XAI and Cursor have just released a

brand new product called Grockbot.

Grockbot aims to allow us to build 247

always running virtual agents.

Essentially, these have their own

virtual computers that they run off,

enabling us to ideally automate far more

work than before. So, for this video,

I'm going to do a complete deep dive

into this product, going through

everything about it, so that you have a

full understanding as to whether or not

this is something that you should be

looking into. And by the end of this

video, I'm going to give you my complete

honest thoughts about this product,

whether or not it's something that you

should actually be signing up for or if

it's just a complete waste of time. And

this is something that you might be able

to do with some other tools already out

there. And before I dive in, this right

here is the virtual computer that this

agent is fully controlling by itself.

So, it's not just the browser. It's a

full-on computer with files and

everything. You can see right here, it's

actually doing some research uh into my

newsletter system on Beehive. You can

see it just loaded the page. If I go

down and click into the middle of the

taskbar with the file manager, you'll

see a whole file manager coming up all

specific to this particular agent

session that's running with its own file

system with images that it can download.

And obviously, this entire agent

computer is connected into a chat

interface, which is able to then

retrieve and pull information out of

that to obviously automate tasks and

pull back information for me to see. I'm

just quickly going to go through their

announcement page so you can see exactly

what this product is best used for. So,

real quick, meet Grockbot. AI teammates

you can give real work to. Bots can sign

into your tools, use them just like you

do, and come back with finished work.

You can see the interface right here

with the bots on the left hand side, the

chat interface right here. So, it looks

very similar to any other chat interface

that you've ever used for AI, of course.

But, as I scroll through this page right

here, really the main powerful part of

this is going to be the computer use

part of it. At least from my brief

amount of testing, this has probably

been the best integration of computer

use agents that I've probably ever used.

In this particular demo, it's asking the

user to sign into Salesforce in the

virtual computer. So, it has access to

that account rather than an API. It's

just giving it full access to Salesforce

to be able to click around and navigate

in their account from their perspective.

Scrolling down a bit further, message

bots like teammates. So, one really big

part of this is that they also have an

iOS app, which I will showcase in just a

minute, which I do think is a really

important part of this because these

virtual computers are virtual computers.

It's not running this locally on our

computer as computer use has been

through the likes of Claude Code or

Codeex. And just scrolling down a little

bit further, work with many bots at

once. Of course, you can create multiple

different bots. You can set them up all

with routines to be running essentially

24/7, constantly checking different

pages for different reasons and just

running and doing real work all the time

in the background. And one really

important thing that I am going to cover

in just a minute is showing a bot how

it's done. And we're going to be able to

essentially send a video of me doing a

piece of work on my computer to the

agent and it's going to be able to watch

back that video and fully recreate and

redo exactly what I was doing on the

computer. And so this particular feature

is honestly something that I have never

seen done before and no doubt about it

is going to save a lot of time when it

comes to just communicating how to do

particular tasks to an AI to get it to

replicate that. Okay, without further

ado, let's actually jump into Grockbot

to see exactly what it's capable of.

Now, to get Grockbot installed, all you

need to do is head over to their main

landing page, which is x.ai/bot.

And you'll see right here, download for

Windows, or if you're on Mac, of course,

it will just say download for Mac. hit

the download button and it's going to

install as a separate application on

your computer. This is not something

that we're going to be using through the

web. And once you've done that, it's

going to look something like this. So,

at the moment, I've already created

three different agents, which I'm going

to go through right now so you can see

exactly what kind of work I've already

been able to automate with this. And

don't worry if you do just want to try

this out. You don't have to pay anything

at all because they do have a oneweek

free trial. So, you can literally sign

up for free and start using it. Play

around with it and just see what you can

do with it. To be totally honest though,

it is very limited. you're not going to

be able to send that many messages

before hitting the session limit. So,

just be aware of that. Now, the first

thing that you're going to be prompted

to do is creating your first agent. If I

go to the top left and click on this

plus button right here, this is going to

allow us to create a new bot. And all I

need to do is just type in a name for

that bot. So, if I wanted to do a

schoolbot for the school.com community

that I have, I can just type schoolbot

and click create schoolbot. This is then

automatically going to load into the

chat interface right here. And right off

the bat, just based on the name that

you've given it, it's automatically

going to send a couple of initiation

messages to you. It obviously knows my

name. You can see here that it's doing a

little bit of thinking about the name

that I gave it. And we can see here that

it's already telling us, what do you

mainly want me handling on school? And

you can see it's already given me some

suggestions on what the school bot could

potentially be. Pretty much just because

I named it the school bot, it's already

been able to do this. Obviously, we can

just cancel out of this if we want to

just not go ahead with this right away.

But one thing that I can do is just

click on the name of the bot at the top

here. If I click into this on the right

hand side, it's going to allow us to

give it a little bit more information

about what this bot uh in particular we

might want it to do. So in my case, if I

wanted to create this bot to access my

school community from my account and

answer community questions

automatically, that is something that I

might be able to just type a title and

description for automatically here. And

then once I've updated that description,

the bot itself is going to obviously

know that information and then start to

work towards actually building that

automation out. And now that I've just

clicked out of that option on the right

hand side, you'll actually see that this

screen will pop up in the right corner

right here. You can see it says school

bots screen. And this is the virtual

computer that this individual bot now

has full access to. So if I just click

on the open button right here, it's

going to open up this virtual computer

in a full screen tab right here. And

here it is. It is a full operating

system that our agent is fully using.

It's got full file management as well as

Chrome installed. Now, we're also able

to control this computer, not only just

the agent. So, I can go ahead and click

on Chrome right here, and it should boot

up Chrome just like this. You can see

that I'm actually already logged into my

Google Chrome account as well. And that

is because it has shared context across

all of the other computers in this setup

as well. So just quickly jumping out of

this because in my previous session

where I was setting up a beehive

automation uh in this computer right

because I logged into my Google account

in order to sign up for this uh website

this schoolbot now has in its Chrome

browser access to my Google account as

well. So now it's able to fully access

anything uh that I do Google signins for

which is quite a lot of things. And so

it's pretty capable already. And just to

prove that this is an actual computer,

we can click in the middle part of the

taskbar right here and pull up an entire

file management system as well. So you

can see here we have a bunch of files

and file systems uh for our agents. Now

once again, this particular virtual

machine is shared across all of your

agents. So this is one computer that is

shared across all of your agents. So if

you ever need to delegate tasks between

one agent to the other, it will have the

full access to that same computer that

it's using. I have set up three

different bots already on this doing

three different types of tasks. So I'll

just quickly go through those. So this

bot right here is accessing Beehive,

which is a newsletter platform that I

use every single week for my weekly

newsletter. And pretty much what I want

to have happen is every time we do a

weekly newsletter, I want this bot right

here to come back and report to me all

of the statistics in terms of open rates

and click-through rates and overall

performance of the newsletter and just

read all that statistics for me and come

back and give me a summary. And so

essentially in order to set this up, all

that I did is I provided it with a

demonstration of how to go through the

beehive navigation on their website,

find the statistics, and then obviously

it's able to screenshot that, read it,

and then repeat it back to me. As you

can see here in my conversation with the

agent at the top here, you can see that

it was automatically able to know that I

wanted to do Beehive just because I

named it the Beehive bot. And then

you'll see my message on the right hand

side here which says learn from

demonstration. And so all that I needed

to do if I click into the computer

right, you'll see at the top of the sort

of right hand options, teach a task. And

if we click this, it will automatically

start a recorded session in this box in

our computer right here. we're going to

be able to click around and do a

particular thing, a particular task

right here. And then once we've done

this, it's going to be able to stop the

recording or we can stop the recording

and then we're going to be able to send

that off to our agent. It's going to be

able to watch that recording frame by

frame, see exactly everything that we

did and repeat that particular task as a

skill. So, we can see here from the

messages that it hit back, we did that

demonstration. It was able to read

through this. At this point, you'll see

here the video review hit a limit. That

was the limit of the free trial that I

was running. So, it's pretty limited

that free trial to be honest. But once I

upgraded the plan, then I was able to

continue on. And you can see here, it

turned that entire video demonstration

into a skill that it now has and can

repeat whenever needed. And so, I can

click into this orange highlighted text.

And this is going to pull up the private

skill, as they call it, for pulling

beehive post statistics specifically for

my use case. And we can have a look at

the instruction set here that it

automatically created by literally

watching the video of me doing that

task. And it's going to be able to do

this task over and over again. Uh

because it has all this information that

is recorded and it has this skill now.

And then once I created that skill, I

just told it to do that again, do that

particular task again. And it was able

to pull back all of those statistics by

going through that entire process. And

now anytime I need the latest statistics

on my latest newsletter, I can very

simply get this agent to do that. The

next bot that I set up is one that

monitors the Grock X account for any new

posts or announcements. So it would

immediately notify me of anything that's

come out so that I'm super informed on

all of the latest product releases from

this particular company. And essentially

all that I did is I said at the top

here, I want you to monitor the Grock X

account and send me a message every time

a new announcement is made on that

account. After this, it just went ahead

and started to build out that process.

It asked me how I wanted to pull through

those posts. And I pretty much just told

it to use the browser. So using this

computer that we have right here. And

then right after this what it did is it

automatically created a routine which

checks every 30 minutes on this

particular web page of any new posts or

any new announcements. And if there has

been a new announcement in that 30inut

block that it's checked then it's going

to automatically message me right here

about that announcement. So this is

essentially a 24/7 running agent that we

now have checking this particular post.

I don't have to ever come back to this.

It's going to run for me. I can exit out

of this and it's going to be running and

will notify me whenever. You'll see on

the right hand side that we have the

routines right here. So I can click into

a routine and we can see the entire

setup instruction set for this

particular routine. So we can scroll

through the instruction set that it

created in order to do this monitoring

every 30 minutes. We can see when it is

set up when to run. We can do this

manually if we'd like. But we can just

tell the agent to build it for us, of

course. And we can see the run history

here as well. You might see some X's

right here for these particular

timestamps. That's just because once

again my free trial of the rock ran out

whilst these were running. But once I

did upgrade my account, it obviously

started working again and was able to do

the checks. And then lastly, I set up an

email bot as well. I just told it that I

wanted to review my email inbox and it

then prompted me to connect my Gmail

through an MCP.

And then immediately this was able to

pull back a series of uh emails for me.

And if I just click on the plugins

button at the bottom left, you will see

all of the plugins that we can natively

integrate into very easily through an

oorthth connection. These are all of the

MCPs that are available. Very similar

connection amounts to what you'd expect

from Claude or OpenAI previously. So

this particular setup, not exactly super

revolutionary. Obviously, we're not

using any of the browser or virtual

machine automation in this case. So

nothing super revolutionary happening

here. I would say mostly the export

automation and the beehive stats uh is

more powerful because of that virtual

machine. Now, before I dive into setting

up this schoolbot, I just want to go

through a couple of my thoughts, at

least at this stage, as to whether or

not this is something that's worth

signing up for. Of course, at this

stage, it really does depend on how

you're using AI right now. So, at the

moment, if I were to automate any kind

of work or tasks on the computer through

Claude, essentially, what I'd have to do

is click into the plus button and

connect a series of connectors into

multiple different platforms and through

some sort of an MCP call, tell Claude to

add some information into my CRM or

extract some information. And that's

essentially the process that we'd be

using in order to automate different

tasks into those applications. And of

course, through this, we can also set up

routines very similar to what the

Grockbot has as well, where let's say at

9:00 a.m. every morning, we want to

check our email inbox and send me a

summary. Something like that is very

straightforward. It doesn't require a

virtual machine to monitor anything

because the task itself has enough MCP

and APIs able to fulfill that task. So,

there is certainly a lot that we can do

here through MCPs. But I do know that a

very large percentage 80 plus% I think

of the internet itself doesn't have any

APIs or MCPs certainly that we can

interact with. And so there certainly

are a lot of limitations when it comes

to just using this to automate all kind

of work that you want to do because not

everything can just be done through

these calls. And pretty much the exact

same goes when it comes to just chatting

through or automating work through

chatbt's main interface. Once again,

very similar. We can add connectors and

MCPs to this connection here. But we

don't have the ability to automate any

computer level work through this

interface right now. We of course have

the ability to do web search at the

bottom here. We can allow it to research

on the internet and find things and

extract information from websites. But

when it comes to account level

permissions and logging into certain

forms and clicking through multiple

different tabs and doing all of that

actual work, that is not something that

is natively available through one of

these platforms. Now, of course, one

thing that we can do is head over to a

tool like Clawude Code or OpenAI's

codeex where we do get a lot more

flexibility in automating our work

because these platforms are running

locally on our computer. One thing that

I can do is click the plus button down

here and do the browse the web option.

And this browse the web option

essentially opens up a brand new Chrome

tab in which it is able to navigate

through the page, click into different

options, log into our accounts, we can

help it do that. And so then we do start

to get that computer use through this

particular feature. But once again, this

is not a virtual machine. We aren't able

to get any kind of a file manager to

download files or anything like that.

And to be honest, my experience with

this browse the web feature has been a

little bit janky through Claude to be

totally honest. We ourselves have to

manually initiate that. So I have to

come here and tell it to browse the web

in order for it to do it. And because

this is running locally on our computer,

if you've got a slow computer, this

might be a little bit tricky to run. And

you most certainly can't obviously share

that context with your phone or anything

like that. And one more thing to mention

is that because this is sort of a

sandboxed virtual computer, it is quite

a bit safer as well. Because if we

wanted to maybe download a CSV file from

some sort of a website and automate that

or change some information or extract

information out of that, we're going to

be downloading that into our virtual

computer and manipulating and extracting

and deleting files all within our

virtual computer as opposed to our own

local computer. So it can just help to

avoid that risk of your claude agent or

OpenAI codeex agent accidentally

deleting the wrong file or you know

missing files and all that stuff.

Anyways, I'm going to go ahead and see

if we can automate responding to school

posts in my school community. So what I

might highlight is community engagement.

I'm going to click on this and we'll see

what we can do when it comes to uh

responding to school posts in my school

community. It's now asking which one to

use. I'm going to go ahead and just get

the link for this. So, I'll just copy

the link to my school community, right?

Jump back into the agent and just paste

it right here. Submit. And we should see

on the right hand side with our computer

agent. It says it's opening it right

now. We should see that this link gets

opened on our school bot. We can see

it's actually loading at the moment. We

should see that page come up. There we

go. The page is now loaded as well. So,

of course, we will need to log into

this. We can see that we're not

currently logged in at all, but I can

obviously now manually log into that

account. And now it's asking me how it

wants to reply to the members where it's

going to do a draft. It's going to do

post simple replies and just ask for

tricky ones or it's going to tell me

every single time. I might just say for

this case draft for my approval. We can

actually see that the agent is scrolling

through all the different posts in the

community at the moment. Uh anyways, but

I'm just going to tell it to wait for my

approval at the moment. Okay, we can see

here that it's actually gone ahead and

just read through a bunch of posts in

the community already and it's asking to

do a draft reply already. Uh it's not

signed in, but it seems like it just

wants to start to uh start to respond to

these questions already. So, what I

might do is just click on one of these

posts. This post is about price points.

So, maybe I'll click on this one and

just see if we can do a draft reply. And

in the meantime, what I would probably

do is I'm going to click into open. And

then I might just quickly log into my

school account. So, I just logged into

my account. We can see here it's now

changed to my account. And I'm just

going to let it know that I've just

logged in. And so, now that I have

logged into this particular virtual

machine right here, it should have that

uh in its session cookies. So every time

that I need to run something on school,

any one of these virtual computers that

we have is going to be able to access my

account uh and it's going to be able to

do whatever kind of automation that I

need. So at the moment, I'm essentially

just letting the agent uh go through

school and just figure out everything

that it needs to do in order to complete

this task. But what I probably would do

if it was a more complicated task or one

that just wasn't super obvious is that

what I would do is I would pretty much

just click into the the virtual computer

here and do teach a task. And this would

allow me to once again uh go through the

virtual computer myself and record

exactly how that task should be done. So

it can then see exactly how to do it

itself. You can also click at the bottom

left on the plus button, right? And then

you'll also see teacher task is also

down here as well. And it just does the

same thing. So, for this particular

automation setup, what I would probably

do is tell the agent to let's say every

morning at 7 a.m. to create a routine in

which at 7:00 a.m. every morning check

my school community to see if there's

any new posts that have come through.

There are any new posts to maybe

automatically respond to some of the

posts and try to help out with whatever

their questions are. And this is

something that really does require a

virtual computer or some sort of

computer use in order to do this as best

as possible. And this is because

school.com especially is just a platform

that has got notoriously bad APIs and

just missing APIs at all. They certainly

don't have an MCP server and so

automating anything on school is very

very tricky. And this is the case for a

lot of platforms out there on the

internet. Like I said, majority of

websites and things that are done on the

internet do not have easy to access or

easy to use uh APIs or MCPs in order to

do specific tasks. And to be totally

honest, it's not fully feasible in order

to just use APIs or MCPs in order to do

everything because we ourselves use

computers through computer use. I would

certainly say that this is going to be

the future of just automating computer

work is literally just getting the agent

to use a computer to do the work. Okay,

we can see here it's now gone ahead and

given us the draft reply for exactly

what it wants to send a reply for. And

we can see that we're getting the option

here now to post it or tweak it or

change it or whatever. I'm going to go

ahead and click post it. I'm probably

going to delete this right after, but

just to prove the point, we should see

in the computer. I'll click into this.

Uh, it should go through and reply to

this post all automatically without me

having to do anything. It is now opened

up the post once again, which is great.

We should see it scroll down a bit, I

believe, and then reply to the post.

Okay, so it's now scrolled down to the

bottom. And in just a second, I assume

it's going to paste in that comment and

then hopefully post the comment as well.

Okay. So, we can see it is now typing

that comment at the moment. You can see

it just did the first sentence right

there. And now it's just added in the

rest of the comment there as well. And

it should in the second send this off. I

believe it looks like it actually failed

to send it. Not sure what happened here.

I think it might have accidentally

clicked the cancel button. So, you can

see it's obviously not perfect at all at

doing everything uh right away. Just

looks like the message just disappeared

just then. What I would probably

recommend doing is going through and

doing the teacher task first. So, maybe

I can do that now. Now, if I click into

the open button and I do teach a task,

what I might quickly do is just type a

comment. And then if I hit the comment

button right here, if I just click

comment and then we've got that sitting

there. So, what I might do is now pause

the recording and we should see that the

learn from demonstration has now sent it

off uh to the agent right here. I am

going to quickly delete that post. But

hopefully from that recording that I

just sent it, it should now know to

click the comment button specifically

and just get a better understanding as

to how to make a post. But we can see

here it says watching the demo you just

recorded learning the steps now. So

obviously for your more complex tasks

would definitely recommend doing this

essentially without doing this. The way

that it works is that it pretty much

just clicks around different buttons the

things that it just assumes that what it

should click and then just looking at

all the button availabilities that it

gets. Okay. We can see here just from

that recording uh is actually created a

skill around that. If I click on this,

you'll see that we now have a comment on

a school post skill that's been created

with all the instructions set on how to

exactly do that. It's provided us with

the steps as well. In this case, I'm

just going to tell it to finish posting

the actual reply. And it does look like

it's doing that at the moment. Uh if I

click into the agent here, we can see

that it is beginning to reply. Okay, so

it is still trying to reply to this

post. To be honest, I'm pretty

disappointed with the results of it

trying to reply here to this post. Uh it

seems like it's kind of just stuck at

the moment. It says that it had to uh

restart the page because the composer

has stalled. You can see there. Not

exactly sure why this is such a

difficult task for the agent. So, not

exactly sure what's going on at the

moment. Hopefully, it's going to be able

to post it in just a second. Uh but

maybe a bit of a limitation of the agent

at the moment. Okay. So, finally that

post has gone through. I did take a

little bit of pushing to get it to work,

but if I click open, you'll see that the

post has now gone through and sent

finally from this agent. Kind of

disappointed to be totally honest in its

ability to to make that post. It seems

like it was continuously missing the

comment button to actually send the post

off and would just keep sort of

refreshing the page in order to do this

task. So, a little bit disappointed. Not

sure if it's just this web page

specifically that it's just been

struggling with. But something to keep

in mind when it comes to this kind of

browser automation is that it is

definitely not perfect right now. But I

do hope that now because it has done the

post once and it's fully understood

exactly how how to do it. Then when I

need it to do another post, hopefully

it's learned from its first uh iteration

and then it's going to be able to do it

much better over time. Okay, the next

thing I wanted to quickly cover is also

the mobile app a part of this as well.

So I would say that the mobile app as

I've been testing it so far is actually

a really big component of this app.

Essentially what we can do is just

download Grockbot from the app store

directly onto our phone. log into my

account. And hopefully you can see on my

phone screen right here, all of the

sessions of my agents are on the phone

app right here automatically. And if I

were to click into the bot itself, you

can see the conversation of the agent

that I was having on my computer all

loaded here because this is all

happening in the cloud and it's not

happening locally. And the thing that I

think is the most cool about this is

that if I click into the top right of

the phone, if I show you right now,

you'll see that it is connecting into

the virtual computer of this particular

agent. And you can see right here, this

is now connected to that computer agent.

And so if I jump back onto my computer,

if I click open of this particular agent

right here, what I can do is now go onto

my phone. And what I can do is now

control this computer from my phone. If

I click onto the plus button real quick,

there we go. I just added a new tab into

Chrome from my phone and now that's

updated on the virtual computer on my

computer right here. And you can see

that this is what's popped up. The

keyboard's popped up so I can type stuff

as well. But I can mostly just click

around this web page. And so that's

super awesome that we can monitor and

watch this computer that is running

autonomously for us 24/7 in the

background. We can control it with a

keyboard. We can do whatever we need to

from our phone. And this is happening

even with our computer fully turned off.

And this obviously means that we can

send our standard messages into the

agent as well. So I can go ahead and

send it a message to uh you know do some

sort of a task on the computer from my

phone. So I can literally control an

entire computer uh yeah from my phone

without having to even have a computer

of my own. So there really are some

pretty epic use cases here. This is

something that probably wouldn't replace

my claude code or my open AAI codeex

system, but definitely if I was on the

go, I could use this mobile app to

initiate some sort of a task uh into

this virtual computer to do something

for me. And the seamless integration

between this web app and the phone app

as well is just so good. We haven't

really seen a seamless integration like

this quite yet. So definitely quite a

big plus when it comes to this feature

is going to be that mobile app and just

the seamless integration to be able to

communicate with all of these computers

and all of the bots uh with a seamless

integration is pretty awesome. Now, as

for the best use cases for this right

now, I'd probably say it's not going to

be used for building out any sort of

applications or anything too intense

like that. You'd still be using your

Claude code or your OpenAI codeex for

that. It's mostly going to be best used

for the use cases that I've showed here

in terms of monitoring pages or checking

in on things periodically or doing some

sort of a task on a computer with

existing setup systems. So, you wouldn't

build an entire CRM through this.

Although, maybe you'd use it to navigate

through your CRM that you have built. Or

if you've got multiple different

softwares or you need to move

information around different areas, you

could just come to your agent writer

which has got that computer, send that

task off. It's going to be able to do

that. But this is pretty much it when it

comes to this new Grockbot product. My

current honest thoughts on this is that

it's definitely a pretty promising

product or approach when it comes to

just overall automation of work. I do

think that the best way to, you know,

automate this online work is literally

just by using the browser or using a

computer exactly how we do rather than

reinventing the entire internet around

brand new MTPs and APIs to access these

web pages. makes much more sense to me

if we can just get the agent to use the

computer instead exactly like how we've

been using it in the past. And it seems

like the Grock bot setup right here is

taking that sort of computer first

approach to automation. So once this

does get really really good, there's no

doubt about it. Using an interface like

this is going to allow us to literally

automate anything you could possibly do

on a computer. But there definitely is a

way to go for it to be perfect. So to be

honest, I probably wouldn't recommend

this right away right now. And pretty

much the main reason for that is that

this is still very expensive. If I

scroll down to the bottom here, the

pricing for this at the moment, at least

at the minimum, is $200 a month. Now,

this gives you Cursor Ultra. So, you're

not just uh paying for Grockbot, of

course, for $200 a month. You're getting

access to the entire Cursor Ultra plan.

So, this is obviously a pretty powerful

plan when it comes to AI usage for this

amount. Of course, you're getting a lot

of AI tokens. So, if you are somebody

that is hammering Clawude or OpenAI

pretty hard at the moment, obviously

paying for a Cursor Ultra plan is going

to give you a lot of token amounts and

usage just like your current OpenAI or

Clawude plans give you. But, of course,

this does mean that you'd probably want

to swap from your Clawed subscription to

your cursor subscription instead because

of course it is quite expensive. You

wouldn't want to be running multiple of

these subscriptions, especially if

you're getting the same amount of token

usages. But if you are looking to

literally just set up Grockbot,

obviously paying $200 a month just for

Grockbot is probably way too much. And

you can get quite a bit of the way there

just by using the traditional browser

automation individually. For example,

through claude code, we can obviously do

a browse the web here, which just plugs

into Chrome and then allows it to

navigate through a web page. But the

problem with that is that we once again

have to manually sort of open up that

web page on our computer. It only runs

locally. we can't do that 24/7 as

opposed to a entire virtual machine

which is what this is which can run

24/7. I can turn off my computer and

this system will still run in the

background. So there definitely are some

advantages to it. I definitely think

it's a step in the right direction. But

what I would like to see is this kind of

feature or product uh integrated into

the likes of Claude Code or Codeex in

the future or maybe XAI will integrate

this product more deeply into their

Grock build system. Anyways, I hope

you're a lot more informed now about

what Grockbot is and whether or not you

should or shouldn't use it. Just for my

last note, I recently discovered that

90% of you watching right now are not

subscribed to the channel at all, which

I think is pretty insane. So, if you

could just scroll down a little bit, hit

the subscribe button, like the video,

turn on the notifications, that would be

amazing. I've got a lot more content

coming on all of these new releases that

are coming out, as well as all the

different ways that we can effectively

utilize and leverage all the existing

platforms out there like Claude Code and

Codeex.