Used to think LLMs have no moat as models get better and better and the switching cost is near zero. Recently started asking chatgpt some questions based on my behaviors and prev responses...realized the moat is in models understanding user behavior extremely well.

Try asking some interesting questions... "based off what you know about me..."