PDF to Markdown
Pull the text out as Markdown, with headings and lists detected.
Everything runs inside your browser. Your files never leave your device.
Drop your file here
.pdf · they stay on this device
Ask about PDF to Markdown
Questions about what this tool does, which option to pick, or what it can and cannot handle.
The question you type here is sent to an AI provider to be answered — your files and whatever you put in the tool above are not, and the assistant cannot see them. Answers are generated and can be wrong. The tool itself is not guessing: it runs deterministic code on your device.
About the PDF to Markdown
Getting the text out of a PDF as Markdown is useful for moving content into a wiki, a static site, a note-taking application, or anywhere that wants structured text rather than a fixed page.
This extracts the text and infers structure from it — headings from relative font sizes, paragraphs from spacing, lists from leading characters — and produces Markdown.
The inference is the interesting part and the reason to expect a draft rather than a finished conversion. A PDF has no concept of a heading. It has text drawn at a position in a size, and a heading is simply text that happens to be larger and to have space around it. That heuristic works well on a report and poorly on a design-led document where size varies for visual reasons rather than structural ones.
What comes through reliably: paragraphs, headings in a conventionally structured document, simple lists, and the reading order for single-column layouts. What does not: tables, which lose their structure because a PDF table is lines and text at coordinates rather than rows and cells; multi-column layouts, where extraction follows the internal ordering rather than the visual one; footnotes, which land wherever they were drawn; and every image.
Scanned PDFs produce nothing at all, and it is worth understanding why. A scan is a picture of a page, and there is no text in it to extract. That requires OCR, which is a different operation with different failure modes.
Expect to edit the result. This does the tedious ninety per cent — getting the words out in the right order with the paragraphs intact — and leaves the judgement to you.
How to use it
- 1Drop in the PDF.
- 2Read the extracted Markdown, checking the heading levels the tool inferred.
- 3Rebuild any tables by hand, since a PDF table has no structure to extract.
- 4For a scanned document, run OCR first — there is no text in a scan to extract.
Questions
- Why did my tables come out as a jumble?
- Because a PDF table is lines and text placed at coordinates, with no rows or cells recorded. Extraction can recover the words but not the grid, so tables have to be rebuilt by hand.
- Nothing was extracted from my document.
- It is almost certainly a scan — a picture of a page with no text in it. OCR is the only route to text from a scanned document, and it is a separate operation.
- How does it decide what is a heading?
- By relative font size and the spacing around the text, since a PDF records no structural information. That works on conventionally formatted reports and poorly on design-led documents where size varies for visual reasons.
- Why is the text in a strange order?
- Extraction follows the document's internal ordering, which for multi-column layouts, sidebars and pull quotes may not match the order a person reads in. Single-column documents come out reliably.