HomeVideos

10 Incredible Things ChatGPT Browser Use Can Do (Automate Everything)

Now Playing

10 Incredible Things ChatGPT Browser Use Can Do (Automate Everything)

Transcript

530 segments

0:00

You probably know that the Chat GPT

0:01

desktop app has a built-in browser, but

0:04

most people don't realize just how

0:06

powerful it really is. In fact, it's

0:08

probably one of the most slept-on

0:10

features inside of Chat GPT. The more I

0:13

use it, the more I find automations and

0:15

workflows that were previously either

0:17

impossible or extremely difficult to set

0:19

up. In this video, I'm going to show you

0:21

everything you need to know about the

0:23

Chat GPT browser to master yourself. And

0:25

then I'll share 10 use cases ranging

0:27

from the simple to the mind-blowing.

0:29

Let's start out by opening up the Chat

0:31

GPT desktop app. I want to start by

0:34

showing you how the browser works and

0:36

all of its features. If you feel like

0:37

you've already got that down, I

0:38

encourage you to stick around anyway

0:40

because I may surprise you with

0:41

something new. But if you're confident

0:43

you know everything about the browser

0:44

and you just want the use cases, you can

0:46

skip around using the timestamps down

0:48

below. Once you have the Chat GPT

0:50

desktop app open, click on this icon on

0:53

the top right to toggle open the side

0:55

panel. And you may see two choices in

0:57

here. You may see files and browser, or

0:59

you may just see browser. Obviously, we

1:01

are going to click on browser to open

1:03

the browser. And this is it. I'm going

1:05

to click on this button right here to

1:08

expand it to make it larger and easier

1:11

to look at. And this is pretty much a

1:13

standard browser. So up here, you can

1:15

enter in any URL. You can open up

1:17

multiple tabs, and you can use the URL

1:20

up here as a search bar. And just like

1:22

with a regular browser, there's a back

1:24

button, there's a forward button, and

1:25

there's a refresh. If this is your first

1:28

time using the browser, one thing you

1:29

may want to do is click on these three

1:31

dots over here on the right, and then

1:33

click on import cookies and passwords.

1:36

And this can bring in your browsing

1:37

history, your cookies, and your saved

1:39

passwords from the browser that you were

1:41

using before like Chrome. You can just

1:43

select the correct Chrome profile, and

1:46

then click on import. So yeah, it's a

1:48

browser, but unlike a traditional

1:50

browser, this has Chat GPT built in. So

1:54

at any point, you You click down here

1:56

where it says work with ChatGPT. That

1:58

will expand it and you can ask ChatGPT

2:01

questions about the tab that you have

2:02

open, about all the tabs you have open,

2:05

or you can even ask it to take over your

2:07

browser for you and start clicking

2:08

around. There's also an annotation

2:10

button up here. So, if you click on

2:12

this, you can now select elements on the

2:15

page and ask ChatGPT questions about

2:18

those elements. And you'll understand

2:20

more what this feature does and why it's

2:22

useful when we run through some of those

2:25

use cases in a second. But before we do

2:27

that, we have to look at the browser

2:29

settings and change some things there.

2:31

So, to unselect the annotation, you can

2:33

click up here where it says annotating.

2:35

That will unselect it. And now we can

2:37

come into the settings by clicking on

2:39

our name on the bottom left and then

2:41

click on settings. And then on the left,

2:44

under integrations, select where it says

2:46

browser. Now, there are a lot of

2:48

settings here, but there are only a

2:49

couple that I suggest that you change.

2:51

So, first, make sure you have this

2:53

toggled on where it says let ChatGPT

2:55

control the built-in browser. Then, if

2:58

you want, under general, I have where it

3:01

says web URL and link open destination,

3:04

I chose ChatGPT. Because what that will

3:07

do is if ChatGPT in the chat sends you a

3:11

link and you click on it, it will open

3:13

it up in the ChatGPT browser. If you

3:16

don't want that, you can select your

3:17

default browser and then it will open

3:19

Safari or Chrome or Firefox, whatever

3:21

you use. But I like using the ChatGPT

3:24

browser for that. That way I don't have

3:26

too many windows open. I also do the

3:28

same thing for the next option, which is

3:30

local URL open destination. Down here is

3:33

an option for annotation screenshots. I

3:35

just showed you annotations, but if you

3:38

have this enabled, this gives ChatGPT

3:40

more context when answering your

3:42

questions. It does say that it increases

3:44

your plan usage. However, I found it to

3:46

be very useful, so I leave that one on.

3:48

Scrolling down, we have some settings

3:50

for auto fill and passwords and

3:51

downloads. if you want to change the

3:53

download location. But, the really

3:55

important thing is down here under

3:57

permissions. First, look at approval.

4:00

This is where you can choose whether

4:01

ChatGPT can open websites on its own

4:04

without asking for permission, or if it

4:07

should always ask for permission before

4:09

opening a website. You can make that

4:11

choice for yourself, and we have a

4:13

similar choice for browsing history. You

4:16

can choose if ChatGPT can always look at

4:18

your browsing history, or if it first

4:20

has to ask permission to access your

4:22

history. But, there are always these

4:24

specific site permissions that override

4:27

those defaults. So, let's say you're not

4:31

yet comfortable giving ChatGPT full

4:33

access to the web without it asking you

4:36

for permission. So, you would select

4:38

right here where it says always ask.

4:40

But, let's say there are a couple of

4:41

sites where you do feel comfortable, all

4:44

you have to do is click on add here, put

4:46

in the name of the website, and then

4:48

choose the rule for it. So, for this

4:49

one, I want ChatGPT to always have

4:52

access to facebook.com. Let's say I have

4:55

a schedule set up where every day it's

4:57

going through my Facebook feed and

4:59

looking for potential leads. I want that

5:01

to be done automatically on its own, so

5:04

I need to give it permission to access

5:07

Facebook all the time. So, that's why

5:08

I'm doing that here, and then I can

5:10

click on add. And so, even though

5:12

ChatGPT normally under these settings

5:15

has to always ask for permission before

5:17

opening websites, if the website is

5:19

facebook.com, it no longer has to ask me

5:22

for permission. So, what's great about

5:23

this is that you can get very granular

5:25

with the controls. I'm going to delete

5:27

this one and switch this back to always

5:29

allow though, cuz those are the settings

5:31

I prefer. Okay, so now that your browser

5:33

is set up and you have a general idea of

5:35

how it works, let's dive into some of

5:36

the use cases so you can get a better

5:38

idea of what you can do with ChatGPT

5:41

browser use. This first use case is

5:43

pretty simple, but it's worth

5:44

mentioning. You can use the ChatGPT

5:46

browser as a browser. And because you

5:49

can open multiple tabs and it stays

5:51

logged in for you between sessions, it's

5:54

a pretty capable browser, especially

5:56

because it has Chat GPT built in. So,

5:58

when you come across something like an

6:00

article that you want AI to summarize

6:02

for you, all you have to do is click

6:04

down here where it says work with Chat

6:06

GPT and type in your question like,

6:09

"Summarize this article." Now, usually

6:11

when I'm using the Chat GPT browser, I

6:14

like to switch which the model slider

6:16

down so it's a little bit faster because

6:19

I don't want a slow model when I'm

6:21

browsing the web or doing things. I want

6:23

it to work a little bit quicker. When

6:25

you have it in this mode where it's full

6:27

screen like this, you'll be able to see

6:29

your thinking and sorry, the model's

6:31

thinking and the answer by clicking on

6:34

this chevron right here and that will

6:36

open it up. You can see right here it is

6:38

reading the article and pulling out the

6:40

main points. So, this is all of its

6:42

thinking and then here is the summary.

6:45

But, if you don't like this where you

6:46

have to expand this up and down, you can

6:49

always click on the top right here where

6:51

it says restore panel width and then

6:53

you'll have the browser over here on the

6:55

right and the chat over here on the

6:57

left. And you can hover over the split

7:00

right here and hold it down to make

7:02

either side as big or small as you want.

7:05

Now, for this second use case, what I'm

7:07

not going to do is I'm not going to show

7:09

you how Chat GPT can fill out forms for

7:12

you because that's what everyone else on

7:13

YouTube demos. And I don't think it's a

7:15

very good demo because browsers already

7:17

have autofill built in and that's much

7:20

faster. But, there was a similar

7:23

situation that came up for me recently

7:25

where Chat GPT really was helpful. Every

7:28

year for my business, I have to file an

7:30

annual report with my state and every

7:32

year I forget how to do that. There's

7:34

some website I have to go to and I never

7:36

remember where it is. This year, I just

7:38

told Chat GPT to start the process for

7:40

me and to get me as far as it could

7:42

before I had to take over. And because

7:45

ChatGPT already knows a lot of

7:47

information about my business, it was

7:48

able to get me pretty far. Of course, I

7:51

didn't just sit and watch it do this. I

7:53

had this running in the background while

7:55

I was working on another task. So, that

7:57

taught me that if there's a task that

8:00

needs to be done online and I would

8:02

ordinarily ask ChatGPT how to do it,

8:05

what I really should be doing is just

8:07

asking ChatGPT to do it for me. Then,

8:09

while it's working, I can spend time on

8:11

a task that I actually do know how to

8:13

do. For this third use case, I'm going

8:15

to show you another task that you can

8:17

offload to ChatGPT. Talking to customer

8:20

support. So, let's say I purchased a

8:22

drill recently from Home Depot that I

8:24

want to return. What I'm going to do is

8:26

just ask ChatGPT to do that return for

8:29

me. Can you use your browser to open up

8:31

homedepot.com and open up the live chat

8:34

and return the drill that I recently

8:36

purchased, the email information, the

8:39

information about it's in my email, I

8:40

mean. And while it's working, you may

8:42

have noticed that I specifically told it

8:44

to open up the browser. I find that that

8:46

often helps to either say that or to say

8:49

use browser use because sometimes it

8:51

tries to do everything through plugins

8:55

and it doesn't realize or remember that

8:57

it has access to a browser. So,

8:59

sometimes using that language can help.

9:02

Now, as we are talking, you will see

9:03

right here that this little window pops

9:06

up. This is a preview of what it's

9:08

doing. So, even if you have the side

9:10

panel closed and you're not able to see

9:13

the browser, you always see a preview of

9:15

it right here. If you wanted to, you can

9:16

hide it, but to see the full thing, you

9:18

can expand the side panel right here.

9:21

And then, here is the Home Depot tab,

9:23

which it has opened. So, it already

9:25

accurately founded found the chat right

9:27

here and it clicked on start a return.

9:31

And now, Home Depot's asking for the

9:32

order number. It correctly pulled that

9:35

and pasted it in. It got that from my

9:37

email, and now it's off to you actually

9:39

do the return. So, that's again

9:41

something I can have running in the

9:43

background while I'm working on

9:44

something else, and if a problem comes

9:46

up, ChatGPT will let me know, and then I

9:49

can take over the browser and fix any

9:51

issues. But, I've used this already, and

9:54

it's already been able to return a bunch

9:55

of stuff for me, which is really nice

9:57

because, you know, that's something I

9:58

could sit and just watch and do myself,

10:01

but it's nice when it's just automated

10:03

and running for me. For this use case,

10:06

I'm going to show you two cool things

10:07

that I use ChatGPT browser for. First,

10:10

you can have it open multiple tabs for

10:12

you, and with that, it's really good at

10:14

helping you compare different products

10:16

when you're shopping for something. So,

10:18

let's give it a test right here. Hey, I

10:20

have a Starlink a Starlink Mini that I

10:23

want to attach to the top of my truck

10:25

camper. Can you find like four or five

10:28

different products that would do that? I

10:30

want ones that have good ratings and

10:32

that would be easy for me to mount. Pull

10:35

all of them up as separate tabs in the

10:36

browser. And here we go, it's getting to

10:38

work opening up all five of these or

10:41

four of these four or five of these

10:42

different tabs with the different

10:44

products. And hope you're seeing like a

10:45

little bit of a theme here, right?

10:48

Because in the past, we used to ask

10:50

ChatGPT how to do things, and then we

10:52

would follow those instructions, but as

10:55

long as you can do it on the web, you

10:57

should just be asking ChatGPT to do all

11:00

of it for you. And here, it's not only

11:02

gone out and done research like in the

11:04

past, we would just have a list like

11:06

this where we would have to click and

11:08

open all of them. We don't have to do

11:09

that anymore. It now opens up all the

11:11

tabs for us, and we can see all of those

11:14

right here. But, of course, it works the

11:16

other way as well. So, if you were doing

11:18

research on your own, let's say you

11:19

didn't want to ask ChatGPT to do the

11:21

research, and you just on your own found

11:24

all of these different options, you can

11:26

then come into ChatGPT here and say

11:28

something like, "Create a table

11:30

comparing the features of all of these

11:31

mounts." And it can look at all the tabs

11:33

that you have open when it's writing its

11:36

response. And there we go.

11:39

I mentioned before how ChatGPT will

11:41

usually try to use a plugin before it

11:44

uses browser use. And that makes sense.

11:46

It really should do that. But there are

11:48

some situations where it can actually

11:50

use both together in a way that's very

11:52

beneficial. Here, for example, on the

11:54

left I have opened ChatGPT, and on the

11:57

right I have opened the Notion app. This

12:00

is not inside of ChatGPT. This is a

12:02

separate app. And if you're not familiar

12:04

with Notion, it's basically a

12:05

note-taking app, and you can see one of

12:08

my notes right here. So, ordinarily what

12:11

most people do is they use the Notion

12:13

app and they work inside of here. And

12:16

then, when they want ChatGPT to work on

12:19

a document or page that they've created,

12:22

they use the Notion plugin, which you

12:24

can find just by doing the at symbol and

12:26

typing in the name of the plugin, which

12:28

is Notion.

12:29

And then, they have to describe what the

12:32

page is that they want ChatGPT to help

12:35

them with. So, I may say, "Can you find

12:37

the page called ChatGPT versus Claude in

12:40

my video database and edit it to make it

12:42

shorter?" But there's an easier way to

12:45

do this. So, let's full-screen the

12:47

ChatGPT app, and I will come over to

12:50

Notion on my browser. So, now I have the

12:52

same page open in the ChatGPT browser

12:56

instead of inside the Notion app. And I

12:58

can use this the exact same way. I can

13:00

come in here and make edits. And so,

13:02

instead of using the Notion app, I'll

13:05

start using Notion inside the ChatGPT

13:08

browser. That way, if I want ChatGPT's

13:11

help, all I have to do now is come down

13:13

to ChatGPT down here and say to it,

13:16

"Edit this to make it shorter." I don't

13:18

have to tell it to use the Notion

13:19

plugin. I don't have to tell it which

13:21

page I want it to edit. It has all of

13:24

that context because it's able to see

13:27

what I have open in the browser. And

13:30

what's great about this is that it's

13:31

going to use the browser for context,

13:34

but it's going to use the plugin to

13:36

actually take action because that's more

13:38

efficient. And you can see that

13:40

happening right here. It's editing the

13:43

Notion page using the plugin. And then

13:46

over here on the right, you can see it

13:47

made those changes. Use case number six

13:49

is one of my favorite little hacks. You

13:51

can have AI use AI. So here in the

13:55

browser, I have opened grok.com and I'm

13:58

signed in. I don't use Grok all that

14:00

often, but I do find that it's helpful

14:02

when I'm researching trending topics

14:04

because it has real-time access to

14:06

Twitter. So what I do, you know, is I've

14:09

set up a skill called Ask Grok. So

14:12

anytime I ask ChatGPT to do some

14:15

research for me on a topic, it knows

14:18

that in addition to doing its own

14:20

research, it will open up the browser

14:22

and also ask Grok to do research and

14:25

pull in the results from that and

14:26

include that in its full analysis and

14:29

the report that it makes for me. It just

14:31

makes it a lot more comprehensive. Now,

14:34

Twitter has an official API, which I

14:37

could probably set up, but this was just

14:40

a lot easier for me to set up. I didn't

14:42

want to have to learn how to set up the

14:44

API for Twitter or pay for it either. So

14:47

I just use this and it seems to work.

14:50

This seventh use case is kind of

14:52

similar. Let's say you want to use

14:54

ChatGPT with a certain app or tool, but

14:58

when you come over to plugins, they're

14:59

not available here or they're not

15:01

available on the free plan of whatever

15:03

app you want to connect with ChatGPT. A

15:06

lot of times, you can just use browser

15:08

use instead. So for instance, I'm going

15:11

to come over to LinkedIn. Here we go.

15:13

And even though LinkedIn has a plugin,

15:16

it's quite limited in what it can do.

15:18

And it can't do what I need it to do,

15:20

which is to really look through LinkedIn

15:23

and look through my feed and find new

15:26

leads because I'm thinking about

15:28

starting an AI agency. But, with browser

15:30

use, ChatGPT can do that. So, if I send

15:33

that along, it will start that task and

15:35

it will take over this browser and start

15:37

looking around for me. And there we go,

15:39

it's starting to scroll. So, basically,

15:41

this works for any website where you

15:44

need to be logged in to really access

15:47

the information that you want to see and

15:49

that you want ChatGPT to be able to use.

15:52

So, for example, I've seen other people

15:54

use this for online support forums. So,

15:56

ChatGPT could compile a list of all the

15:59

biggest complaints for the software that

16:02

the the online support forum is for. Be

16:04

aware, though, that doing something like

16:06

this does sometimes violate the terms of

16:09

service for certain websites. So, I just

16:11

wanted to give you that warning. Use

16:13

case number eight, I haven't actually

16:14

heard anyone else talk about this

16:15

before, but you can give ChatGPT its own

16:18

email address, its own VoIP if you

16:21

wanted, or really any chat service that

16:23

you can access online. So, in addition

16:25

to having its own email address, it also

16:27

has its own Google Chat reminders, and I

16:31

have it often send me messages on here.

16:33

So, I will sometimes tell it, "Hey,

16:35

start a task for me, and then when

16:37

you're done, message me." And it knows

16:39

that message me means open up the

16:42

browser, open up Google Chat, and send

16:44

me a message on there letting me know

16:46

that you're done. And the reason I do

16:48

this is because even though I sometimes

16:50

get notifications on my phone from the

16:52

ChatGPT app when it's done with the

16:53

task, I don't always get them. So, this

16:56

makes sure that if it finishes

16:57

something, I actually get alerted about

16:59

it. So, really useful, so I'm not

17:01

sitting in front of my computer all day

17:03

waiting for it to finish. I can walk

17:04

away, get a Google Chat message when

17:06

it's done, and I can come back and see

17:08

the completed task. Again, there are

17:10

other ways to do this, but this is fast,

17:12

it's free, and it's easy. This ninth use

17:14

case is really helpful if you like to

17:17

use ChatGPT to vibe code websites or

17:20

apps because you can open it up in the

17:23

browser, look at it, and tell ChatGPT

17:26

what changes you want to make using the

17:29

annotation tool. So, let me expand this

17:32

and let's say I don't like some of the

17:35

graphics here like how I hover over this

17:37

and it changes to orange. So, what I'll

17:39

do is I'll click on the annotate tool

17:41

and I will select those elements. So,

17:43

that's that one. And if I hold down

17:45

shift, I can actually select multiple

17:47

elements. Most people don't know that.

17:50

And then you can add the comment right

17:51

here. So, I'll say, "I don't like that

17:54

this switches from blue to orange." Then

17:57

if you click on enter, that will be

17:59

added to the chat, but if you hold down

18:00

command and click enter, that will just

18:03

be sent to ChatGPT. You can see it's

18:05

working down here and working on those

18:07

changes. And as it's working, what it

18:09

can also do is on its own after it fixes

18:11

those changes, it can take over the

18:13

browser and look at it itself to make

18:16

sure that it worked properly. If it

18:19

hovers over these elements and they

18:21

still turn orange, it'll know that it

18:23

didn't fix it properly and it'll go back

18:25

and fix it. And finally, ChatGPT's

18:27

browser use is excellent for timing

18:30

online bookings. For example, I want to

18:32

reserve a campsite at Rocky Mountain

18:34

National Park. Some of these more ideal

18:36

campsites though can be pretty hard to

18:38

get. So, I'll just have ChatGPT take

18:41

care of it for me. Find out when Rocky

18:44

Mountain National Park opens up

18:46

reservations for their campsites. Then

18:49

schedule a task so right when the

18:52

availability right when the campsite

18:55

becomes available for the weekend of

18:57

September 18th, you come into the

19:00

website and reserve a campsite for me.

19:03

And now, instead of sitting at the

19:05

computer waiting for those campsites to

19:07

become available myself, ChatGPT is just

19:10

going to handle it for me. So, there you

19:12

go. That's how to get the most out of

19:14

ChatGPT browser use and 10 mind-blowing

19:16

things you can do with it. If you

19:18

enjoyed this video, do me a quick favor.

19:20

Give it a thumbs up and subscribe to the

19:22

channel because I make content like this

19:24

all the time showing you the best ways

19:26

to use AI tools. In fact, I have another

19:28

video just like that right here for you.

19:30

This one goes over the 10 concepts that

19:32

you need to master to get the most out

19:35

of the ChatGPT desktop app. So, if that

19:37

interests you, click on this video right

19:39

here and I'll see you over there in 1

19:40

second. Bye for now.

Interactive Summary

The ChatGPT desktop app features a powerful, often overlooked, built-in browser that can automate complex workflows directly within the interface. By integrating ChatGPT with browser control, users can perform tasks such as researching products, managing administrative forms, interacting with customer support, and even executing time-sensitive bookings, all while letting the AI work in the background. The video provides a guide on setting up the browser, configuring permissions, and utilizing ten practical use cases to enhance productivity.

Suggested questions

5 ready-made prompts