RAG at scale: production-ready GenAI apps with Azure AI Search
summary
Azure AI Search demonstrates production RAG capabilities: it loads 1.4 billion vectors with interactive query latency, raises capacity up to 12x vector density and up to 6x storage at no extra cost, and adds Int8 and packed binary support plus automatic quantization with oversampling rerank to recover accuracy. Integrated vectorization indexes text and multimedia from OneLake (including S3 shortcuts), supports model-catalog embeddings, and powers deployed systems like KPMG 10,000 / AT&T 80,000 users daily.
AI-written summary · about
chapters (10)
- 0:00Introduction to RAG at scale
- 4:46Scaling pressure points
- 7:51Azure AI Search overview
- 12:07Billion-scale demonstrations
- 16:55Vector compression techniques
- 21:57Integrated vectorization
- 26:34Advanced index configurations
- 30:16Multimedia and data sources
- 38:35Optimizing retrieval quality
- 43:22Success stories and conclusion
related talks
01h 16mAzure SQL ❤️ Open AI
44mSupercharge your data with Azure AI Search and OpenAI
14mBring AI to Your Data
45mMaking semantic search & RAG real: How to build a prod-ready app
26mProduction RAG at Scale with Azure Database for PostgreSQL
21mHow RAG is Redefining AI Capabilities
01h 35mTutorial: Build AI-Supercharged RAG Apps with a Vector Database
01h 03mAI Templates