HomeToolsMCPHow It WorksStoriesPhilosophyArchitectureStar on GitHub
All Tools
D
DataFreeOpen Source

DOCLING

Get your documents ready for gen AI

MIT

ABOUT

Documents (especially PDFs, scans, and mixed-layout files) are messy and unstructured, making them hard for AI systems to consume accurately. Docling converts complex documents into clean, structured representations (Markdown, JSON, HTML, DocTags) with accurate extraction of tables, formulas, reading order, images, and OCR text, eliminating brittle custom parsers and enabling reliable downstream AI, RAG, and agentic workflows.

INSTALL
pip install docling

INTEGRATION GUIDE

1. RAG pipeline ingestion — chunk and convert PDFs, DOCX, PPTX for vector stores (Milvus, Weaviate, Qdrant) via LangChain, LlamaIndex, or Haystack integrations. 2. Structured data extraction — parse financial reports (XBRL), patents (USPTO), and scientific articles (JATS) into machine-readable structured data. 3. Agentic AI and assistants — equip Cursor, MCP-based agents, and other AI assistants with document-reading skills using the Docling MCP server and agent skills. 4. Document conversion and OCR — convert scanned PDFs and images to accessible Markdown, HTML, or JSON with table structure recovery and reading order detection. 5. Audio and video transcription processing — parse WebVTT, WAV, and MP3 with ASR models for downstream AI workflows.

TAGS

document-processingpdf-parsingocrraggen-ainlppythonibmllmdata-extractionmcpagentic-ai
Docling — AI Tool | Agentic AI For Good