**Logan Kilpatrick** (0:00)
The paradigm is changing, where you actually build, and people are doing this with other tools in the ecosystem. You build many of these things in parallel at the same time, which is really exciting.
**SPEAKER_2** (0:08)
This is a very complex app that we just ripped up in 10 minutes.
**Logan Kilpatrick** (0:11)
The full end-to-end runtime was 7 minutes, and the app works, which is crazy. It's been fully live for every single Googler for the last month and a half. People across Google, I think, have built 300,000 internal apps already.
**SPEAKER_2** (0:28)
Wow, out of curiosity, are there any interesting apps that you've seen being built internally at Google?
**Logan Kilpatrick** (0:34)
The reality of what's happening right now is like Google is building Google on top of AI Studio. And then I think the fun part is that you'll be able to have a deep research agent working with a Lyria 3 agent. And so at the end, you can literally write the string of code that is like, go do deep research. And instead of me reading a long academic report at the end, turn it into a song. Hopefully, we're delivering on this mission of helping you bring anything to life with Gemini, which is exciting.
**SPEAKER_2** (1:03)
Logan, excited to have you again. Been almost a year since we last chatted. What are some of the exciting things? It's crazy. What are some of the new and exciting things that are coming out of AI Studio?
**Logan Kilpatrick** (1:13)
Yeah, I mean, it's crazy because I've been seeing all the stuff that you all have been doing, which is why I've just been seeing all the posts. Yeah, I mean, it's a super exciting time. I think we just landed Gemini 3.1, which is really exciting. So our new frontier state-of-the-art pro model, which is awesome, so crushes a bunch of the benchmarks. The model is also really good at coding and building front-end apps, which is super exciting. The week before that, we launched DeepThink, which is even more state-of-the-art, uses a parallel reasoning in order to get even better at solving a bunch of really difficult STEM problems, math, science, engineering, etc. Then now I'm super excited that we've brought a bunch of the new model improvements as well as the anti-gravity coding agent, and a whole new full-stack runtime into AI Studio, so that everyone can vibe code full apps, which is really exciting. I think we've been doing the work behind the scenes and actually, a bunch of folks in the ecosystem have been testing and giving us feedback in this experience now fully rolled out to everyone, which is really exciting. So you can really, hopefully, we're delivering on this mission of helping you bring anything to life with Gemini, which is exciting.
**SPEAKER_2** (2:29)
Awesome. I feel like everyone who is watching this right now probably wants to see the vibe coding thing. That's the main thing here. So if you want, we can just jump right in and give the viewers what they want.
**Logan Kilpatrick** (2:40)
Let's do it. Let's do it. Let me share my screen and we'll do some stuff. All right. You should be able to see the current version of AI Studio in the UI, which looks very similar. We'll actually have a new version of this rolling soon, which is really exciting. So hopefully, we'll see that. If you want to go and get a quick look at some of the new experiences and an example of some of the things that we've enabled, one of the big new ones is other than just the coding agent being a lot smarter and able to do stuff, is actually building multiplayer experiences. So this is a lot of fun because you can just share a link and that we can do this together. Actually, we should try doing this and build a game together. Let's go. I'm going to go back to the start screen, and I'm going to do my YAPTA app to voice the text, and we'll build something.
**SPEAKER_3** (3:31)
Cool.
**Logan Kilpatrick** (3:33)
All right. So I want a multiplayer dino run game like the Chrome loading page has, and it should look just like the Chrome dino run experience when you get a 404 or whatever it is, except I want to be multiplayer. So there should be like a lobby, I should be able to load in, and then if there's other players who are loaded in, there should be like a clock timer, and then we should all, once there's at least two players, it should kick off, and we should be able to play against each other in dino run. So make that app, make it multiplayer, and let's try it out. Yeah. So we'll YAPTA app this, and then we'll kick this off and see what happens. It'll take a little bit of time to make this. So this is actually one of the, maybe one of the trade-offs from a new capability perspective is the model is like much more capable for certain generations though. It actually takes longer because it's building a full backend and doing all this other stuff for you. The nice thing though is you can actually leave this, or you can close the tab and come back to it later. It's completely async happening on the backend. If you want to follow up, you can, it'll be done later on, so you don't need to actually sit here and watch. We can look at a couple of other examples or we can do some magic editing and cut to when it's done in a couple of minutes. Yeah. Any questions? I mean, the other core parts of the experience, we updated some of the sharing components of the experience. Publishing is a little bit refined, so there's some nice quality of life improvements across the experience.
35 more minutes of transcript below
Try it now — copy, paste, done:
curl -H "x-api-key: pt_demo" \
https://spoken.md/transcripts/1000651996090
Works with Claude, ChatGPT, Cursor, and any agent that makes HTTP calls.
From $0.10 per transcript. No subscription. Credits never expire.
Using your own key:
curl -H "x-api-key: YOUR_KEY" \
https://spoken.md/transcripts/1000756412942