Where I highlight a different approach to using LLMs and expand it into the whole repeatable thinking recipes bit.