Hey Clief Notes fam! 👋 I'm Ali, been here for some time, read the ICM paper, watched Jake's videos and I love them. But I have yet to get it. I feel that for many use cases we don't need an agent whether using ICM methodology or any other way to build agents.
ICM seems to be a simple idea but is hard to explain and everyone is trying to explain it (like monad tutorials). I guess I am not getting it because as Jake often says in his shorts and videos "You don't need all of that", I haven't tried that "that" that I am still unable to get it.