Meet Docling: The “Pandas” for Document AI
summary
Docling is an open‑source Python package that converts PDFs, DOCX, PPTX, HTML, images and Markdown into a structured DoclingDocument and exports to Markdown or JSON, and it already has 25,000 GitHub stars. The demo shows a DocumentConverter + CLI, tables exported directly to pandas DataFrame, OCR optional, picture enrichment via local or hosted VLMs (e.g. Ollama) producing visual descriptions and page bounding boxes for visual grounding. Runtime for the full pipeline averages ~0.7s per page; a hybrid chunker yields 67 text / 5 table / 23 image chunks and feeds LlamaIndex/LangChain agentic RAG workflows via an MCP server.
AI-written summary · about
chapters (9)
- 0:00Docling introduction
- 5:22First steps with Docling
- 12:28Tables and figures in Docling
- 21:41Docling conversion pipeline
- 26:42Enhanced document chunking
- 35:35Multimodal apps with Docling
- 44:01Docling in Agentic AI
- 57:01Q&A
- 1:00:35Performance optimization
related talks
35mDocling: Get your documents ready for gen AI
01h 25mLearn to Unlock Document Intelligence with Open-Source AI
01h 00mComplex Data Ingestion with Open Source AI
01h 19mLearn to Unlock Document Intelligence with Open-Source AI
29mConquering PDFs- document understanding beyond plain text
24mTeaching PyTorch To Read Your Worst PDFs With Docling
31mConquering PDFs: document understanding beyond plain text
26mAdvanced Document Parsing for AI Workflows