ChatGPT Doesn’t Just Do Chat: How to Build Real Data Models Inside ChatGPT o3—works for Claude and Gemini as well

Source: Nate’s Newsletter Date: 2025-05-21 URL: https://natesnewsletter.substack.com/p/chatgpt-doesnt-just-do-chat-how-to

Summary

Advanced LLMs (ChatGPT o3, Claude 3.7, Gemini 2.5 Pro) have crossed a threshold from conversational assistants to “Everything Engines” capable of building rigorous analytical models — actuarial, predictive, scenario-based — that previously required dedicated data teams. Three concrete examples demonstrate the capability: AI product success rates, housing market entry probabilities, and whether layoffs actually result from automation.

Implications

Enterprise adoption thread. The democratization of data modeling is a direct threat to the analyst labor market: work that previously required specialized skills and dedicated teams is now achievable by practitioners with good prompting skills. Finance, operations, and strategy teams are the most directly affected.

AI economics thread. The “can’t tell because they don’t look different” framing captures a real adoption dynamic: capabilities have improved dramatically but the interface looks the same, so users aren’t aware of what’s now possible. This discovery gap is where Nate’s newsletter creates value — closing the lag between capability and awareness.

Watch: Whether enterprise data teams respond to this capability expansion by upskilling into more complex analytical work (model validation, business translation, data strategy) or whether the roles get reduced as AI absorbs the entry and mid-level analytical work.