what i keep coming back to with @ritualnet is that it doesnt feel like a model first approach at all it feels more like they are designing a place where models can exist long term not just run once give an output and disappear but actually persist interact and evolve within a system most ai setups today still feel very stateless you prompt you get a result and thats it nothing really carries forward unless you force it to ritual seems to be pushing toward something where state actually matters where outputs can feed into future actions where models can interact with contracts data and even other models in a more continuous way and that changes the vibe completely because now youre not just thinking about solving a task youre thinking about maintaining behavior over time if that direction works then the interesting part is not going to be how smart a single output is but how systems behave over longer horizons how they adapt how they coordinate and how they keep producing value without constant human input so instead of chasing peak intelligence it feels closer to building something that can sustain itself and that feels like a very different game than what most people are still focused on