Don't Tell Me RAG Is Easy - Building Retrieval-Augmented Generation Systems

Explore the complexities and challenges of implementing Retrieval-Augmented Generation (RAG) systems in this 36-minute conference talk from DevConf.US 2025. Discover why building effective RAG solutions goes far beyond simply dumping documentation into a database, as speaker Major Hayden shares real-world experiences and roadblocks encountered while developing a product with RAG as its core component. Learn about the technical limitations that arise when using large language models (LLMs) and how RAG addresses knowledge gaps through just-in-time data retrieval from databases. Understand the available technologies for building RAG stacks, gain insights into proper implementation strategies, and discover how to avoid transferring human complexity into already complex technical systems. Benefit from practical lessons learned about the intersection of human knowledge organization and machine learning systems, while getting guidance on navigating both technical and organizational challenges in RAG development.