IMO LLMs make for a great rubber duck https://en.wikipedia.org/wiki/Rubber_duck_debugging