Google AI Studio’s New Tool Nobody Else Has (gemini 3)
Let's learn how to use Google AI Studio (gemini 3)📅 2025-10-25
Transcript
[00:00] Google's new AI coding tool has a feature that no other platform has. Cursor AI doesn't have it. Chat GBT codeex doesn't have it. And no, Cloud Code doesn't have it either. And when you see this specific feature Google AI Studio has compared to these other ones, it'll make a lot more sense. As we saw with earlier videos on this channel, this is the Vibe coding tool. So, it's a vibe feature. Welcome back. Yo, we got another Google AI Studio video. This is going to be absolutely amazing. I want to do this video because there is a really cool tool here that I noticed that is absolutely peak vibe coding. As a side note, stick tuned on this
[00:30] channel. I got a lot of requests from people saying, "Hey, Corbin, this is a cool builder, but how do we take this code and connect it to Firebase and really do the next steps here when making it to a.com, adding a database, adding authentication?" Don't worry, I got you covered. I'm in the works of creating a very long series here that's going to show you everything and how to do that step by step starting from scratch in the Google AI Studio. But this cool little feature here, y'all, is called annotate app. The way you can think about this is quite literally it makes your front end a whiteboard. So, for example, start sketching. You know
[01:01] what? Don't call it Gemini thumbnail editor. I'm going to grab the little text tool here. It's called best editor ever. Enter. And then you go somewhere else and you're like, wait, you know what? I don't want a little generic image icon. No, no, no, no. Give me some coffee beans. Like, make this icon coffee beans. Notice this workflow, though, and why this is very unique to Google AI Studio. Because this is ultimate vibe coding. We've gotten to the point where it's not audio anymore. It's not a text input. No, no, no, no. Just straight up draw on the front end. When you've drawn and you liked everything you have, go to add to chat.
[01:31] Click this and then simply hit enter. So, we should expect a nice little caffy emoji here and then this to change the name here. While we wait for that to cook, here is two things I want you to know. The first thing is that this tells me that Google AI Studios direction when building out applications in this context is going to be leaning in heavily in the vibe coding niche. And why I say that is that I was playing around when creating an app earlier today and I noticed that there is a lot of limitations around this studio. One of the major ones that I found pretty frustrating was my ability to identify
[02:03] other external APIs and identify them in a secure manner using the EMV file. If you don't even know what I'm saying, I'm speaking alien language to you. So essentially if I go to code here and I come down to Gemini service, this allows us to access Google's API ecosystem, eg nano banana or Google maps or xyz Google API. Notice the API key here, Google geni API key. But we never see the API key and that's because it is stored in the process.vapi key. This is standard software development. Essentially what's incurring here is that it's being read
[02:34] in the EMV file. But they do this in software so that when I push this to something like GitHub, I don't reveal my key. Therefore, anyone could steal it. The issue being though, and let me know in the comments, I might be wrong or maybe by the time you watch this, they push out an update to solve this. I can't literally create the here to associate this key. Or alternatively, if I create a key like it's very it's almost too siloed of a coding environment for me to really do big changes. Like for example, if I wanted to integrate Firebase directly into this
[03:04] logic, but beyond that, let's jump back over to the app real quick. Let's see if the front end looks good. It's loading in. Okay, not bad. It definitely messed up, but that's the whole point of vibe coding. So, we got best editor ever with a nice little tilt. It didn't change this. So, let's try again. We're going to hit the annotate app again behind me. It's behind me. And what I can do here is maybe we got to do like an arrow. Wrong way. Arrow. Nice. And we can start sketching here. And we can add the text here and be like make this image icon an icon of a coffee cup. Maybe you have to be more specific. We like drag the
[03:35] little start sketch thing. Boom. This is very much like a whiteboard which is kind of cool. So I can like circle this be like you know what make this purple. Make this pink. It's already purple. Don't worry I'm not color blind. There we go. And then I can essentially say add to chat or am I color blind? And we're going to hit enter. Let me go back to that tangent. So second thing and obviously the Google AI Studio team obviously knows this. They know that right now if I try to integrate Firebase directly into this codebase in this environment, it doesn't work as effectively due to the nature of Firebase requiring so much high-risk variables to connect things like
[04:06] authentication, to connect things like database, to connect things like hosting. Yes, right now if I hit deploy, we get a live website link. So that is why I'm going to dedicate that entire series I referenced earlier to showing you the real way to approach this way of coding. Because as you can already tell, this is different than Claude. This is definitely different than Chad GBT codeex. This way of coding has a place and that's why we're able to do cool things like quite literally draw on our front end and get a change. That's cool. But let me create this series for y'all where I can connect the dots. And the
[04:36] way I see functionally Google AI Studio working in the dev workflow now is phase one. Not necessarily full-on MVP prototyping, a little past that. I genuinely think and I'm going to show in that series. We're going to build out basically all the front end and all the value of our application in Google AI Studio. Where Google excels is their ability to really lean into their API ecosystem. Therefore, we're going to take this thumbnail editor app, upgrade it to V2, and do most of the back-end logic for the value we provide and consumer, eg the edited thumbnail, and
[05:07] make the front end look good with a bunch of dummy data for the signin, sign out. The second part of that series will then take this code and integrate things like authentication and database. And then the third part will be like monetization and all the fun stuff. But look at this. Absolutely beautiful. It made my upload image pink. And we finally got the coffee. So that just about does it y'all. That is the annotation feature that is unique to the studio here. Let me know what y'all think about the studio whether you use it or not. And as you already know these style of videos. I'll see you in the next. Did Google AI Studio vibe so hard that they quite literally made front-end development like a whiteboard marker
[05:39] that you can just draw around like, "Hey, I don't want this. Oh, I do want this. Oh, make this a coffee icon. Oh, no. No. Make this button pink kind of video.