Retrieval Augmented Generation in the Wild

Explore the practical implementation of retrieval augmented generation systems in this conference talk that examines the recent explosion of retrieval-based AI applications. Learn about the retrieval loop, its purpose in AI systems, and how human feedback enhances performance across various applications including document question answering and autonomous agents. Discover the key challenges developers encounter when building these systems, including selecting appropriate embedding models, determining optimal chunking strategies, and ensuring retrieval relevance. Gain insights from real-world experience building embeddings-based retrieval systems and understand what the future holds for retrieval in AI contexts, presented by the co-founder of Chroma, an open-source embeddings store designed specifically for AI-native applications.