Many engineers use generative AI for building PoCs, internal tools, or learning new concepts. However, the author finds LLM explanations overly simplistic and sometimes annoying due to emojis and style. Instead of passively consuming AI-generated content, they developed a structured approach to learn complex topics like chip manufacturing.

The method involves using models like Claude Code or Open Code in 'plan mode' to build foundational knowledge, verify accuracy, then simulate the topic in a low-poly, RollerCoaster Tycoon-style animation. The simulation includes UX features like responsive design and playback controls, hosted via GitHub Pages.

This approach resulted in Chip Tycoon, a visual journey from raw sand to finished chip delivery at a data center. The animation accurately maps each manufacturing step, helping retain knowledge better than traditional reading or bullet points. Enhancements could include photorealistic 3D models and interactive puzzles based on prior steps.

The author also created similar simulations for rocket engines, LLMs, F1 engines, and EUV machines.