. You need an account with a real email to download results.
Headings become # heading levels and tables become Markdown pipe tables. The output keeps the document's structure rather than dumping plain text.