Logan Kilpatrick

reinforcement learning environments and you can make them really good at domain specific tasks.

And then ideally you learn a bunch of capabilities that generalize across a bunch of different stuff.

that is turned out to be like somewhat true in certain areas.

And like, maybe we'll end up being more true over time, but it's not like a clear slam dunk.

Like you could make, and there's lots of good examples of this, like actually making an AI system, non LLM frontier model that can play all these games was a problem that was solved probably like four or five years ago.

And yet at that time,

there was nothing like LLMs.

They didn't have this general purpose intelligence.

It wasn't actually that useful.

Like making something by default that is good at playing games doesn't end up being like beneficial for most people.

But making something that's really smart and good at things that can also play games, hopefully will be the way.

And I think of like LLMs as this like delivery mechanism for intelligence that like hopefully continue to generalize across all these different domains.

I think for, um, so we're not like one of the, I was just in a meeting and we were talking about this.

One of the explicit things with AI studios, like we're not trying to solve every problem.

Um, so like really there are great tools that are out there in the ecosystem.

Some, some of which are made by Google, like Gemini CLI, some of which are made by the rest of the ecosystem.

And really the sweet spot of what we can do in AI Studio is sort of get you a feel for what the models are capable of.

Hopefully with all the vibe coding stuff we're doing and with build mode, get you a working prototype and then go and get you out into like a full-fledged sort of professional developer product.

And that could be your IDE of choice, your CLI of choice, et cetera.

And I think that's where