Raw LLM Responses

the idea that apps are transitional is one of those things that sounds obvious once you say it out loud but nobody really acts on it. what i think happens first is not a wholesale replacement of apps but a thinning layer. the ai interface sits between you and your existing tools, and over time the tools that expose good APIs get absorbed into the workflow while the ones that do not just become obsolete. we are already seeing this with things like zapier and n8n but the ai layer makes it conversational instead of visual. the interesting part about the karpathy wiki comparison is that personal knowledge bases might become the new "apps." instead of opening an app, you query your data layer and the ai figures out which source to pull from. the value shifts from owning the interface to owning the data structure underneath. one thing i would love to see explored more is how permissioning works in this model. right now app permissions are app-scoped. if an ai agent is accessing multiple data sources on your behalf, the permission model gets a lot more nuanced.