Very cool. I’ve been daydreaming about whether LLMs can be used to reason through gaming decisions.