Scalable RAG with Azure Cosmos DB and DiskANN
summary
Azure Cosmos DB now adds native vector search with the DiskANN index and demonstrates a Python SDK Jupyter Notebook that builds a small RAG pipeline and shopping chatbot. DiskANN compresses vectors for in-memory speed, stores a full‑precision graph on SSD, and re-ranks results with full vectors for high accuracy. The demo uses a new vector distance system function with cosine similarity, combines vector search with WHERE filters, a semantic cache, and shows deployment options via Azure Web App or Gradio. SDKs include Python .NET and Java.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 0:48Vector search announcement
- 1:27Cosmos DB vector advantages
- 2:45Understanding DiskANN
- 4:27Vector search live demo
- 7:53Shopping assistant demo
- 8:48Q&A and resources
related talks
19mHow Microsoft powers planet-scale AI apps with DiskANN
50mHow to design and build AI applications with vector search using Azure OpenAI & Azure Cosmos DB
12mBuilding Generative AI Apps with Azure Cosmos DB
03h 08mAzure Cosmos DB Conf 2025
03h 12mAzure Cosmos DB Conf 2024
40mUnleashing the Power of Vector Search with Azure SQL Database
15mAzure Cosmos DB as Vector Database
27mKeynote