Google's anti-gravity just released and

in this video I'm going to show you how

to use it, how to set it up on your

local machine and I'm also going to show

you three major features that you need

to know if you're using anti-gravity

here to do vibe coding. So first thing

first I'm going to show you the one of

the feature which is the agent manager.

You can think of it like a orchestrator

that orchestrate multiple AI agents

running at your same times inside of

your coding IDE so that we're not

wasting any time for our vibe coding

here. And the second feature is the

browser agent. This will basically help

us to navigate to our applications and

examine and also help us to figure out

what can we improve for application. And

lastly, I also want to show you is the

ability to have multiple models inside

of our coding editor. For example, we're

not limited to just the Gemini models.

You can see here we also have the

ability to use claw models, the GBT

models, any models we like. Okay, so

pretty much these are the three main

features that I'm going to show you for

anti-gravity. And I'm also going to show

you the development workflow that I'm

going to follow to make sure that we're

getting the highest accuracy for using

Google anti-gravity for the lab coding

here. And to do so, we're simply going

to provide the prompts on how we're

going to build things. Then we're going

to pass it to the plan mode here to

basically create a task list on how the

AI agent here is going to perform the

task. Then we're going to pass the task

list here, the instructions to the large

language model to build the

applications. Then we're going to test

out the applications using the browser

agents that we just talked about.

basically have it to generate a report

on what things did well and basically

continue to cycle through until we have

a perfect application built. So pretty

much that's what we're going to cover in

this video. So with that being said, if

you're interested, let's get into it.

All right, so to get started with the

Google anti-gravity, first thing first,

we're going to download this on our

operating system. So in this case, I'm

going to download this on my Mac OS. And

once I download this, open the

applications here is navigating me to

the onboarding session. So I'm just

going to follow this setting up import

from VS Code settings. Now, in terms of

the editor themes, I'm just going to

choose the import themes that we have,

which is the solarized light, and click

on nest. Here, in terms of the

antigravity agents, we're just going to

choose the recommended, which is the

agent assisted development. And here in

terms of configurations, I'm just going

to use the default one here and click on

next. All right. So, once we set up with

the onboarding process, here is

basically our anti-gravity co-editor

here. So, one of the main feature from

anti-gravity is the agent manager. So

here if you want to access the agent

manager simply just going to do command

E or you can click on the open agent

manager right here and here you can see

this will basically navigate us to the

agent manager. So basically what this

agent manager does think of it like an

orchestrator which will help us to

orchestrate multiple agents running at

the same time. So for example here you

can see we have different workspace that

we have created inside of our agent

manager for example we have our main

website we have our bookkeeping

applications and here we can also be

able to add a new one here by simply

just clicking on the open workspace

right here. So what we're going to do

here is I can be able to open a new

conversation for example I can be able

to use the context here to reference any

files or folders inside of our projects.

I can also be able to attach any files

or images we have and simply I can also

be able to change the conversation mode.

For example, if we want to do a plan

mode, if we want to do a fast mode like

direct executions, right? And also we

can change different models. For

example, we can use Gemini 3 Pro and

also claw as well as GBT as well. So

that we're not limited to the specific

model that the Gemini provides. So in

this case, I'm just going to um mention

a prompt here saying that please help me

to look through the entire codebase and

try to understand everything and give me

a rundown on what is currently going on

with this application. So I'm just going

to provide a simple prompt here. Try to

have it to index our codebase. So I'm

just going to submit this request and

let it do its thing. And while it's

doing that, I also want to create a new

conversation here with the ARF type

website. So here I'm just going to say

based on the uh index.html here. So

based on this page, I want you to

examine this page and try to understand

what are some areas that we can improve

to improve the overall high conversion

rate on this landing page. So once we

have our prompt, we're just going to

send this request. And at the same time,

I can also create a new conversation and

be able to have another conversation

with the same workspace that I have. For

example, here I can be able to provide a

prompt that I wrote. And this is

basically a context on what are the

information that we need to be able to

have them to create a better version of

the website. So what I want to do here

is that I want to say based on this

prompt right here that I have, I want

you to do a research online to find out

what are some themes, what are some UI

themes that we can use to copy and be

able to improve our current website

design. And I also want you to open the

browser here to navigate to this website

and be able to take some screenshots and

see and be able to do a comparison. All

right, so now you can see that this is

the prompt. Not a perfect prompt, but

we're just going to submit this request

and see how Gemini handle this. So,

while Gemini process this, one of the

feature from Gemini anti-gravity here is

basically have the ability to open the

browser and analyze the changes. So, for

example, here you can see that it's

going to create a to-do list. So, here

you can see on the right, which is our

UI theme research and comparison, and

what it does first, it's going to

understand the prompt, right? And also

try to analyze the current website by

opening the website browser. And as we

speak, you can see that it opens the

website browser. And basically you can

see that it's going to taking some

screenshots and try to analyze the

entire website. So let's wait for a bit

until it fully analyzed this. All right.

So after it has analyzed the entire

website here you can see that these are

the things that it captured. And here is

basically the research that's going to

do right. So it has analyzed the entire

website by taking the screenshots and

now it's going to do the research on the

UI for the themes like the enterprise or

the high ticket sales and try to select

top candidates here to make decision. So

while I was doing that I'm also going to

jump back to another conversation with

another agent in this case the

bookkeeping applications that we have

been working on and here you can see

that this is the current status right

now right so you can see that has

complete these things and here is

basically the project overview that has

understand and now what we can do here

is I'm going to have a new conversation

with this AI agents by basically saying

that can you be able to open the browser

here and be able to log in with the

following credentials and I'm just going

to provide the credential and basically

what I wanted to do is to I want you to

log into this application, take a

screenshot of the dashboard and also

take a screenshot of all the subpages

that we have inside of the authenticated

account and after that I also want you

to play around with the match page where

user can be able to match the

transactions with receipts and see if

those features are working and if not

then please make sure to log them down

exactly what are the problems and based

on the user experience how can we be

able to improve a better UX for this

application and here I'm just going

going to have Gemini here to send this

request and let it do its thing. All

right. So, while it's doing that, let's

also check back with the Eric Tech

website. So, right here you can see that

it has done the research for the UI

themes. And here you can see that has

completed the research and the

comparison task. And now, if I were

navigate to the website here, you can

see that this is what it looks like. So,

you can see we have our Amazon, we have

our Microsoft, and here we also have our

call to actions. And also, we have our

education footer. And here is basically

the light themes. We can choose the

light themes here. And this is basically

our light theme right here. Okay. So

overall you can see the style here the

theme looks much more better compared to

the last one. Now whilst doing that

let's also check back with the

bookkeeping applications and we can see

here that it in fact did open the

application in the browser and try to

navigate to different parts of the

applications like the receipt page you

can see here and after it takes some

screenshots and it will also navigate to

the transaction page and after

everything's done it will basically try

to analyze everything and eventually uh

it should be able to generate a plan

based on what we have here. So here you

can see it generates a detailed report

with the screenshots and recommendations

in the UX recommendation review. So

let's take a look at what this generate

for the UX review doc. All right. So

here you can see that inside of this

review report you can see we have a

bunch of screenshots for each pages like

receipts statements and match pages. And

here you can see that inside of our

findings we have all the findings that

we find for each page as well as the

problems the recommendations and

basically we can also be able to add

comments onto this report to basically

have it to do a revision of the report

and be able to continue from there. All

right. So so far we have went over is

the agent manager on how we can be able

to have the agent manager here to

orchestrate different AI agents running

in the background and also the browser

mode which we can have our AI agent here

to open the browser for applications and

try to verify or analyze the changes

that we have inside of our application.

So now what I want to talk about is how

we can be able to use different large

language model. For example, here you

can see we have a new prompt and here we

can choose our AI agents or in this case

our large language model to be able to

have a different model here to process

our request. So in this case I'm just

going to send this request and the goal

here is basically adding a new section

called project section inside of our

static landing page. All right. So

eventually here you can see that it has

added this feature for the feature

project section inside of our website

with all three video demo cards are

working correctly with the hover

effects. So right now if I were to look

at the preview this is the screen

recording that it has made. So if you

scroll down this is the feature and if I

were to scroll down this is the feature

but it doesn't show the uh image from

the Google drive link. So in this case

I'm just going to come back. So here I'm

just going to say that the current cars

for the projects does not show any

images. So please make sure to add the

following after the image URL to be able

to have it to display the image. And

basically this is the slash view that I

was talking about to basically add it to

the each thumbnail that we have. And I'm

just going to send this request and see

how it works. And in the meantime here

I'm also going to um copy the rei file

for each projects that I have. So for

project one this is the rei file. And

for the project two

this is the REI file. And then for the

project three, this is the REMI file.

And please make sure to update the right

information for the project description

for each cards. So in this case, I'm

just going to send this request to claw

4.5 here to then have it to update the

project cards. So in this case, we can

see it's going to process this request.

And let's wait for a bit until it fully

process this. So now you can see that it

has done the verification for the

project and here if we will scroll down

you can see that if I were to open the

application this is what it looks like.

So this is our static page which will

have our feature projects. So this is

basically our three projects that we

have. We also if I were to click on this

you can see it navigate us to the right

projects inside of our YouTube channel.

So it also has the descriptions for the

projects, the skill sets, the tools that

we use and all those kind of things.

Now, before we jump into the next

section, let me give a quick shout out

to the sponsor of this video, Testbrite.

Testbrite is an AI agent that's

specifically for software testing. With

the release of the Testbrite MCP, you

get a chance to use the Testbrite MCP

inside of your coding IDE like clock

code, cursor, windserve, and more. With

simply just adding the configurations

inside of the MCP settings, you can use

it right away. With the test MCP, it not

only understands what you want because

it first read through the codebase,

understand the documentations as well as

validates the results that your Asian

roads and it is doing that by

automatically generating the test plan

for the PRD documentations and producing

the test cases and test coverage without

any manual inputs. And then from there,

it will basically start to execute the

test and then be able to send the

reports back to you by telling you

exactly what's broken. with normal

coding accuracy of 42% from other coding

agents. We can be able to improve that

with test MCP with the feature delivery

accuracy of 93%. So if you're interested

to try it out, you can check out this

video that I made or you can check out

the link in the description for more

details. Now one of the cool feature

that I realized from anti-gravity is the

generate commit message. So here you can

see that for the source control let's

say if I were to delete bunch of like

end to end testings or testing files or

images for the test results I can

actually be able to click on generate a

commit message based on the changes that

we have made and what's really cool is

that before I have to use AI here to you

know generate this based on the onstage

changes I have to write it down but this

one this time it's going to give us a

shortcut button here that can help us to

be more specific exactly on what are the

changes that we have made for the

onstage changes. So here you can see

delete all the steel results, the

reports and the associate tests as well

as the helper files for the end to end

testing. So in this case I'm just going

to commit this change and push the

change onto the branch here. You can see

that it has our uh commit message that

we just created with AI. All right, so

pretty much that's it for the review on

the anti-gravity from Google. And in

this video we went over how to use the

agent manager here. Basically try to

spin up multiple sub agents running at

the same time in the background. Right?

We can be able to have AI agent here

working on the same projects or a

different projects at the same time. And

the other one here is we can also be

able to use the browser agent from

Google Anti-gravity which can be able to

navigate to the application in a Google

Chrome browser. And it can also be able

to verify the changes taking screenshots

and try to identify where the issue is

and how to better fix it. And lastly, we

also went over how to select different

models. For example, we're not limited

to the Gemini models that we have, but

we also have the option to choose Claude

and also GBT model as well. And lastly,

we also went over a development workflow

here simply providing the prompts and

then we pass it to the anti-gravity

using the PLMO here to basically

generate a to-do list on how the AI

agent here is going to perform the task.

Then it's going to basically execute the

task using the model of our choice. And

then eventually we're going to use the

browser agent here to test it and

generate a report. And eventually we can

be able to take this report and continue

to prompt it and refine our application.

So pretty much that's what I showed you

in this video. And if you do found value

in this video, please make sure to like

this video, consider subscribing for

more content like this. But with that

being said, I'll see you in the next

video.