AI-Ready Document Extraction

Convert PDF, DOCX, XLSX, PPTX, images (OCR) and 76+ formats into clean, structured Markdown ready for LLM prompts, RAG pipelines, and AI workflows. Powered by Kreuzberg WASM, 100% client-side.

Preparing engine...

The extraction engine runs entirely in your browser. It only needs to load once — next time it will be instant.

PDFDOCXXLSXPPTXImagesZIP+70 more

Document Extraction Guide

How It Works

This tool uses Kreuzberg, a Rust-based document intelligence engine compiled to WebAssembly. It runs entirely in your browser, no upload, no server.

Kreuzberg uses format-specific parsers (PDFium for PDFs, native parsers for Office formats, Tesseract WASM for OCR) and reconstructs:

Headings (H1-H6) from font size and style analysis
Tables as Markdown pipe tables with headers
Lists, bold, italic, and hyperlinks
Metadata: title, authors, dates, page count
OCR text from images and scanned PDFs

Supported Formats

Documents: PDF, DOCX, DOC, ODT, RTF, EPUB
Spreadsheets: XLSX, XLS, ODS, CSV
Presentations: PPTX, PPT, ODP
Images (OCR): PNG, JPG, TIFF, BMP, WebP
Email: EML, MSG
Archives: ZIP, TAR, 7Z (extracts contents)
Text / Markup: HTML, XML, JSON, YAML, Markdown, plain text

Use Cases

RAG pipelines: feed clean, structured text to LLM retrieval-augmented generation systems
Documentation migration: convert legacy documents to Git-based docs (Docusaurus, MkDocs)
Data extraction: pull tables and structured data from PDFs and spreadsheets
LLM context: structured Markdown yields better AI comprehension than raw text
Accessibility: extract text from images via OCR

Limitations & Tips

Multi-column PDFs: two-column layouts may interleave; single-column works best
Images in documents: enable "Extract images" in options to pull embedded images
First load: the WASM engine (~23 MB) is downloaded once and cached by your browser
Max file size: up to 512 MB, limited by browser memory

For best results with PDFs, use documents from word processors (Word, LaTeX, Google Docs).

Frequently Asked Questions

Is my file sent to any server?

No. All processing happens locally in your browser using WebAssembly. Your file never leaves your device. No uploads, no API calls, no storage. You can verify this in your browser's Network tab.

Why does it take a few seconds to load?

The Kreuzberg WASM engine (~23 MB) needs to be downloaded and compiled on first visit. Your browser caches it, so subsequent visits load instantly.

What file formats are supported?

Over 76 formats including PDF, DOCX, XLSX, PPTX, ODT, RTF, EPUB, CSV, HTML, XML, JSON, YAML, images (PNG, JPG, TIFF, BMP, WebP) with OCR, emails (EML, MSG), and archives (ZIP, TAR, 7Z).

Does it support scanned PDFs and images?

Yes. OCR support is available via Tesseract WASM for image-based pages and image files. It can extract text from screenshots, scanned documents, and photos.

Does it extract tables?

Yes. Tables in PDFs, Word documents, and spreadsheets are detected and converted to Markdown pipe tables with proper headers and cell alignment.

Can I convert between document formats (e.g. DOCX to PDF)?

No. This tool extracts content as Markdown from any supported format. It is a document-to-Markdown converter, not a format-to-format converter.

What is the maximum file size?

Files up to 512 MB are supported. Processing happens in your browser's memory, so very large files may take longer on devices with limited RAM.

Is it free? Do I need an account?

Completely free, no account needed. No hidden charges, no premium tiers, no usage limits.

JSON

Validate, format & convert JSON

UUID

Generate UUID v4, v7 & ULID

Hash

Generate MD5, SHA & BLAKE hashes

Encode

Encode & decode Base64, URL & Hex

AI-Ready Document Extraction

Document Extraction Guide

How It Works

Supported Formats

Use Cases

Limitations & Tips

Frequently Asked Questions

You Might Also Like

JSON

UUID

Hash

Encode

Document Extraction Guide

How It Works

Supported Formats

Use Cases

Limitations & Tips

Frequently Asked Questions