EP 22

In this episode of The Build, Cameron Rohn and Tom Spencer dig into rapid AI agent development and practical startup strategies that shape modern product velocity. They begin by framing the current landscape of AI agents and memory systems, assessing claims versus measurable engineering work and highlighting the role of Langsmith and MCP tools for orchestration, evaluation, and experiment tracking. The conversation then shifts to developer tools and workflows, where Vercel and Supabase surface as pragmatic choices for deployment, edge functions, and persistent state—connecting technical architecture decisions like vector stores, caching layers, and stateful agent design to real release velocity. They explore building in public strategies next, describing how transparent telemetry, public roadmaps, and incremental demos accelerate community feedback and open source adoption while preserving optional monetization paths. The hosts also analyze entrepreneurship insights, from choosing MVP scopes and pricing models to hiring for systems-level thinking in AI teams. Across the episode, they compare various development approaches and trade-offs in framework selection, memory architecture, and agent orchestration. The discussion concludes with a forward-looking call to action: prioritize pragmatic architectures, instrument experiments with MCP tools, and keep iterating in public so developers and founders can build resilient, composable AI products.