In this episode of The Build, Cameron Rohn and Tom Spencer dissect the evolving landscape of AI agent development, focusing on practical strategies for building robust tools and businesses in public. They begin by analyzing the OpenAI MCP and its implications for agentic workflows, spotlighting how recent integrations—such as Codex and the Open Deep Wiki Feature—expand the developer toolkit. The conversation then shifts to architectural decisions, where they compare the merits of deploying on Vercel versus leveraging Supabase for data management, and discuss how Langsmith and MCP tools streamline experimentation and monitoring. They explore real-world use cases, including the implementation of agent supervisors like Second Opinion OAP and the technical nuances of integrating ChatGPT with Anthropic models. Drawing on their own startup experiences, Cameron and Tom share insights into building enterprise-ready connectors and agentic applications, emphasizing the importance of data sensitivity and ETL pipelines as outlined in the OpenAI Evaluation Framework. The discussion also touches on open source contributions, community-driven development, and the challenges of creating impactful versus stock content. By the episode’s close, listeners gain a nuanced understanding of how technical architecture, creative tooling, and open agent platforms like Google X XR can accelerate innovation. The takeaway: for developers and entrepreneurs, building in public and leveraging cutting-edge frameworks are key to shaping the next wave of AI-powered startups.