Everything Gets Rebuilt: The New AI Agent Stack | Harrison Chase, LangChain

The podcast explores the evolution and architecture of AI agents, particularly focusing on the infrastructure required for agents that plan, use tools, and manage memory. Harrison Chase, co-founder and CEO of LangChain, discusses the shift from early agents that struggled due to model limitations to the current generation, which benefits from improved models and a better "harness" or framework. He emphasizes the importance of the harness in enabling effective model interaction with its environment, highlighting components like detailed system prompts, planning tools, subagents for context isolation, and file systems for LLM context management. Chase also touches on the convergence of conversational and coding agents, the role of sandboxes for secure code execution, and the significance of long-term memory in agent development.