The Build

we are back with jetpacks, as promised.

see buildaipod.com for show notes and extras

details to follow

Details to follow

In Episode 21, Tom Spencer and Cameron Rohn break down the current state of AI — from market hype to hardcore engineering practice.Topics include:Michael Burry’s short on Nvidia and PalantirIs there really an AI bubble — or just a new kind of economy?Breaking Nvidia’s CUDA lock-in with modular AIGoogle’s “Nested Learning” and Anthropic’s interleaved thinkingBuilding AI copilots and MCP serversLangSmith experiments, evaluators, and continuous optimizationMicrosoft’s Copilot Studio and enterprise automationWhat real AI engineering looks like in production🎧 Subscribe for weekly deep dives into AI products, agent frameworks, and research.00:00 – Airport stories, brisket, and warm-up banter03:00 – MCP servers and Polygon data experiments05:00 – Minimax and Anthropic’s interleaved thinking07:00 – Google’s “Nested Learning” paper and continual optimization08:30 – NeurIPS, AI research culture, and the VC invasion09:30 – Is there an AI bubble? Michael Burry’s Nvidia short11:00 – Palantir, Nvidia, and the tech bubble debate14:00 – CapEx growth and the “AI money loop”17:00 – Are AI companies actually profitable?19:00 – Free users, monetization, and ChatGPT’s economics20:30 – The real differences from the dot-com era22:00 – Nvidia’s margins, chip efficiency, and modular AI challengers25:00 – Breaking CUDA lock-in and the rise of hardware portability27:00 – Local inference, hybrid models, and agentic operating systems33:00 – Chrome OS, MCP in browsers, and local AI34:00 – Anthropic Excel plugin and Kimi Thinking model benchmarks37:00 – MCP server demos and architecture discussion43:00 – Building an AI options trading copilot46:00 – Visualizing strategies, composable components, and LangGraph50:00 – How MCP connects data and trading logic55:00 – Skill systems, consistency, and reproducibility in LLM apps58:00 – LangChain documentation and developer experience1:00:00 – Combining MCP data for richer insights1:03:00 – Converting trading logic into agentic workflows1:06:00 – Building autonomous trading systems on LangGraph1:08:00 – LangSmith experiments, datasets, and evaluators1:13:00 – Backtesting AI outputs and customer feedback optimization1:20:00 – Comparing models and evaluators in LangSmith1:24:00 – Microsoft Copilot Studio and Power Automate for enterprise AI1:29:00 – Wrapping up: AI compliance, tooling, and what’s next