Docling: Get your documents ready for gen AI
summary
Docling is an open source Python library that converts PDFs, DOCX, PPTX, HTML, images and Markdown into a lossless DoclingDocument and easily exportable HTML/JSON/Markdown, claiming 30,000+ GitHub stars and ~2,000 contributors while joining the Linux AI & Data Foundation. It reconstructs reading order, complex tables and figure–caption links, offers OCR and VLMs (SmolDocling/Granite), integrates with LlamaIndex/LangChain/LlamaStack, has a CLI and docling-serve and reports 50x cheaper than large VLMs for scanned docs.
AI-written summary · about
chapters (15)
- 0:00Docling project intro
- 1:24What Docling does
- 4:12Key features demo
- 7:25Hidden prompts security
- 9:11Docling processing internals
- 12:03Vision language models
- 15:59Compact VLM training
- 18:30Granite Docling multilingual
- 21:39Doc tags for structure
- 24:34AI research standardization
- 25:19Two processing pipelines
- 27:26Document extraction agents
- 28:50Agent tools demos
- 29:12Langflow OpenRAG RAG
- 30:15Q&A
related talks
31mDocling: Get Your Documents Ready for Gen AI
01h 07mMeet Docling: The “Pandas” for Document AI
49mNext-Gen Documentation with AI
16mInteract with Your Technical Documentation via Generative AI
30mMaster 2.5 GB of unstructured specification documents with ease
23mDocumentation in the Age of AI: Why Docs Are a Product
28mA Retrieval Augmented Generation system to query the scikit-learn documentation
50mContinuous and Automated Documentation Docs As Code in Practice