← MDifying document to Markdown converter
A Project is only as good as what you put in it. Uploading twelve PDFs straight in works, but you cannot see what the model actually reads, and you cannot fix it when a table comes out as a paragraph. Converting first gives you files you can read, edit and trim before they go anywhere.
Convert a file now — it is free and runs in your browser
Select up to twenty files together. They convert one after another and the list shows which are done — a file that fails does not stop the rest, it just says why in its own row.
One archive of .md files, named after the originals. Two files that shared a name are told apart by their source format — report.md and report (docx).md — so nothing is silently overwritten.
This is the step that pays. Delete the boilerplate, the cover pages and the legal footers; fix a table that came out crooked; split a long report into the two subjects it actually covers. Every one of those changes is easier now than as a retrieval setting later.
Plain text, one subject per file, no binary formats in the mix. The token estimate shown with each conversion gives you a rough sense of what you are spending before you commit a document to the Project.
You can, and for a clean, text-based PDF it is fine. The problem is the ones that are not: a scan with no text layer contributes nothing, and a two-column layout or a wide table can arrive scrambled without any sign that it did. Converting first makes that visible while you can still do something about it.
Twenty files per run and 50 MB per file on a computer. Both are limits of what a browser tab can hold, not plan tiers — run a second batch and nothing changes. Files are converted on your own machine, so a Project built from confidential material never sends that material anywhere, including to us.