1 How RAG research prevents disasters

Retrieval-Augmented Generation (RAG) is an architectural pattern that enhances large language models (LLMs) by connecting them to external knowledge sources in real time. Unlike traditional search engines that return documents for humans to read, or standalone language models that rely solely on training data, a RAG system performs the synthesis itself. It retrieves relevant information to inform AI-generated responses and generates answers grounded in that evidence. This distinction is critical: search tools find sources; RAG systems use sources to construct reliable, verifiable answers. RAG systems offer significant advantages over standalone LLMs, but to build a successful one, it's important to understand how they work and recognize the limitations of different approaches.

1.1 How this book teaches RAG

1.2 How do RAG systems work?

1.3 How RAG research addresses failure points

1.3.1 The knowledge boundary problem

1.3.2 The hallucination challenge

1.3.3 The private knowledge access problem

1.3.4 RAG is an enduring architectural pattern

1.3.5 Why RAG justifies infrastructure investment

1.4 Failure points within RAG systems

1.4.1 RAG failure examples

1.4.2 Research-backed solutions

1.5 Naive, advanced, modular, and agentic RAG

1.5.1 Naive RAG: Establishing the foundation

1.5.2 Advanced RAG: Optimizing the components

1.5.3 Modular RAG: Adaptive and composable systems

1.5.4 Agentic RAG: Autonomous information-seeking systems

1.5.5 RAG evolution

1.6 Summary