Quot is an extremely powerful AI model.

In today's video, I'm going to show you

how you can leverage a high reasoning

model like Obus 4.5 to build out a real

application. By the end of this video,

you are not only going to learn how to

start leveraging these AI models to

create real code, but you are going to

learn something fundamentally

groundbreaking, which is you genuinely

don't need to know how to code anymore

to code real software when using these

models because it's gotten to the point

that they're that good. Does that sound

good? Let's jump in. Welcome back,

y'all. Today's video is sponsored by

Anthropic Obis 4.5. Y'all, if you know,

you know. This thing is killer.

Therefore, I discussed with the team and

I think the best way for me to show you

how to leverage this model is we're

going to use cloud code. I'm going to

show you how to integrate it into a

integrated development environment like

VS Code. Don't worry if I'm speaking

crazy to you. This is extremely simple

process. And then we are going to create

a real application together. I'm going

to show you a real software's journey

when it comes to building out the

architecture, the plan associated, and

then the end execution point. There's

going to be an application where you'll

provide a YouTube link. The YouTube link

will then be reflected in the

application, eg the thumbnail for the

video, the title of the video, the

duration of the video, and then finally,

the end value point that we'll provide

the consumer will be the ability to take

that underlying title and provide five

variations based off a specialized

prompt we create with Opus. Let's do it.

So, first things first, I'm going to

leave these cloud code docs in the

description down below. These are

extremely helpful. Nice little tip for

y'all. If you want to understand the

docs more or maybe learn at a chatbased

level, simply hit copy page here, come

over to your Claude chatbot, and then

just paste it here and then say like,

"How do I do this? How do I do that?"

Quite literally paste. And what you'll

notice is that we have an amazing UI

that shows you the page. This is going

to give context to Claude here to help

you out in the direction of this

tutorial. But first things first, we

need to get in an environment where we

can actually build this. Think of it

like how do you play Minecraft? You got

to install Minecraft. So scrolling down

here, we need to do a couple things. So

first thing you need to do is simply

install VS Code. Once you have VS Code

installed, you're going to open up and

see a window like this. Step number one,

you see this button that says open? Hit

open. And then it's going to prompt you

to either choose a folder or create a

new folder. Create a new folder. Once

you create your new folder, you can call

it like YT app. Call it whatever you

want. This is going to be where we do

all of our work in. Think of it like

starting a new project and hit open. We

are in YT app. I like the name. But now

we need Cloud Code. So to do that, we're

going to go to our extensions here.

Click this. You are then going to type

Cloud Code. Obvious, right? You will

notice there is a ton, but we need to

choose from Anthropic because they have

the best integration. So click this and

simply hit install. As you see right

above me, trust publisher and install.

Once it is installed, you'll notice it's

prompted here. come up here to the

little claude icon. Click it. We have a

couple different options here. Now, for

me personally, I have a Claude AI

subscription, which is nice. So, I'm

going to go ahead and use this method.

It's going to bring you to this

absolutely beautiful page. I love the

art. Simply hit authorize, y'all. And

thank you, Anthropic. We are going to

build something great. Once you do that,

you come back here and you will notice

we have our Claude code chat. We have

our nice little clawed squid. I love the

squid. And we can begin. First things

first, anytime you want to start a chat,

come here to the top right and hit the

claude icon and we have clawed code

here. So, if you made it up to this

point, we have set up everything we need

to do to start coding with AI. This is

where the fun begins because now it's

like, you know, we purchased a Ferrari

on the lot and now we're like, you know

what, I want to actually drive the

Ferrari. So, let's drive the Ferrari

here. And we got to create our first

base layer. When I say base layer,

essentially a thing in software

development we call a tech stack, the

actual application. If you don't know

what I mean by that, think about like

Chrome or Safari. That's an application.

That's code. So, a really very popular

one that a lot of developers like to use

is TypeScript. And before I put in this

prompt, let's actually switch this from

ask for edits to edit automatically cuz

essentially we're just creating the

architecture first. So, here is our

first prompt. Create a TypeScript app

and have it be a simple landing page

with an input link for a YouTube video.

We don't need to get technical right

now. Right now, we need an actual

machine to play around with a actual

landing page. So, first step here is

we're going to get a really nice to-do

list from cloud. Thank you. And the

second step here is going to be

rendering this page. If you're new to

coding, I suggest you just let AI take

the wheel. So, we say yes for this. And

if you want to skip these little prompts

here, we can always do yes and don't ask

again. But as you'll see here, we're

getting a nice little to-do list of

first off initializing the project for

the TypeScript app, creating an HTML

landing page of YouTube input, creating

TypeScript file for handling video

input, adding CSS styling for the

landing page. When creating

applications, if you're new, I want you

to think of it very simply. So, there's

a thing called a front end and there's a

thing called a backend. The front end is

what you're watching this on right now.

Whether this is on X or on YouTube, you

are watching this on a nice little thing

called UI, user interface. When you

click that like button right now, you're

interacting with the front end. But when

you click that like button, what happens

is the back end hears that and the back

end's like, whoa, there there's a

there's a like on the video. And that is

how it communicates. So, we got the

front end and the back end. So, we're

going to go ahead and let this initiate.

When you get prompted like this, just

say yes and don't ask again. We want to

give the AI the full wheel here. People

use different app text stacks depending

on the specific use case. What I

encourage you to do if you're brand new

is leverage the text stack of

TypeScript. It works in the front end

and the back end. Other text stacks

include things like using Python in the

back end, but usually that is for more

machine learning oriented types of

workflow. What you need to understand

fundamentally is that these AI models,

no matter the application you're trying

to build with it, you can vibe with it,

it can build it. And these things are

extremely impressive. So right now it

finished the first task. What you'll

notice is that and how you'll know it

finished the first task is that it's not

chatting anymore and it gives us what to

do actually next. So in order to see

this rendered you'll notice that it says

build the type script. So we're going to

do that. I'm going to come here to the

top right. I'm going to click this. This

is going to create our build. I know it

looks I know terminal can be scary.

Don't worry. This is very simple.

There's a couple ways you can approach

this. You can be extremely viby and say

build it for me or you can simply just

grab the command here and put it in. So

I do npm run build. Hit enter. What I

want you to understand fundamentally is

look at this terminal. It says YT app.

Do you remember where YT app was? Yes,

that was our application up here or just

a simple folder on your computer. When I

do npm run build, all that's happening

is that it's taking the code like the

code static. It exists but it's building

it. So we can see it visually. Now if

you don't want to use terminal, you can

also simply ask the cloud agent to do

it. In local development, in order to

see a website on our computer, we call

it localhost 3000. This is a port, but

you don't have to make it confusing.

Essentially, this allows it so that we

can see this code on a web browser link

that is only accessible by you, which is

nice. And we're able to see this once

it's deployed in a development server.

We can go ahead and check it out. So,

I'm going to hit command and click this

link. And boom. If I go up here, you'll

notice in the top left local host 3000.

That's pretty cool, right? You literally

just did two prompts or actually just

one prompt and we were able to get a

live rendering of what this looks like.

It doesn't just stop there, though. So,

next we're going to build out a little

bit more functionality here. But one

thing I'm noticing is that it does keep

prompting me and asking me, do I want to

do XYZ? Like, should I do this command

yes or always allow? In order to fix

this, if you want this to just go like

take the wheel, go down to settings, hit

settings, come over here, hit Claude,

and then scroll down. You're going to

say allow dangerously skip permissions.

Hit allow. It sounds scary, I know, but

if you really want to push this AI model

to the limits, I suggest you turn that

on. Now, that done, we're going to look

at another really cool feature here,

which is going to be our ability to do

plan mode. This is I love this. This is

especially important if you are new to

coding or you're about to do a really

big next step. So, our current

application allows us to put in a video

link, but it's not connected to YouTube

API. It's like how is it going to get

the data? Like, it doesn't know how to

do that. And in addition to that, we

need the extra logic of being able to

actually analyze the YouTube video title

and then get variations. This sounds

confusing, but watch me prompt it. So,

we're going to do this prompt. Let's

make a plan. I want to be able to

provide a YouTube link as we saw for a

landing page. You use YouTube API to get

relevant metadata like the thumbnail,

image, duration, and the title of the

video. I want to have the Cloud AI give

me five different variations of this

YouTube title and reflect in the website

page. One thing you need to understand

about software is that you might be

asking yourself, Corbin, what the heck

is an API? I want you to think of an API

like our ability to grab data from

anywhere. So for example, another

example of API would be Instagram has an

API or meta API where your ability to

grab metadata on the Instagram post for

the title. Corbin, what is metadata? The

YouTube title, the duration of the

video. API is going to be your key to

access data and spread across different

ecosystems when building out softwares.

I love API. What's a great example? This

cloud API. This allows us to integrate

an AI model into a software and provide

AI model outputs using their API

endpoint. API is important. Learn more

about API. API allows us to access data

from anywhere. I love the data. Okay,

we're going to hit enter here.

Genuinely, Opus 4.5 is killing the game.

It seems like Enthropic has really

narrowed down on coding models and the

ability to write actual good code. So,

when planning with Opus here, it just

knows what to do. This is really next

level stuff, y'all. It wasn't always

like this. And I would say within the

last two to three months, the leap that

the tech has gotten to is tremendously

higher than it used to be. So, I will

say this, if you ever coded with AI in

the past and you had issues, it's

changed tremendously. And you're

watching this 6 months to a year from

now, oh, you're just having fun cuz

these models are even better. So, during

plan mode, what's nice is look at all

this. Look how amazing this is. This

gives me the ability to really read and

take actionable steps working with the

model. So, which YouTube API should we

fetch the video metadata with? YouTube

data API v3. Let's just do Google's

official one. How should we integrate

cloud AI to generate five title

variations? What you'll notice is that

it gives different options, but Opus

4.5's reasoning is so high level that

it'll be able to be like this is the

right one. So, usually you want to go

recommended if you don't necessarily

know what these other mean. If you don't

necessarily know what these other ones

mean, cloud API, duh. Of course, you

want to use cloud API here. Where should

the title, variations, metadata be

displayed below the video player makes

sense? Or you can push back side by side

with the video replace input section or

even other. So you want to give your own

input. So maybe these options aren't

relevant to what you were envisioning in

your brain. But for me, all these seem

solid. Submit answer. So two things are

going to happen here. First thing we're

going to do is we're going to set up a

ENV. Now don't make this confusing. An

EMV is what we call an environment. Uh,

not the environment like you're going on

a nice walk. An environment where you

can store highrisk variables. Corbin,

what's an example of a high-risisk

variable? This sounds scary. These API

keys, right? So, the Claude AI API key

we're going to create here. That's high

risk. You can share it with developers

that you know or maybe on your team, but

you don't share this with anyone. You

don't you're not going to put a tweet

out and be like, "Yo, this is my Claude

API key." Don't do this. If you do, DM

me. No, I'm joking. But these are high

risk. So we put these kind of variables

in a thing called an env. And we

reference them in the code this way.

You'll see what I'm talking about pretty

soon here. It simply is just another

file here. So when a plan is ready, you

can go ahead and review it. So I'm going

to scroll up here, see all the relevant

files that we created, and then it will

essentially ask me, do we accept this

plan? Yes. And auto accept. Let's do it.

What you'll notice is, look at this,

y'all. This is all just prompts, y'all.

I haven't touched a single file yet. And

you're about to see a real application

be made. And look at this.

Like it just goes down the list. I like

the list. Let's let it work here and

build it out. In the meantime, while

it's building it out, we are going to

create our keys. So, the first key we're

going to need to create in order for the

AI model to access this API is going to

be the anthropic key. So, I'm going to

go to anthropic here and hit create key.

Simply choose your workspace. You can

name your key. So, I'm going to say my

YT app here and then hit add. It will

create this key. This is high-risk.

Don't share it. Have it ready, though.

Next, we're going to need the YouTube

API key. In order to access that, we

need to make a GCP account or just

Google Cloud, right? Come over here to

API and services. Go to credentials and

we're going to say create credential.

We're then going to say API key. This

will then show you your API key. So make

sure to copy it. We're going to rename

this to YouTube right here. And then we

can go down to restrict key. We're going

to select APIs here. And all we need to

care about is going to be the YouTube

API, which actually isn't showing up. So

let's enable it and make sure it is

there. So we're going to enable API and

services. What we want to do here is hit

enable API and services. Again, we're

going to go to YouTube. YouTube. We're

going to go to YouTube data API v3. Why

is it v3, Corbin? Because it's better

than v2 enable. Once it is enabled,

you're going to come back to credentials

here. You're going to go to your API

key, and then you are simply going to

hit restrict key, select APIs, YouTube,

and there we go. Select it. Okay, save.

Nice. Cloud is still working, but we

have the relevant variables that we

need. So, the next thing we're going to

do here is we're going to say new file.

Actually, it's already been created for

us. I love it. ENV. Click this. And look

at Claude. Just an absolute beast.

Claude said, "Time for you to put them

in." I'm like, "All right, I got you. I

got my keys. I got my keys." So, you're

going to put our a YouTube API key here.

We're going to put our cloud API key

here. Don't make it difficult, y'all.

Simply delete this part. Paste your key.

Delete this part. Paste your key. And

the keys we just got. Now, if this EMV

file wasn't created, and you don't see

this right now, right click, new file,

env. It's that simple. And then hit

enter. And then on top of that, if

you're like Corbin, I created the ENV

file. What do I call the variables? So

one just talk to the AI model claude

here it'll help you but two these are

pretty solid variable names although to

be honest with you we might need to

rename these and add vite in the

beginning or vit as the French will say

and v is the compiler and builder for

this yes here and keep proceeding

regardless add your keys here so once it

finishes it's going to show you how to

run the application and second it'll

show you that we need to add the env key

like we did earlier I'm going to add a

clarifying question though real quick

and obviously you can ask clarifying

questions as you build your application

but I'm going to say added the API keys.

Do we need to add the vit at the

beginning of the EMV variables? Let's

see what it says here. The best part

about coding in this context is that not

only is this a 24/7 teacher, but you

learn as you go. So, it looks like it's

not required. So, that's good. We're

going to go ahead and rebuild our

application here and then we're going to

try it out. So, it went ahead and tested

for me. And this is really cool. What

the AI model can do within cloud code

here is the ability to run terminal

commands to test and verify endpoints

are working and everything is connected.

So, I'm going to try this. I may need to

restart the server, but supposedly it

could be good to go. So, coming back

over here, I'm going to refresh this

page. And the page is refreshed. What I

suggest you to do in Chrome is hit

inspect. Simply come up here to the top

right, hit console. This is going to

give you a little bit more context of

what's incurring here. As we see there's

already an issue here. So, what I would

like to do and when using AI models to

code is when I see something like this,

I'll simply copy it. I come over here

and be like, I see something red. It's

an error. What do I do? And what is

great is that these models are so smart

and intuitive that it'll be able to

understand what is incurring. So it

looks like it fixed it. It looks like

it's rebuilding the front end here. And

it looks like we are going to try

another YouTube link. So if we come back

over here, we can reload the page.

YouTube API ready. Give us a nice little

console log here. We're go ahead and

take one of my videos, an amazing video,

and I'm going to paste it here. And I'm

going to hit load video. Let's see what

happens here. Oh, okay. This is how the

next I like it. I really like it. Um, a

couple things and things like this just

impress me is the ability for it to make

really good UI, but also like do things

that you weren't expecting. I wasn't

expecting for it to really load in the

video and actually be able to play the

video. That's cool. But also notice like

it provided the metadata here. We do

have a little bit of an error here. So,

we'll figure that out right now. But so

far, I'm loving this. I love how it

shows the nice little thumbnail as well.

So, what I'll do is what you'll notice

is that okay, so the YouTube API worked

nice, but we're not getting the AI

generated title variations. So, what

I'll do is like if you don't know what's

happening, simply copy like all the logs

here and then simply paste them here and

then be like, I don't see any title

variations

question mark. As you get better with

coding, you'll be able to narrow down

and speak in more developer jargon. But

for now, that's what's really cool about

Claude is that you can just simply say

what you don't see. So, I'm going say

yes here and let it just check it out.

And you'll see that it ran into an error

when accessing the claw AI and it's

going to fix it for us. It's going to

understand how to effectively do the

correct type of payload to receive the

title variations. One thing in life is

that the more you do, the more you

learn. And in the beginning of using

this, we're going to say yes, build the

back end. I love the back end. Beginning

of this is going to be a little bit like

uh Corbin, I have no clue what the heck

that chat's saying. But don't worry,

over time slowly, you're going to

understand more and more what a payload

is, what the code is, why are we running

a separate command for front end, why

are we running a separate man command

for backend, it all becomes connected

over time. So, we're going to try again.

I have a YouTube link load video. So,

let me show you another tip when

debugging. You'll notice that I don't

see the titles. So, what I can do is I

can screenshot this. What's cool about

Claude Code here is that we can actually

upload screenshots as well. You hit this

little plus button or you can simply

drag it in like this. I love it. I still

don't see the title variations. Can you

add logs? These are the console logs

that you saw in the browser and a

loading state for when the titles are

being processed. You know, like when you

use a software or an app and it's like

that little loading thing. I want to see

that. So, here we go. A lot of times

when coding with AI, it's not that it

can't do it, it's that it doesn't have

enough information. So, console logs and

loggers in your back end are going to be

your best friend. This is going to give

the YouTube, sorry, the Claude AI the

ability to really fundamentally

understand what the error is and fix it.

What's really cool here is now we're

going to get a console log of when this

API request is given to Claude. So, here

we go. It has successfully rebuilt the

application. It will also give me

directions of how to actually do what I

just told you when it comes to the

console tab, which is cool. I'll come

back over here. I'm going to reload the

page. I'm going to put in the link and

I'm going to hit load video. And this is

important because what you'll notice is

that no title variations called API may

have failed. It's giving me an

indication. And then what you'll notice

is that it gives all this information

here. You don't need to know what's

occurring here. All you need to do is

simply copy all of this. And then some

of you'all might be like, Corbin, what's

this 404? It says fave icon. If you

don't know what a fave icon is, look at

your browser right now. You see that

little box like the YouTube box like

that says like the YouTube logo in the

tab? That's a fave icon. So I know it's

not that. You can still copy that if you

don't know it's that. So you can copy

that if you want to. Copy console logs.

Paste them here. Enter. And what you're

seeing right here is the workflow, the

backtoback as a true vibe coder. So, I'm

gonna say yes to this. And what you'll

notice is that with these clear console

logs, it's able to identify the issues

clearly. And there's my issue. My credit

balance is too low. So, I need to add

credits to my billing. And I've gone

ahead and added credits. But I want you

to understand, I'm showing you this

workflow. So, if you run into errors

that aren't associated with billing, you

know how to solve them, which is

fundamentally a very important skill in

app development, especially with these

new AI models. Also, what's really cool

is that it gave me a direct link to

solve it. I love it. I'm just going to

confirm that I added the credits so it

knows we're on the same page here. And

then we are going to try out again. So,

come back here. We'll go ahead and put

in our YouTube link again. Below video.

And here we go. So, I want you to notice

two things. First thing is that the logs

look like they worked. And coming down

here, we have all of our AI generated

title variations based off why AI

software is the next app store. From app

store to AI store. Okay, I like it. The

AI revolution. Building tomorrow's

digital economy. It's digital. the

coming of the AI gold rush. AI apps, AI

vibes. I love it. Now, if you want to

start a new chat, all you need to do is

simply come up here to the cloud icon,

click cloud, and get going. Now, one

thing I want you to understand is look

at the application I just built, and

that was all done in a single chat, and

I think in under 20 minutes. That's

fast. That's extremely fast. And if you

genuinely had no clue how to code, you

could build the same exact thing. What I

encourage you to do is step by step do

exactly what I just did right there. And

that's going to unlock in your brain

like wait a second this is different

because it is. So that just about does

today's video. Make sure you leave a

like. It is completely free. And as you

already know these style of videos, I'll

see you in the next. Did Corbin just

quite literally show me one of the most

powerful models on the market called

Opus 4.5 and show me how to build an

entire application in under 20 minutes

type of video. I like it.

I really like