PDF to Markdown
Text with page breaks kept, for notes and for feeding an LLM.
or drop it here
Nothing is uploaded. Your file never leaves this tab.
What PDF to Markdown is
The same glyph extraction as plain text, with each page preceded by a heading. The page break is the one structural boundary a PDF genuinely records, so it is the only structure that can be reproduced without inventing anything.
What this tool does that others do not
Page headings as the navigable structure
They make the output usable in notes, and they give a language model a reliable way to cite where something came from — which is the most common reason people want a PDF as Markdown at all.
No invented structure
Headings inside the page, bold runs, lists and tables are not recovered. A heading in a PDF is text that happens to be larger; there is no marker saying "this is a heading". Guessing from font size works on documents produced by one template and fails on everything else, which produces confident nonsense.
How to use it
Run it and check the page headings line up
They are the structure; everything else is body text.
Paste it wherever it is going
Notes, an issue, or a model prompt. The page markers survive all three.
Do not expect tables
A table in a PDF is ruled lines and positioned text with no recorded relationship between them.
Worked example
Real input, real figures.
Preparing a report for a language model
You give it
A 40-page PDF report.
You get
Markdown with 40 page headings and the text under each. The model can then cite a page number, which is the one landmark the source document genuinely contains — and the reason page markers are worth more here than invented heading levels.
Against a converter that infers structure
| Aspect | This tool | A structure-guessing converter |
|---|---|---|
| Headings | One per page, from a recorded boundary | Guessed from font size |
| Lists and tables | Not claimed | Claimed, frequently wrong |
| Failure mode | Plainly plain | Subtly wrong |
Guide
For notes, and for feeding a language model
The same extraction as PDF to text, with each page preceded by a heading so the structure of the document survives as something you can navigate. That makes it useful for pasting into notes, and useful for handing a document to a language model where page boundaries are often the most helpful landmark available.
Headings inside the page, bold runs, lists and tables are not recovered. Those are visual properties in a PDF rather than recorded structure, and inventing them from font sizes produces confident nonsense.
Questions people ask
Straight answers. Each one links directly: copy the address after opening it.
Are headings and lists detected?
No. A heading in a PDF is text that happens to be larger; there is no marker saying "this is a heading". Guessing from font size works on documents produced by one template and fails on everything else, so rather than produce Markdown that is subtly wrong we produce Markdown that is plainly plain.
Why page headings?
Because the page break is the one structural boundary a PDF genuinely records. It gives you something to navigate by, and when you paste this into a model it gives that model a reliable way to cite where something came from.
Will tables come through?
No. A table in a PDF is ruled lines and positioned text with no relationship recorded between them. Recovering it needs ruling-line analysis that has no browser-side implementation.
Nothing leaves your device
Your file is read into this tab and worked on there. It is never uploaded, which is why there is no queue, no daily limit and no account, and why the tool still works with the network switched off.
All 15 PDF tools, and what we do not build