EP - 10 - GPT 5 , China Open Source Model Onslaught, Claude SubAgents and Alpha Earth - Part 1

In this episode of The Build, Cameron Rohn and Tom Spencer dig into the future of AI agents, models, and startup strategy. They begin by mapping current model and tool developments, reviewing GLM 4.5 release, Quinn 3 model updates, Claude Flow Package, and integrations with LangChain and Langsmith to illustrate practical agent workflows. The conversation then shifts to developer tooling and deployment, evaluating Vercel and Supabase for realtime stacks and MCP tools for observability and orchestration in production. They explore technical architecture decisions with an emphasis on memory systems, the Browse Completion Dataset, and Flexible AI Stacks that enable Coin Model Integration Workflow and modular agent design. The hosts weave in entrepreneurship insights about monetization and markets, discussing Llama Ecosystem Services, the Chinese open source onslaught, and possibilities for UAE tech collaboration and mapping via Google Mapping Tools. Throughout, they balance building in public strategies—release cadence, transparent telemetry, community contribution—with tactical guidance on memory, caching, and model routing. Practical takeaways highlight how to compose Claude subagents, leverage Langsmith for debugging, and deploy with Vercel + Supabase while using MCP tools for lifecycle management. The episode closes urging builders to prioritize modular memory, open ecosystems, and visible iteration as the path to durable AI products and company growth.