Context Engineering for Agents - Lance Martin, LangChain

In this episode of the Latent Space Podcast, Alessio and Swyx host Lance Martin to discuss context engineering in AI agents. They explore the distinctions between prompt engineering and context engineering, emphasizing the challenges of managing context from tool calls in agents. The conversation covers strategies such as offloading, retrieval, and reducing context, along with the nuances of multi-agent systems and the importance of context isolation. They also touch on the "bitter lesson" of AI, highlighting the trade-offs between structured approaches and more general, adaptable models, and the role of frameworks in AI engineering.