Turn scans into editable text
On-device OCR for scanned PDFs, images and screenshots.
PDF to Markdown
Text-based PDFs — anything exported from Word, LaTeX, a browser or a report generator — are converted to Markdown with headings inferred from the document's own font hierarchy.
Drop a .pdf file here
Word, PDF, Excel, PowerPoint & more
Choose a fileor drag and drop it here
ANYMARKDOWN FOR iPHONE & iPADYour next step.
A full Markdown workspace.
Enjoy the web converter? Take AnyMarkdown with you. Scan documents, refine your writing and keep your library close — wherever you work.
Get AnyMarkdown on the App StoreFree to try · In-app purchases for Pro · iPhone & iPad
Edit, preview, make it yours
Syntax highlighting, a visual table editor and live preview.
A library that goes with you
Organize your documents, sync with iCloud and export Markdown, HTML or PDF.
How it works
- Drop your file on the box above, or click to choose one.
- The converter loads once, about 1.8 MB, and runs in this tab. Your file is not uploaded.
- Copy the Markdown, or download it as a
.mdfile.
What survives
- Body text in reading order, with column layouts linearised
- Headings inferred from relative font size and weight
- Lists recovered from their bullet and number glyphs
- Ligatures and hyphenated line breaks rejoined into whole words
What does not
- Scanned and photographed PDFs contain no text layer. These are detected and the affected page numbers are reported — this converter does no OCR.
- Complex multi-column magazine layouts can interleave when the PDF's internal text order disagrees with the visual order.
- Tables in PDFs have no structural markup to read, so they come out as text rather than as pipe tables.
Questions
Can it read a scanned PDF?
No — a scan is a picture of a page, and extracting its text requires OCR, which this converter deliberately does not do. It tells you which pages need OCR instead of returning an empty result. The AnyMarkdown iOS app does OCR on-device if you need it.
Why did my table come out as loose text?
PDF has no notion of a table. It stores glyphs at coordinates, and a visual table is just text that happens to line up. Recovering it reliably needs layout analysis beyond what a text extractor can do.
Is my file uploaded to a server?
No. The converter is a WebAssembly build of the anydoc engine that runs inside your browser tab. Your file is read into memory, converted, and discarded when you close the page. Nothing is transmitted, stored, or logged.