HomeVideos

Give Me 28 Minutes, I’ll Give You 10,000 Hours of ChatGPT Knowledge

Now Playing

Give Me 28 Minutes, I’ll Give You 10,000 Hours of ChatGPT Knowledge

Transcript

922 segments

0:00

So, look. If you're happy doing the same

0:02

boring, repetitive tasks every day for

0:05

the rest of your life, well, then you

0:06

can click off right now because the next

0:08

30 minutes are for people who want out

0:10

of that. You see, most people use

0:12

ChatGPT to plan their groceries or to be

0:15

their therapist after 5 hours of

0:17

doomscrolling on TikTok wondering why

0:19

life is so tough. And look, if that's

0:21

you, I hate to say it, but you're

0:23

probably only using about 2% of what it

0:25

can actually do. And that's not your

0:27

fault. I mean, this thing doesn't come

0:28

with an instruction manual. Now, if

0:30

you're happy paying 100% of the price to

0:32

use 2% of its features, well, then be my

0:35

guest, but this video probably isn't for

0:37

you because today I'm going to show you

0:39

how to learn 98% of ChatGPT in the next

0:42

20 to 30 minutes. I'm going to give you

0:43

my favorite beginner, intermediate, and

0:46

advanced tips that will help save you

0:48

time and get ahead of 99% of people who

0:50

pay to use it the wrong way. So, let's

0:53

start nice and easy. You see, every

0:55

single time that you open up ChatGPT and

0:58

you ask it something, you're literally

1:00

talking to a stranger. I'm not joking.

1:02

You see, by default, ChatGPT knows

1:04

nothing about you, nothing about your

1:05

job, what you do, how you like things

1:08

explained, what you don't like,

1:10

literally [music] nothing. So, imagine

1:12

you hire a personal assistant. This is

1:14

the smartest person you've ever met, and

1:16

they could help you with almost

1:17

anything, but every morning they wake up

1:20

and forget everything about you, who you

1:22

are, what you do, how you want things

1:24

done. Look, you wouldn't want that. It

1:25

would drive you insane, right? And yet,

1:27

that's exactly what most people do when

1:29

they use ChatGPT. They open it, they hit

1:31

a new chat, and they start from scratch

1:33

every single time. Now, what if I told

1:35

you there was actually a setting in here

1:37

that changes all of that, and it only

1:39

takes about 1 minute to fix this? Well,

1:41

congratulations because it exists, and

1:43

it's actually called custom

1:44

instructions. Here is where you find it.

1:46

You're going to go to settings, then

1:47

you're going to go to personalization,

1:49

and in there you're actually going to

1:50

see the custom instructions right there.

1:52

Inside of it, add what you want ChatGPT

1:54

to know about you. Tell it who you are,

1:56

what you do for work, what you're trying

1:58

to get done, and anything else that

2:00

helps it understand the context. And

2:02

then, tell it how you want it to

2:04

respond. The more specific you can be

2:06

here, the better your outputs are going

2:09

to be. Do you want short, direct

2:11

answers, or do you want it to explain

2:12

its thinking? Do you want a certain

2:14

tone? And just as important, what do you

2:17

want it to avoid? Now, look, I know that

2:19

sitting there trying to write all of

2:20

this from scratch is a lot. So, what

2:22

I've gone ahead and done is I've

2:24

actually added my prompt that I use

2:25

below this video. I actually use this to

2:27

set up my own ChatGPT. It interviews you

2:30

one question at a time, and [music] it

2:32

actually hands you back exactly what to

2:34

paste into that box. So, it's in the

2:35

description section below. I want to

2:37

say, I know that this is tedious work,

2:39

but trust me, once it's done, it's out

2:41

of the way, and you will thank me later

2:43

for showing you this hidden feature.

2:45

Now, once you've done that, there is

2:47

actually a second layer sitting right

2:49

next to it that you also may have missed

2:51

as well. And people mix these two

2:52

[music] up all the time. So, I want to

2:54

make the difference clear inside of this

2:55

video. Custom instructions are what you

2:58

tell it once, and whatever you add there

3:00

stays there unless you go back in and

3:02

actually change it. Memory is different.

3:04

Memory is actually what it picks up over

3:06

time, the more that you actually use it.

3:08

I want you to think about it like this.

3:09

Custom instructions is like giving

3:11

someone you work with a one-page brief

3:13

on day one. Here's who I am, here's what

3:15

I need, here's how I like things done.

3:17

It's as applicable on day 100 as it is

3:20

on day one. Memory is actually what

3:22

happens from that point on. It's every

3:24

conversation, it's every project, it's

3:26

every time [music]

3:27

that you corrected it. Every time you

3:29

said, "No, not like that. Like this

3:31

instead." It's basically like you

3:32

training your ChatGPT. And with this

3:34

feature, it only gets better over time.

3:36

But here's the weird part, though. Most

3:38

people never even bother to turn the

3:41

memory feature on, which is absolutely

3:43

insane to me, because why not make your

3:45

ChatGPT better and stronger every single

3:47

day? It's like hiring someone knowing

3:49

that they will never get better at what

3:51

they do. It makes no sense. So, here is

3:54

how to activate the memory feature.

3:55

You're going to go to settings, then

3:56

you're going to go to personalization.

3:58

Inside of it, you're going to see memory

4:00

sitting right there underneath the

4:01

custom instructions. Please make sure

4:03

it's switched on. And there's actually

4:05

quite a cool feature here as well. If

4:06

you're curious or you ever want to see

4:08

exactly what it's picked up about you,

4:10

you can actually go to manage memories

4:12

and it's all there. And if something's

4:13

ever wrong, the cool part is is that you

4:15

can manually go in and edit it yourself.

4:17

But, people have a big misconception

4:19

with this. They think that ChatGPT only

4:21

gets better when OpenAI actually updates

4:23

their models. And sure, that's partly

4:25

true, but sometimes it's also true that

4:26

it gets worse. We've all been there.

4:28

They've updated a new model and we're

4:29

like, "This was so good last week. Why

4:31

did we ruin it?" But my point is is that

4:32

this is actually another way that you

4:34

can upgrade your ChatGPT. It's a mini

4:36

upgrade, if you will. Okay? With this

4:38

feature, you are training your ChatGPT

4:40

to slowly learn how you think. Now,

4:42

sure, it's not the most groundbreaking

4:43

feature I'm going to share here today,

4:45

but it's easy to set up and it will make

4:47

your experience 10 times better. So,

4:48

with that said, let's move on to

4:50

probably one of my favorite tips in

4:52

today's video. Tip three is devil's

4:53

advocate. Now, this last beginner tip

4:55

will pretty much go against everything

4:57

that every other video tells you to do,

4:59

or better yet, just forgets to tell you

5:00

to do. It's what I call devil's

5:02

advocate. And honestly, this is one of

5:04

the most important beginner tips. Let me

5:06

ask you a question. Have you ever used

5:07

ChatGPT and felt like it's just agreeing

5:09

with everything that you say? You're

5:10

going to type in, "Am I right in

5:12

thinking that the world is flat?" And

5:13

it's going to find a hundred ways to

5:15

agree with [music] you. Look, I've been

5:16

there before. And that's why for this

5:18

final tip, I want you to force your

5:19

ChatGPT to disagree with you. Force it

5:22

to play devil's advocate. You see, most

5:24

people are using ChatGPT for validation,

5:27

to stroke their ego. As I said, you

5:28

could type in the most egregious

5:30

question and ChatGPT will still find a

5:32

way to go, "Yes, you're absolutely

5:34

right. That was a great question." And

5:36

look, if you think about it, the reason

5:37

it does this makes total sense, okay?

5:39

OpenAI is a business and just like any

5:42

other good business, what is their

5:44

number one main objective? It was built

5:46

to agree with you, to keep you happy, so

5:48

that way you keep coming back. But,

5:50

that's also dangerous, which is why if

5:52

you don't know what you are doing,

5:53

ChatGPT is just going to run laps around

5:56

you. And if you use AI just to agree

5:58

with everything that you already think,

6:00

well then you're just training yourself

6:01

to trust [music] it without any filter,

6:03

which is exactly what I want you to

6:06

avoid. You still [music] need to be the

6:07

person steering the ship, the one in

6:09

control. Look, I'm going to give you an

6:11

example of this. One of the questions

6:13

that I always ask my team here at our

6:15

company consultant.com is, "What are my

6:17

blind spots? What am I missing in the

6:19

situation?" Now, you might be wondering,

6:21

"Okay, why do you like to ask your team

6:22

these questions?" Because I know that

6:24

sometimes I don't have the full picture.

6:26

So, before you ask ChatGPT anything

6:28

important, tell it what to do. Tell it

6:31

to tell you your blind spots. Tell it to

6:33

play devil's advocate. Don't just ask a

6:35

question hoping they will give you the

6:37

answer that you want it to give you. You

6:39

can explicitly say things like, "Don't

6:41

agree with me, play devil's advocate,

6:43

challenge my thinking," or things like,

6:45

"Give me the strongest possible case for

6:47

why I could actually be wrong." And

6:49

look, when you do this, you go from

6:50

speaking to a yes man to someone who's

6:52

actually going to challenge your

6:53

viewpoints and your worldview, so that

6:55

way you can make better decisions and

6:57

make much more well-rounded conclusions.

6:59

And by the way, while we're on this

7:00

topic, one more thing I want you to be

7:02

aware of. ChatGPT is known for

7:05

hallucinating. [music] It will

7:06

occasionally get things wrong with 100%

7:08

complete confidence. But, the problem

7:10

is, how do you know when it's

7:11

hallucinating? You don't, which is why

7:13

you have to be hyper aware of this. So,

7:15

here's a system that I have personally

7:17

developed to help me detect when my

7:19

ChatGPT is hallucinating. At the start

7:21

of any chat before I do anything else, I

7:23

will always tell it to refer to me by my

7:25

first name. Then, if it ever answers

7:27

questions without, "Great question,

7:29

Eman," for example, then that is a sign

7:31

that it's probably losing the plot. At

7:33

that point, I'm actually going to stop

7:35

and double-check all the information

7:36

that it gave me, and that is how I keep

7:38

it in check. All right, there we go. So,

7:40

with all that being said, that is the

7:42

beginner tips done. Most of them, once

7:44

you sit down and do it, are pretty

7:46

simple. You've got custom instructions,

7:48

memory, keeping your chats organized,

7:50

and forcing it to play devil's advocate.

7:52

Those are the foundations. Now, we're

7:54

stepping it up a notch with the

7:56

intermediate tips. So, now that we know

7:58

how to get ChatGPT to understand us

7:59

better, let's look at how we can

8:01

actually understand ChatGPT better. You

8:03

see, ChatGPT isn't just a tool. It's

8:05

really a stack of different modes, where

8:07

each mode serves a different purpose.

8:09

And understanding what modes are best

8:11

for each task is what will get you

8:13

closer to learning 98% of ChatGPT.

8:15

Knowing this alone can put you ahead of

8:18

98% of people, because most people are

8:20

not switching modes. They are just using

8:23

the standard mode for every task. There

8:25

are actually six modes that I want you

8:27

to pay close attention to. There's voice

8:29

mode, canvas mode, study mode, agent

8:31

mode, image mode, and deep research

8:33

mode. But, before we continue, I could

8:36

sit here and give you all the tips and

8:37

the tricks in the world, but what's the

8:39

point if you haven't got a way to take

8:40

advantage of it? So, today, I'm going to

8:42

give away another MacBook. To win the

8:44

MacBook, all you have to do is watch the

8:46

rest of the video and let me know which

8:48

tip did you find most useful. And by the

8:50

way, if your comment is an AI comment,

8:52

trust me, I can tell, you will not be

8:54

selected. So, watch the rest of the

8:55

video, leave your favorite tip and why

8:57

was your favorite tip, as well as leave

8:58

a like, and that's all you have to do.

9:00

I'm going to pick a winner and post it

9:01

as a pinned comment and in the

9:03

description of next week's video. Oh,

9:05

and by the way, the winner from last

9:06

video is actually in the pinned comments

9:08

below, along with instructions on how to

9:10

claim it. So, with that being said,

9:11

let's actually start with the first

9:13

intermediate tip. This is one that fixes

9:14

something that's been a real pain for

9:16

you in the past. You see, you ask

9:17

ChatGPT for something and it comes back

9:19

with something close to what you wanted,

9:21

but not quite there. Maybe one paragraph

9:24

needs a little bit of work, maybe the

9:25

tone is slightly off in one or two

9:27

sections, and normally, what would you

9:29

do in this situation? Well, just like

9:31

most people, you're probably going to

9:32

spend time retyping the whole thing,

9:34

hoping that it fixes just that one bit.

9:36

And half the time, it changes another

9:38

section that you were actually happy

9:39

with. And look, I know how annoying that

9:41

is because it's happened to me so many

9:43

times. And look, I don't think I'm the

9:44

only one, right? You see, that was until

9:47

I found this small feature called canvas

9:49

mode. It works just like, I guess you

9:51

would say an artist canvas does, which

9:52

[music] is you're not trying to finish

9:54

the painting, but you're focusing on the

9:56

place where you can shape the work to

9:57

your liking. What it does is that

9:59

actually opens up an additional editing

10:01

workspace right next to your chat. And

10:02

instead of just getting a wall of text

10:04

back, you can actually get something

10:05

that you can work with directly. You can

10:07

refine and tweak as you like. You can

10:09

highlight the bits that you want

10:10

changed. You can ask for just that

10:11

section to be rewritten. You can adjust

10:13

the tone. And it only touches the bits

10:16

that you ask it to amend. Everything

10:18

else can be kept the same. Now, once

10:19

you've actually used this a couple of

10:20

times, going back to the old way of just

10:22

retyping the whole thing is going to

10:24

feel absolutely absurd. Trust me, you

10:26

will never do that again. The problem is

10:28

most people don't even know that this

10:29

mode exists. And that's why they only

10:31

end up using 2% of what ChatGPT actually

10:34

has to offer. Now, speaking about

10:36

getting more out of ChatGPT, here's one

10:38

that might actually remind you of your

10:39

time at school. Einstein once said

10:41

something along the lines of education

10:43

is actually what remains after one has

10:45

forgotten what they actually learned in

10:47

school. Look, I never even finished high

10:48

school, but I would say that I've

10:49

learned a lot in my life.

10:51

>> [clears throat]

10:51

>> And that's the point. Real learning

10:52

isn't what you can repeat the day after

10:54

you read it. It's what's still in your

10:56

head weeks or even years after you've

10:58

learned it. Look, do you know that

10:59

feeling where you read something or

11:01

maybe you watch a video and you walk

11:02

away thinking, "Yeah, I I I get that."

11:05

Then someone actually asks you question

11:06

about it and you realize that you only

11:08

understood half of it. That is the

11:10

difference between feeling like you know

11:11

something and actually knowing it. And

11:13

that is exactly what study mode allows

11:16

you to do. If you're learning a new

11:17

topic, you can actually ask study mode

11:20

to actually give you a breakdown of that

11:21

topic and then it will test you on it.

11:23

It's actually going to give you some

11:24

multiple choice questions to test if you

11:26

really understood the topic. Look, this

11:28

one reminds me of being back at school

11:29

taking an exam. Not something that I

11:31

would say I particularly miss, but

11:33

definitely a useful tool that you can

11:35

use to make sure that you speed up your

11:37

learning process. I don't use this

11:38

feature too much, but when I really want

11:40

to learn something quickly and make sure

11:42

that it sticks, then this is the first

11:44

thing I go to. So, to actually activate

11:46

this mode, all you have to do is click

11:47

the plus button, type in @study, and

11:50

then select the study mode. Then,

11:51

literally paste in whatever it is that

11:53

you're trying to learn, and it will take

11:54

it from there. All right, so next up is

11:56

image generation. If you saw my Claude

11:58

video, maybe you remember me saying that

12:00

image generation was actually one of its

12:02

weak spots, whereas with ChatGPT, it's

12:04

one of its biggest strengths. So, I

12:05

actually asked it to create an image of

12:07

me, Iman Gadzhi, drinking coffee just to

12:09

see what it would do, and the output

12:10

actually kind of surprised me. It looks

12:12

like me, and I actually get my editors

12:14

to throw it up on screen. Now, to use

12:16

it, here's what you're going to do.

12:17

You're going to describe exactly what

12:18

you want. And then look, if it gets

12:20

close, but something's slightly off, you

12:22

don't need to start over. Just tell it

12:24

what to change in the same chat, similar

12:26

to what I showed you with canvas mode,

12:28

and you can actually also upload your

12:30

own photo and have it edited from that.

12:32

Now look, to be honest, I don't use it

12:33

too much. Well, when I was doing the

12:34

interior design for our office, I

12:36

actually did use it for that function,

12:37

but in general, I'm not really sitting

12:39

here creating images all day. And

12:41

honestly, my designers have told me that

12:43

it's not really replacing their design

12:44

work, either. But for the average person

12:47

who just needs some quick images here

12:48

and there, it actually does the job

12:50

quite well. Now, the last three modes,

12:52

which was canvas, study, and image, are

12:54

quite straightforward, because you can

12:56

use them easily without changing your

12:57

settings. This next one is a small step

13:00

up, which is actually why I left it for

13:02

a little bit later on in the

13:03

intermediate category, but trust me,

13:06

it's worth it, because it completely

13:07

changes how you do work. Look, if you're

13:09

anything like me, you hate typing, okay?

13:11

I hate it with a passion. My team knows

13:13

it. I would rather just get on the

13:14

phone, or even better, send a voice

13:16

note. So, if you're anything like me, I

13:17

have some good news for you. ChatGPT

13:19

actually has a voice mode, and it is a

13:21

game-changer. But let me tell you where

13:23

this has actually come in clutch for me.

13:25

I do this thing called Russian banya

13:26

every Sunday night. It's my Sunday night

13:28

ritual. I book out the whole place and

13:30

what I do is I sit there, I sweat it

13:32

out, I clear my mind and normally that's

13:34

a couple hours where my phone is in my

13:36

locker and my brain is just switched

13:37

off. But a couple months back, I decided

13:39

to try something out. I had my AirPods

13:41

in, ChatGPT voice mode on and instead of

13:43

just sitting there in silence, you know,

13:44

once I was actually outside the sauna, I

13:46

started talking to it, okay? I had a

13:48

full-blown back and forth convo with it

13:51

about a specific business problem that I

13:52

was having. And I actually ended up

13:53

having a very similar conversation to

13:55

what I would have with my own COO. And

13:57

that's really when I realized that

13:58

ChatGPT isn't just a typing tool. I can

14:01

actually use it in times where I would

14:03

just be ruminating about stuff in my

14:04

head, on a drive, walking, waiting for a

14:07

coffee. If you set this thing up

14:08

properly, it gets even better. What

14:10

you're going to do is you're going to

14:11

pick one device as your main hub and

14:13

that is where all of your core work

14:16

lives. Then, you're going to install the

14:17

ChatGPT app on your other devices. So,

14:19

maybe if you have another phone or a

14:21

laptop or iPad, then you're going to go

14:23

into settings and set up connections.

14:25

Inside of here, you're going to connect

14:26

your apps and devices so everything can

14:28

talk to each other. Now, wherever you

14:30

are, you can actually turn on voice

14:32

mode, you can brief it on where your

14:34

projects are at and you can actually ask

14:36

it for next steps and talk through the

14:38

work in real time. And what's super cool

14:40

is that if you ever need to draft

14:41

something, let's say you need to

14:42

organize a file or maybe update a doc

14:44

and then email it to someone, well then

14:46

you can just say it out loud and it will

14:47

do it for you on your main device. And

14:49

I'll come back to that in just a second.

14:50

But here's an insane part about this.

14:53

And you know what, in all fairness,

14:54

maybe I should have put this in the

14:55

advanced section, but whatever, I'll

14:56

mention it now cuz we're discussing it.

14:58

There's actually an advanced version

14:59

[clears throat] of this that lets you

15:00

share your screen or camera in real

15:02

time. Meaning, you can literally say

15:04

that you are stuck on something, like

15:06

let's say a spreadsheet or whatever, and

15:08

share it directly and just say, "Here's

15:09

where I'm stuck, what should I do?" And

15:11

it can see what you are seeing and talk

15:13

you through the fix step by step instead

15:15

of you having to spend time describing

15:17

the problem or even worse, having to

15:19

type it in. Oh, and by the way, there's

15:20

actually one more thing worth setting up

15:22

here. If you've got a compatible Apple

15:24

device, you can actually turn on your

15:25

Siri, which is normally pretty useless,

15:27

let's be honest, and you can turn it

15:28

into ChatGPT via Apple intelligence,

15:30

meaning you get actual answers instead

15:33

of Siri just giving you the top results

15:35

of a web search. That means you can ask

15:36

it more complex stuff, not just timers

15:39

and weather. And that means that you get

15:41

one consistent assistant [music]

15:43

across your phone instead of trying to

15:44

bounce between Siri for basic things and

15:47

then ChatGPT for everything else. This

15:49

is so cool because you can literally

15:51

work from anywhere after you set this

15:53

up. Now, you know that whole connection

15:55

setting that we just set up for the

15:56

voice mode. Well, that actually leads me

15:58

onto the next tip. I want you to think

16:00

about when you first got your

16:01

smartphone. Whatever you needed to do,

16:03

well, there was an app for it. To check

16:05

the weather, you would just download

16:06

your favorite weather app on your phone,

16:08

or to book a flight, you would just

16:09

download another app for this, for

16:11

music, another one. And what happened is

16:13

that with time, your whole phone slowly

16:15

started turning into this home screen of

16:18

tiny little tools, each one of them

16:20

doing their own separate job. And I'm

16:21

sure that if you don't live under a

16:22

rock, that's still how you do things

16:24

now. What if I told you ChatGPT has its

16:27

own version of this? It actually has its

16:29

own inbuilt app store. And what if I

16:30

told you that most people never even

16:32

knew it existed or how to use it. So,

16:34

instead of downloading separate apps for

16:36

everything, you can just connect them

16:37

straight into your ChatGPT. And the even

16:39

better news is this. Once they're

16:41

connected, ChatGPT can do the work

16:43

inside of those apps for you, as I

16:45

mentioned in the previous tip that I

16:47

gave you. Let's say you need an email

16:48

sent. Let's say you need to check your

16:50

calendar. Let's say you need to book a

16:51

flight. Well, congratulations because

16:53

ChatGPT can do all of that for you. That

16:55

is your first step in having your own

16:57

personal assistant. Look, I'm sure that

16:59

every single person who's watching this

17:00

channel, especially if you've been

17:01

watching for a few years, will get to

17:03

the point where you are incredibly

17:05

successful and you hire a personal

17:07

assistant. And by the way, I can tell

17:08

you that once you get there, it is

17:09

honestly one of the best ways that you

17:11

can spend money in your life. But, until

17:13

you do, you can start here, okay? You

17:16

don't have to go spend 50 or 100 grand a

17:18

year on an EA or a PA. All you need to

17:20

do for now is just connect the right

17:23

apps. On the left sidebar, click on

17:25

plugins and connect any apps that you

17:27

want to. The first two that I would set

17:29

up are Gmail and Google Calendar. That's

17:31

a pretty good starting point. Now look,

17:33

I'll be honest, I'm not really a morning

17:34

person. I never have been, I never will.

17:36

So the first 30 minutes of every single

17:38

day, my brain is not fully awake yet.

17:40

And the last thing I want to do is open

17:42

up my inbox and get hit with 50 emails

17:45

before I've even had a coffee. So,

17:47

here's what I do. I ask ChatGPT to

17:49

summarize everything that came in

17:50

overnight, and then I ask it to filter

17:53

out what needs a reply today, what can

17:55

wait, and so on. And by the time I am

17:57

fully awake, I've already removed all of

18:00

the unnecessary clutter. Same thing with

18:01

my calendar, by the way. I ask it where

18:04

the gaps are in the week, so that way I

18:05

don't need to think about where I can

18:06

squeeze in a tennis session or some

18:08

boxing. And by the way, it gets even

18:10

better. There are some extra apps as

18:12

well. If you like music while you work,

18:13

you can actually connect your Spotify. I

18:15

like listening to Brain.fm, so I get it

18:17

to play some of that. There's even

18:18

Booking.com there. Yes, it can actually

18:21

help you look at flights and hotels,

18:22

too. The point is, ChatGPT doesn't need

18:24

to be just you asking it questions. It

18:27

can actually get stuff done for you. And

18:29

hold onto this, because later in this

18:30

video, I'm actually going to show you

18:31

how this will connect in advanced it. So

18:33

I'm going to share that with you in just

18:34

a little bit. For now, what you need to

18:36

understand is that once these apps are

18:38

connected here, ChatGPT can actually go

18:40

and use them for you. You see, now that

18:42

you've got your apps connected, here's

18:44

another cool use of it. Schedule tasks.

18:47

Look, if you feel like you have a lot on

18:48

your plate and you want to be more

18:49

efficient, this is a great one. This is

18:52

where you give ChatGPT the power to do

18:54

stuff for you when you want it to. It's

18:56

funny, because I used it for this very

18:58

video. You see, every week, or at least

19:00

I try my best every week, I put out a

19:02

new YouTube video, exactly like the one

19:04

you're seeing here right now. And every

19:05

week, the video needs to be posted on

19:07

time. Now, do I want to rely on my

19:09

memory for that and remember to do this

19:10

myself every week? Absolutely not. Do I

19:12

want to chase someone else down every

19:14

single week and remind them? Even worse.

19:16

So, I literally have a scheduled task

19:18

for this. What it does is it

19:19

automatically sends a message to my head

19:21

designer every week reminding him that

19:23

the video needs to go up. I don't write

19:24

it every time. I don't think about it

19:26

every week. I set it once and now it

19:28

turns into a background task for me. And

19:30

that's why I love this feature. It's

19:32

leverage. [music] You set it once and it

19:34

keeps working for you without you having

19:36

to carry around all these thoughts in

19:37

your head. Now, you might be asking

19:39

yourself, "Well, he man, how do I know

19:40

which task should be scheduled and which

19:43

one shouldn't?" Well, the way you should

19:44

think about it is this. Anything that is

19:46

monotonous should be automated. If

19:48

something happens on a schedule, it

19:50

should be scheduled. But, you also don't

19:52

want to schedule your whole life away,

19:53

okay? You need to be intentional about

19:55

it. For example, I wouldn't schedule it

19:57

to send me some random motivational

19:59

quote every morning. Sure, I guess

20:00

that's cute, but it's not actually doing

20:02

anything for me. Also, I wouldn't get it

20:04

to book a table for me for a one-time

20:06

dinner on a random Thursday. So, to

20:08

actually set it up, what you're going to

20:10

do is you're going to go to the

20:10

scheduled on the left sidebar and it's

20:13

probably going to give you some options

20:14

to get started. You're going to pick the

20:16

date and the time and you're going to

20:17

tell it exactly what you want it to do.

20:19

Now, it could be a reminder to someone

20:21

on your team. It could be a weekly

20:22

check-in on a project or anything else

20:24

that you find yourself doing over and

20:26

over again. And then, once it's set, you

20:29

don't have to think about it again. Now,

20:30

before we get to tip number 10, I'm

20:32

going to be very honest with you. If

20:34

you're someone who's got all the time in

20:35

the world to waste, well, then this tip

20:37

isn't for you. You can skip to the next

20:38

one. But, if you're someone who never

20:40

has enough time in the day or you're

20:42

like me and you need to learn something,

20:43

[music]

20:44

but you're too impatient {slash} maybe

20:46

sometimes lazy to dig through 50

20:48

different articles or documentaries

20:50

before getting an answer, well, then

20:51

congratulations, this one is for you.

20:53

It's called deep research mode. And the

20:54

name kind of says it all. You're going

20:56

to give it a proper research question,

20:57

not some throwaway one-liner. A question

21:00

that you need a real detailed answer to.

21:02

Then, it goes to work. It searches, it

21:05

reads, it cross-references different

21:07

sources, and a few minutes later, it

21:09

comes back with a structured report.

21:10

Now, I actually use this a lot, by the

21:12

way, and it's actually part of the

21:13

reason that this very video exists. You

21:15

see, every few weeks, I like to

21:17

understand what is actually working on

21:19

this channel. What is it that you guys

21:21

are enjoying, and what are you guys

21:22

enjoying on other channels? What kind of

21:23

content that I'm putting out do you

21:25

like, and what do you find useful to

21:27

you? AI content seems to be resonating

21:29

right now, but I'm also looking for what

21:32

other creators are covering that [music]

21:33

we haven't touched yet. As well as where

21:35

the gaps are. So, I gave Deep Research

21:38

that exact task. I told it to go look

21:40

into how AI content is performing across

21:42

YouTube right now, what themes keep

21:44

coming up, what audiences are asking

21:46

for, and what's not being covered well.

21:48

[music] And then, a few minutes later,

21:50

it came back with a full breakdown. And

21:52

that's actually part of what shaped this

21:54

video. I also gave it some data on my

21:56

channel, and I saw that the feedback on

21:58

the Claude video a few weeks [music] ago

21:59

was insane. Now, look, I'll be honest,

22:01

could I have done that manually?

22:03

Absolutely, but that would have taken

22:04

hours of [music] me scrolling through

22:05

channels, comments, trying to spot

22:07

stuff. Whereas, with this mode, it did

22:10

all of that for me in a few minutes. So,

22:11

is this something that you absolutely

22:13

100% need? Not really. But, is this

22:15

something that will 10x your [music]

22:16

efficiency, especially if a lot of your

22:18

decisions rely on you doing research?

22:21

100%. So, to use it, it's actually quite

22:23

simple. All you have to do is click the

22:24

plus button next to where you normally

22:26

[music] type your message. You're going

22:27

to select Deep Research and give it a

22:29

specific question. The more specific the

22:31

question, the better. Remember, this is

22:33

not for quick questions. Save this one

22:35

for the stuff where you would have to do

22:36

a lot of research. There we have it, the

22:38

six ChatGPT modes. Now,

22:40

>> [music]

22:40

>> there's one final mode I haven't spoken

22:42

about yet, and there's a reason for it.

22:44

You see, the intermediate modes that

22:45

we've covered so far are quite

22:46

straightforward. This last one is a

22:48

little bit more advanced, so please

22:50

don't worry about this if you're just

22:51

getting started. Focus on the beginner

22:53

and intermediate ones at first. They

22:55

should give you a massive advantage over

22:57

95% of people. Now, before we get into

23:00

it, one last thing. Everything that

23:02

we've covered today is going to save you

23:03

a load of time. Excuse my French. But,

23:06

if you're the type of person who wants

23:07

to use that time to build something on

23:09

the side, build a side income, build

23:11

something profitable, something that

23:13

doesn't require you to quit your job or

23:14

go all in and sacrifice everything.

23:17

Well, then I want to invite you to my

23:18

online event, Side Hustle Summit. It's

23:20

100% free. I'm hosting it from September

23:22

6th to September 13th, and I'm going to

23:25

show you the exact steps [music]

23:26

to build a profitable side hustle. I'm

23:28

going to leave a link to that in the

23:30

description. All right, back to the

23:31

tips. If you're a bit more familiar with

23:33

ChatGPT, I actually want to take things

23:35

up to the next level. Here's how. You

23:36

see this last mode is the closest thing

23:39

to having ChatGPT as your personal

23:41

assistant. This is where you give

23:43

ChatGPT full autonomy of your screen, so

23:45

that way it can actually go ahead and do

23:47

things for you. I want you to think

23:49

about that one annoying task that is a

23:51

pain in the ass every single day. It's

23:52

not the hardest task, but it's

23:54

repetitive. It's filling out the same

23:56

repetitive form for the 10th time this

23:58

month. It's comparing flight prices

24:00

between different sites to check which

24:01

one is ripping you off the least. Those

24:03

are all things that agent mode can do

24:05

for you. All you have to do is give it

24:07

some instructions, like find me the

24:09

three best reviewed hotels in Lisbon

24:11

under $200 a night with a pool. And

24:14

there we go. It goes off, it opens a

24:16

browser, and navigates through different

24:17

websites, and it comes back with the

24:19

best options for you. Now, do you

24:21

remember that tip where I showed you how

24:23

to connect different apps within your

24:24

ChatGPT, and how it's kind of like

24:26

you're having your own app store within

24:28

chat? Well, once you have those apps

24:29

connected, things like your Gmail and

24:31

your calendar, you can actually use

24:33

agent mode to then do stuff for you

24:35

within those [music] places. As in just

24:36

you understand, it can literally do

24:39

things inside of the apps that you've

24:40

connected. And sure, the average person

24:42

might know that they can actually ask

24:44

ChatGPT to go find some good hotels in

24:46

Greece, but I want you to take it a step

24:48

further. With this mode, you're going to

24:50

go from, "Hey, go find me some good

24:51

hotels." to, "Go find me some hotels,

24:54

then check my calendar to check when I'm

24:55

actually free, and then book it in for

24:58

me, and draft me an email to confirm

25:00

this. All of that in one go because

25:02

everything's connected. Do you see how

25:04

when you do this over a long period of

25:06

time, you are going to end up running

25:08

laps around everyone else simply because

25:11

you're going to be 10 times more

25:12

efficient than them. Now, here's the

25:13

important thing though. With this mode,

25:15

I want you to still be the one in

25:17

control. ChatGPT is not going to go and

25:19

book something or buy something or

25:21

submit a form without checking in with

25:23

you first. You will always have the

25:25

final approval. Anytime it's going to do

25:26

something that commits you, for example,

25:28

like spending money or sending something

25:30

or confirming a booking, it will ask you

25:32

first. You're going to then sign off on

25:34

it. All it is doing is just the boring

25:36

legwork on your behalf. I hope you

25:38

understand how this also gives you more

25:40

time to make a better decisions because

25:42

you don't need to do the repetitive

25:43

work. So, to activate agent mode, here's

25:46

what you do. And then all you have to do

25:48

is literally just tell it what you want

25:49

it to do, and it will take it from

25:51

there. A real quick, just one practical

25:53

tip though. Before you run any task, I

25:56

want you to make sure that you close out

25:57

any tabs with sensitive info. I'm

25:59

talking about banking, personal

26:00

accounts, anything that you just

26:02

wouldn't want it near, just in case. I

26:04

would also recommend that you actually

26:05

start with something that's low stakes.

26:07

So, that way you can kind of get a feel

26:08

for how it works. There you have it. My

26:10

top tips for learning 98% of ChatGPT in

26:13

under 30 minutes. Just by knowing what I

26:15

gave you today, you are already ahead of

26:18

95% of people. Look, ChatGPT is a great

26:20

all-round tool. It can write, it can

26:22

plan, it can organize, it can even take

26:24

actions for you. It is useful, no doubt.

26:27

It's the same thing I saw with Claude.

26:28

Every tool has their strengths and

26:30

limits. For example, Claude wasn't

26:32

really good at image generation, and

26:33

ChatGPT hallucinates. It's just what

26:35

happens when something is built to do

26:37

100 things well instead of one thing

26:39

exceptionally well. It's kind of like

26:41

hiring. A generalist employee [music] is

26:43

useful. Okay, you can hand them many

26:45

different tasks, and they will get most

26:47

of them done to a pretty decent level,

26:49

but they're never going to be

26:50

spectacular. For that, you would need to

26:53

go find a specialist. Or think of a

26:54

doctor, if you needed to get brain

26:56

surgery done, would you want a general

26:58

practitioner to handle it? Or would you

27:01

want a brain surgeon? Well, the exact

27:02

same thing applies to AI tools. The

27:05

point is, it's not about which tool is

27:07

the best. It's about which one is the

27:08

best for what you are trying to do. And

27:10

the more specific the task or outcome

27:13

gets, the more nuance there is, which is

27:15

typically [music]

27:16

what tools like ChatGPT struggle with.

27:18

Think about specific tasks like make

27:20

$10,000 a month online, or how to solve

27:22

this one specific problem for a specific

27:24

person or business. Try using ChatGPT

27:27

for that, and let me know how it goes.

27:28

Let's say when it comes to something

27:29

ultra specific like making $10,000 a

27:32

month with a marketing agency, for

27:33

example. There are direct ways to make

27:35

money, and there are indirect ways. A

27:37

direct way would be something that

27:38

identifies specific businesses that need

27:41

help with their marketing, and helps you

27:43

build the solution for them. [music] An

27:44

indirect way would be using tools like

27:46

ChatGPT to save you time, so that way

27:48

you can spend more time on figuring out

27:50

how to make money elsewhere. Do you see

27:52

[music] the difference? Don't get me

27:53

wrong, ChatGPT is an incredibly valuable

27:55

tool. But as I said, it's not going to

27:57

make you a millionaire. For specific

27:58

outcomes, you need specialized tools

28:00

that are built for that specific thing.

28:02

But look, to be honest, that is a topic

28:03

for another day. If you want to be the

28:05

winner of the MacBook, once again, leave

28:06

a like and drop a comment of your

28:08

favorite tip. And with all that being

28:09

said, as always, I'm watching from afar,

28:11

>> [music]

28:12

>> and I'm rooting for you.

Interactive Summary

This video provides a comprehensive guide on maximizing the potential of ChatGPT by moving beyond basic interactions. It introduces key features such as custom instructions, memory, and various specialized modes—including canvas, study, research, and agent modes—that allow users to automate repetitive tasks, improve learning, and achieve significantly higher productivity.

Suggested questions

3 ready-made prompts