How do I turn a PDF into Markdown for an AI?
Convert the PDF's text layer to Markdown rather than pasting raw PDF text, because Markdown keeps headings, lists and tables readable. Only text that already exists in the file comes out — a scanned page has no text to convert and produces nothing useful.
Why Markdown and not plain text
Extracted PDF text arrives as a stream of lines with no structure. A model reading it cannot tell a heading from a clause, or a list from a paragraph that happens to start with a number.
Markdown restores the cheap signals — headings, bullets, table rows — which is most of what a model needs to answer questions about the document.
The conversion
- Open the PDF in the converter. It reads the text layer, page by page.
- Download the Markdown, or copy it.
- Skim the first page. If it is empty, the PDF has no text layer and the file is a scan.
- Paste into the model, or use the optional summary here. Only the first 30,000 characters are sent if you choose that route.
What comes out well, and what does not
Prose, numbered clauses and simple lists convert cleanly. Multi-column layouts interleave — two columns become alternating lines, which reads as nonsense to a model.
Tables are the weak point. A scanned table has no cell boundaries; even a born-digital table often loses its column structure. For a financial table, check the output before you rely on it.
Headers and footers repeat on every page, and they arrive as repeated noise in the Markdown.
Where this stops
No OCR. A scanned document produces an empty or near-empty result, because there is no text in the file to extract. The contract scanner can read page images, but that is a different job and a different page.
No guarantee. The converter infers structure from character positions; it cannot know that a line was meant as a heading when the PDF never said so.
Is my document sent to an AI?
Not by the converter. It runs in the browser tab. The optional written summary is the one feature that sends text, it is opt-in, and it sends only the first 30,000 characters.
Why is the output empty?
The PDF is a scan with no text layer. Convert it to Markdown is not possible without OCR, and OCR is not offered here.
Can I use the Markdown commercially?
The Markdown is derived from your own document. Whether you may use the content is the document's licence question, not the converter's.
This page is general information, not advice for your document. The tools do what they say on their own pages, and every limit is printed there.