Harrison's Hot Takes

Harrison's Hot Takes is a series of articles exploring the evolving landscape of building applications with AI and large language models. A central theme is the shift in how we understand systems, arguing that "traces" are the new documentation for AI in the way code is for traditional software. The posts delve into emerging practices like "context engineering" and offer guidance on architectural decisions, such as when to build multi-agent systems. Readers can also find practical advice on speeding up AI agents and reflections on the three-year journey of building the LangChain framework.