Build AI-Ready Search: Integrating Docling with OpenSearch for Advanced RA...
summary
Docling converts PDFs, DOCX and PPTX into a lossless, structured document format, preserves layout/tables/images and runs locally for privacy; a demo parses a technical report into 40+ chunks, embeds them with a 6B embedding model and serves vectors into OpenSearch 3.5, enabling RAG retrieval and citeable source spans. Default VLM is granite 258M (swappable), pipelines run on CPU or GPU, a Dockling Java ingest plugin for OpenSearch is imminent and the end-to-end demo works.
AI-written summary · about
chapters (10)
- 0:00AI-ready search with Docling
- 2:27Private data challenges with LLMs
- 6:00PDFs problematic for extraction
- 12:02Docling liberates PDF data
- 16:57Docling CLI and API
- 21:04Chunking for RAG with Docling
- 23:11RAG pipeline with Docling and OpenSearch
- 28:25Future integrations and OpenRAG
- 34:24Community growth and contributions
- 37:33Q&A
related talks
48mIntegrating Docling With OpenSearch for Advanced RAG and Agentic Applications
21mBuilding RAG With OpenSearch ML Plugins: From PDFs To Voice-Enabled Search
38mAI-Ready Data Foundation Using Amazon OpenSearch
01h 19mLearn to Unlock Document Intelligence with Open-Source AI
16mOpenRAG: An open-source stack for RAG
27mElevating AI Applications with OpenSearch's Flow Framework and RAG Tool
26mAdvanced Document Parsing for AI Workflows
28mUse OpenSearch as Your Vector Search Database To Optimize GenAI Accuracy!