almost every llm has access to almost the same chunk of information on the web, they overlap hugely on what they are trained on. when everybody has the same llm and everybody asks the same question, everybody gets the same answer. what is rare, is the ability in asking differently, acting differently, and thinking differently.

when intellectual labor gets cheaper, knowledge itself becomes less valuable. it’s no longer about what you know (knowledge), or what you can do (skill). it is about knowing what to do (orchestration), this isn’t knowledge because it’s not a static piece of information, it is more of a creativity with patterns and harness. I don’t think of this as a skill because you can never fully acquire it, it is more like a constant state of trial-and-error.

our weapons are not static insights or information anymore. they are workflows now.

you need to constantly generate new, execution costing, rare information. the key here is not on the information but on the consistency. the moat is not the secret information but the speed and consistency in generating the information. because anything you generate today will either be discovered or copied by others tomorrow. you need to be at the frontier and constantly adapt yourself to the new dynamic that your product brings, identify new problems and opportunities. and it is this set of fresh data that is truly rare in the world where all the information is liquid and shared, we used to spend a lot of time digging and discovering other people’s data (market research comes from statistics, books from the authors, etc.) but this one just got democratized, now it is you that has to be the one to generate fresh, useful data. either by talking to the customers directly or testing ridiculous theories that only you believe in.

another way to frame this would be “mental diligence”. you can very easily tell how much thoughts are given when somebody talks about something. it is very consuming to be constantly thinking about the same thing. but it just became a very important quality. as we now are assisted by the llm, the level of thinking that we do gets directly projected to the models and their returns, which then usually reinforces our confidence in our beliefs. this loop is usually very fast as it only takes seconds for an llm to answer us. but because this happens so frequently and smoothly, we tend to experience a lot less friction in our thinking, we don’t question ourselves as much anymore. it takes a lot more effort to be critical about yourself now. the great degree of freedom in thinking just enslaved a great deal of people under the shackles. it is hard work, but you need to bring up the radical skepticism to get the game going.