Antigravity: Google's NEW Agentic Editor in 7 Minutes

--- type: transcript date: 2025-11-23 youtube_id: wbPpvjcAHew --- # Transcript: Antigravity: Google's NEW Agentic Editor in 7 Minutes In this video, I'm going to be taking a look at Google Anti-gravity. This is the first release from the team, at least part of the team that was originally at Windsurf. Some of the founding team ended up leaving to go to Google. They sold non-exclusive IP rights and this was the product that they ended up building on top of. One of the interesting things with this editor is it's going to feel very similar to something like VS Code or Cursor or these VS Code forks really at the core of it. There's also a lot of new abstractions in terms of how you can interact with the agent as well as how you can also test your product and also have a lot more visibility as well as a platform to actually manage and spawn off different agents. One thing that I do want to mention off the bat is you can try this out for free. It is in a public preview. Because it is a public preview, just know that there very well are going to be rate limits. There is a ton of attention around this right now, but I do hope that they have a paid tier at least to allow people to try this out. maybe even a $20 a month here or something like that. The first thing that I want to show you, so once you open up Any Gravity, the one thing that you'll notice within here is you have this agent manager. Within here, what you'll have is you'll have an inbox where you can spawn off different agents. You can see which agents actually require your attention. If you have a number of different workspaces that you're working in, any number of different projects, you'll be able to coordinate all of that within this view. And then additionally, if you do want to open up the IDE itself, you can go ahead and command E or additionally, you can click to open up the editor in the top right hand corner. As soon as you add in a directory, you'll see it on the lefthand side here. And one of the nice things with this is you'll be able to toggle between the different workspaces to start a new conversation thread for each of the agents just by simply clicking on the dropown. Now within here, what we can do is we can add in images and we can also appmention like you might expect. Then we have two different modes. We have a planning mode and then we also have a fast mode. In terms of the models that we do have available within the platform, we have Gemini 3 Pro on high and then we also have Gemini 3 Pro on low. We do also have, interestingly enough, Claude Sonnet 4.5. And then what is sort of funny is we have GPTO OSS 120, which I can't say I know a single soul, unfortunately, that uses this, at least in a coding context. What is sort of interesting, we have two of the best models in the world, and then one open source model from OpenAI. For some obvious reasons, you can imagine probably why they didn't include GPT5 within here. Okay, the first thing that I'm going to do is I'm going to say I want to create a beautiful SAS landing page that says developers digest. As soon as I send in that request, we'll see within the pane here, we see it working through the task. The agent is spawning off the work, we can see that it's created a simple HTML, CSS, as well as JavaScript file. And now, one of the really cool things and features of this is without me even asking, not only does it open it up within a preview, it will actually go and test out your application. So, if we Here is the first iteration of what it generated for us. So, I don't particularly love this. Like, it's pretty good, I guess, in some regards, but there are some things I don't like. I don't love all the emojis. Obviously, the purple linear gradients isn't something that I love. I can't read all the text colors. Some of the effects are okay. And the styling, at least in terms of some of the responsiveness, isn't quite there, but it does look pretty good, right? In terms of the generation that you would get from other models, this is probably better than most, I would probably say. Now, in terms of actually hopping into the IDE modes, you can simply click to open it within the editor. And then this is going to be the environment that you're probably going to feel much more familiar with if you've used something like cursor, VS Code, or Windsorf. So, you're going to be able to have your agent on the right hand side just like you typically would. And then you're going to be able to send in further edits and be able to configure it just similar to how you sent in that initial prompt to the AI agent. Next one I wanted to show you is they did also incorporate nano banana directly within the product which I thought was pretty neat. What you can do within this is I can say I want to generate an image of a plant store landing page. I want the overall look and feel to be modern and I also want to create this application within Nex.js. And then now what I can do is I can just change my workspace over to my example space here. And then here I can see it's envisioning the landing page. The unfortunate thing here is even though it does have the ability to generate images, it is heavily rate limited right now. Within here, we see it going through and it's setting up the Nex.js application. It went ahead and it ran all the different commands and it's going through and it's installing everything for us. And what I can do within here is again if I just command E, I can open up that editor and then here we go. This is the environment and we can see the exact same thread on the right hand side here. It is sort of nice just to be able to hop between these different environments. So within here, it's gone ahead and it's spun up the development server here. Now, just a few aside. So you're going to be able to see all of the different files that it changes as well as the different terminal commands. Here we go. So it's gone ahead and it's set up our application. It's given us steps in terms of how to run this. And because we're starting to live in the future within a Gravity here, I'm just going to say let's open this up in the browser and go ahead and test out a few features. And I want to have some recordings to actually see making sure that everything works and what have you. Let's test this thoroughly. And here we go. It is beginning to test out our application. The really cool thing with this is what we'll see is it will actually come up with the commands to go ahead and click through our application here. If I just scroll up here, we can see all of the different actions that it's taking. So, it's moving the mouse in the browser. It's waiting a number of seconds. It's moving the mouse. It's hovering over particular elements. It's clicking on buttons like shop now or what have you. We can see it's scrolling and it's reasoning between each of these steps here. This is both very helpful as well as honestly a killer feature. I've seen this in some other products but not quite like this. I haven't seen VS Code have anything like this. I don't think Windinsurf have anything like this. I have seen this I believe in Devon. I believe another product that does have a form of testing like this is emergent labs. I believe they have something similar. But in terms of actually a local tool that I personally would probably spend most of my time in like an IDE, I have not seen this. So this is pretty interesting. And then within here, we can see all of the different things that it went ahead and tested for us. The next one I wanted to show you is the planning nodes. Within here, I'm going to say I want to create another Nex.js application that reads developers digest. Let's have a blog as well as a contact section. Let's give it a beautiful neo brutalist feel with overall modern sort of blacks and whites and then maybe a little bit of a purple and blue throughout the site as well. I'm going to go ahead and send this off. Now, one of the differences with the planning mode is instead of it actually just spawning off and going and doing all of those different tasks, what it will do now is it will develop a plan for us. Additionally, what you can do within the editor is you can leave comments on whether it's the different tasks, the planning stages, or if you want to actually comment on things like different images that you pass in or if it were to take different images or videos and you want to leave certain feedback. So, within here, I could just say skip this step for instance. And what it will do is it will pass in that context into the agent. But it just gives you more surface areas in terms of how you can actually interact with it. Obviously, like I mentioned, you can add in images. Now, another interesting thing with this is Gemini 3 Pro does support videos. I think there will be potentially some interesting things that you could do by actually feeding in the context of a video in the future. At least, I definitely hope that is something coming down the pike. But this is a really interesting option for people that are just generally interested in Gentic IDEs. This is still very new to me, but there are a ton of interesting features within this. Otherwise, I just wanted to do a really quick one showing you Anni Gravity. If you found this video useful, please like, comment, share, and subscribe. Otherwise, until the next