Advanced RAG with Azure AI Search - top questions answered
summary
Azure AI Search shows concrete steps for RAG: new 6–12× SKU capacity lets single SKUs handle far larger workloads, vector quantization cuts vector storage by ~90% while retaining about 90% accuracy, and production deployments hit mid‑millions requests/day (OpenAI use shows 7–8 digit transaction scales). The demo is a patient-facing Tellis chatbot built with Azure Blob/OneLake ingestion, vector search + hybrid semantic ranker fusion, integrated vectorization (preview) and Langfuse observability, and it returns grounded, multilingual answers with citations.
AI-written summary · about
chapters (11)
- 0:00Introduction to Azure AI
- 2:13What is Azure AI Search
- 4:45The importance of RAG
- 9:18Latest AI search features
- 12:36Scaling rag applications
- 15:31Improving search accuracy
- 18:08Data ingestion and pipeline
- 23:03Live healthcare chatbot demo
- 28:29Advanced demo functionality
- 36:22RAG architecture best practices
- 41:50Observability and conclusion
related talks
44mSupercharge your data with Azure AI Search and OpenAI
50mBuilding AI ready applications
44mVector Search Isn’t Enough
01h 03mDrop the Bass with Embedding and Vectors in Azure AI Search
01h 03mDrop the Bass with Embedding and Vectors in Azure AI Search
01h 03mAI Templates
01h 00m5 Levels of AI in Microsoft Azure
50mAzure AI, Power new possibilities for every organization