The Build - LangChain Open Deep Research

In this episode of The Build, Cameron Rohn and Tom Spencer dig into LangChain Open Deep Research and practical AI building. They begin by unpacking an open-source release and its implications for AI developer workflows, referencing Langsmith for experiment tracking, Tavoli Search API for retrieval, and the Integrated Graph Architecture as a backbone for agent state. The conversation then shifts to toolchains and deployment, contrasting Vercel and Supabase for frontend and backend hosting, and highlighting MCP tools for observability and orchestration. They examine API integration patterns using the Google Gemini CLI and evaluate model choices such as Claude Haiku Model and the KWAG Model when composing agents. They explore architecture and workflow frameworks, walking through Plan-Then-Present Prompt, the Multi-Agent Workflow Introduction, and Planner Feature Usage to demonstrate multi-agent coordination and planner responsibilities. Practical entrepreneurship topics surface: building in public strategies, open-source community growth, monetization approaches, and technical trade-offs when shipping fast versus shipping robust. Throughout, they balance technical architecture decisions—modular pipelines, retrieval-augmented generation, and agent orchestration—with developer-focused tooling and startup considerations. They close with a forward-looking takeaway: prioritize transparent iteration, modular architectures, and public builds to accelerate adoption and sustainable product-market fit.