Product · Documents
PDF, Word or Excel to Markdown.
Convert office documents to clean Markdown. Convert ebooks. Convert PDFs. You can also get retrieval-ready chunks. It is the same tool you use for web pages.
See it work
Try it from wherever you work.
1webfold convert report.docx
2webfold pdf report.pdf
3webfold chunk report.pdf --size 4000 --jsonl
What you get
What Documents does.
Many formats
DOCX, DOCM, PPTX, XLSX, XLSM, XLSB, ODT, ODS, ODP, EPUB, RTF (as plain text for now), CSV and HTML, plus PDF and legacy XLS through an adapter.
OCR when there is no text layer
Scanned PDFs fall back to Poppler and Tesseract, if you have them installed.
Chunks ready for retrieval
Split any document into chunks and write them to JSONL, Parquet or Iceberg.
Same tool as the web
One binary for pages and files, so a pipeline does not need two converters.
How it works
Three steps.
- 1
Detect
Identify the format and route it to the right converter.
- 2
Convert
Native extraction first; OCR only when a page has no text.
- 3
Chunk
Optional retrieval-sized chunks.
Limits up front.
Know the limits
- Conversion success is measured, not fidelity. Math, embedded objects, merged cells and EPUB reading order are not scored.
- PDF extraction was scored on public test sets, and Webfold trails liteparse on tables and headings. See the PDF extraction benchmark.
- OCR needs
pdftoppmandtesseractinstalled.
Keep reading.
FAQ
Questions.
Is there a size limit?
On your machine, limits depend on the machine; run untrusted files with restricted filesystem access. The hosted endpoint accepts up to 25 MiB per file, sent in the request.
Use Documents with Webfold.
Run the command-line tool today. Join the waitlist for the hosted API.