Side note: I tried the same questions with some local LLMs I’m running at home—unfortunately, they’re nowhere near as good or useful. I hope local models improve quickly, so we’re not left depending on the good graces of big LLM(tm).