Ep 01 - LangChain updates, Google & Microsoft releases and the Daytona live demo.

In this episode of The Build, Cameron Rohn and Tom Spencer dissect recent LangChain updates and demo a Daytona live system while mapping practical choices for building modern AI products. They begin by unpacking LangChain Platform changes, the Best Model Slide and Python Package Clarification, and how Langsmith and Langraph Cloud fit into developer toolchains for model evaluation and orchestration. The conversation then shifts to infrastructure and integration patterns: using Vercel for frontend deployment, Supabase for realtime storage and auth, and MCP tools and MCP Agents for distributed agent workloads and agent-to-agent communication. They explore architecture and API considerations, comparing agent frameworks, advice for embedding Ai Agents, and decisions around RPCs, eventing, and cost-aware model routing. Along the way they surface building-in-public tactics—iterative demos, transparent telemetry, and community-driven prioritization—and entrepreneurship insights such as verification-focused startups, the Agent Distribution Revolution, and product ideas like an Agent Email Manager. Technical trade-offs (state management, observability, and scaling) are grounded in code-level patterns and developer workflows. The episode closes with a forward-looking takeaway: developers and founders should prototype agent architectures publicly, instrument them with Langsmith and MCP tools, and iterate deployment strategies on Vercel + Supabase to accelerate product-market learning.