Extractor
Extract a document's text without the usual mess.
Most extractors return hard-wrapped fragments and scrambled tables. This runs the parser our video pipeline uses, so structure survives — and it reads DOCX and PPTX too.
- Tables come out as Markdown tables, not cells in reading order.
- Word and PowerPoint files work, not just PDF.
- Markdown or plain text, both downloadable.
Why most PDF text extraction comes out unreadable
A PDF does not store paragraphs. It stores glyphs at coordinates, and the line breaks you see are layout decisions, not sentence boundaries. Naive extraction returns one line per visual line, words split across hyphens, headings indistinguishable from body copy, and tables flattened into a stream of cell contents.
This tool runs the extractor our production video pipeline depends on. It reconstructs paragraphs, detects headings, and emits tables as Markdown tables — because the pipeline needs that structure to build scenes, so it had to be solved properly.
- Markdown — headings and tables preserved, as extracted.
- Plain text — the same content with Markdown syntax stripped.
- Word count, character count, and estimated reading time alongside.
The one tool here without a model in it
Extraction is parsing, not generation, and dressing it up as AI would be a lie. What makes this version better than a browser-side script is not intelligence — it is that a real parser with real table handling runs on a server, and can read formats JavaScript in a tab cannot.
It is still behind the same sign-in as everything else here, because it is the same authenticated endpoint.
What people usually do next
Extracting text is rarely the goal. It is step one of something else — feeding a model, rebuilding a deck, quoting a source, or getting a document into a format people will actually engage with.
If it is the last of those, a text file will not help. A report nobody finished reading as a PDF is a report nobody will finish reading as a .txt. The format that changes the outcome is video.
Frequently Asked Questions
Does it preserve tables?
Yes, as Markdown tables. That is the main thing separating this from a quick browser-side extractor, which returns table cells as a flat run of text.
Which file types work?
PDF, DOCX, PPTX, Markdown, plain text, and VTT. Scanned PDFs do not, because they contain no text layer.
Why does a text extractor need a login?
Because the parsing runs on our servers rather than in your browser, on the same authenticated endpoint the other tools use. The account is free.
Can I use the extracted text commercially?
That depends on the document's licence, not on us. We impose no terms on the output.