Sessions
Advanced RAG patterns for AI Applications
The Retrieval-Augmented-Generation (RAG) pattern has been a key architectural pattern for building AI applications that depend on grounding outputs in real-world, factual, data. However, beyond toy scenarios, it quickly becomes apparent that traditional simple RAG suffers from quality, scalability and efficiency limitations. In this session, we will analyze those problems and survey the architectural patterns and techniques to improve performance and accuracy of RAG in your applications. Additionally, we will explore recent innovations and see how RAG fits into agentic AI space.