L298N

There’s a working robot car sitting in my office. I’ve never built one, and I didn’t read the manual. Here’s what the build actually looked like when the AI had the knowledge and I had the hands. ⚡ TL;DR I built the physical robot car over two sessions — chassis assembly, motor wiring, first movement I have zero robotics experience. I didn’t read any manuals. Claude Desktop & Claude Code spec’d the parts, guided the build, and caught the problems. Fried two L298N motor controllers learning a lesson about polarity. Saw a spark. Smelled the smoke. The PM/dev-team metaphor from post #1 holds — but the collaboration is richer than “I direct, you do.” The people who’ll thrive with AI aren’t necessarily the deepest technical experts. They’re the curious and creative ones. Also: we’ve already cycled through three vision models (Llama → Qwen → Gemma) and three Claudes (Opus 4.5 → 4.6 → 4.7) in three months. Don’t get locked in. Where We Left Off In the first post, I made the case that operating systems built for humans aren’t optimized for AI users, and I described AIOS:ThinkTank — a robot car driven entirely by a local language model — as the experiment testing that idea. ...