How I Built a Production-Grade Agentic AI Customer Support System Using Claude Code

Hello all, my name is Krishna and

welcome to my YouTube channel. So guys,

uh specifically in this AI era, it is

necessary that you start using AI a lot

with respect to any kind of task you do

in your company from project

development, from learning new things,

from asking questions to AI like how you

can improve anything that you are

currently doing and that is specifically

required. Many people when I used to

teach from last two 3 years right many

people used to say AI is just a hype

right and for all those people who still

think AI is a hype god bless you okay so

in this specific video we are going to

show you how you can go ahead and

develop some amazing projects right and

we'll consider a specific use case and

we will be using cloud code and we will

do some amount of wipe coding in order

to complete this specific project Okay.

Uh this project has been recorded by uh

Sunonni Savvita who is an amazing mentor

who's working in a uh company uh called

as PWC and uh right now he will also

talk about like what are the industry

trends that is currently being followed

and how you can also go ahead and use

cloud code for doing most of the task

that we specifically do in development.

Now before I go ahead and start this

specific video, I really want to make a

quick announcement [clears throat] about

our new boot camp that is called as

modern route full stack generative and

agenti boot camp. And this is basically

starting from March 15th, 2026. And the

classes will be on every Saturday and

Sunday from 8:00 p.m. to 11:00 p.m. IST.

Now this particular course is suitable

for anyone who really wants to get into

generative AI, agentic AI, who wants to

quickly get started, how to probably go

ahead and learn things, how to make sure

that you're industry ready to work in

any kind of generative AI and agentic AI

projects. Okay. So here you can see it's

a very detailed syllabus. Here we'll be

talking about LLM foundation, LLM

ecosystem model selection, LLM's

finetuning, production rag, advanced

rag, agentic AI, uh, langraph

orchestration. Along with that, we'll

also be using evaluations and guardrail

techniques along with cloud deployments.

All this information will be given in

this particular description of this

particular video. Uh, please make sure

that you watch this entire video. This

is an amazing project to get started

with. So let's go ahead and enjoy this

video. What if guys if you could build a

entire system or entire project just by

giving some instruction to AI? Uh so

yeah this thing is possible and inside

this video I'm going to show you how you

can build a realtime project using cloud

code in a faster and in a smarter way.

So hello everyone, my name is Sunonni

Savvita and I am back with another

exciting and important video. So inside

this video guys, I will show you how you

can use the cloud code for building any

application. Uh I will give you the step

step-by-step guide how you can set up uh

this cloud code inside your local

system. how you can utilize that and

even I will show you how you can

configure it inside your VS code. You

can configure it in any uh ID whether

it's a cursor, anti-gravity or pycharm

but uh I will show you with the VS code.

Okay. So uh guys uh this is a this is

the website of the cloud cloud.com where

you can find out a different different

services uh related to the cloud. If you

don't know guys, so this cloud is a uh

product of the anthropic. So once you

will search this anthropic.com, you will

lend to this homepage of the anthropic.

Uh this is the name of the company and

the cloud is a product of this company

only. So just uh read out this entire

page and understand uh about their

vision and all everything. Now how you

will uh come to the cloud A. So here you

can see they are giving the option to

try out the cloud a and once you will

click on click on this drop-down once

you will hover your mouse over here you

will get a different different product

uh related to the anthropic okay and

apart from that you can explore the

various model so let's uh check out the

cloud a guys so once you will click on

this cloud a so uh it will redirect you

to this particular page where you can

click on this try cloud a so once you

will click on this try cloud A you will

get the similar page page uh like you

you were uh seeing over the chat GPT uh

now uh they are giving the similar kind

of interface where you can do the

chatting and all everything uh and it is

a journal purpose chatbot. So they are

offering a different different model. So

if you will drop down over here if you

will click over here.

So you will get a different different

models over here like opus, sonet, haiku

right uh uh and uh guys these model this

opus model is a advanced model this

sonet model is a mediocre model and this

haiku model is a uh basically it's a

cheapest model right so they are also

offering a different different category

of the model similar to the openi only

and apart from uh that guys they are

offering a various surveys so you can

navigate to each and every service from

this uh homepage page itself. Now let me

show you couple of more product. So

cloud a as I shown you this cloud a

right now apart from this one they are

offering two main product. One is a

cloud a code which we are going to learn

in today's uh video and the second is

the cloud a cowwork. So what is a cloud

code guys? So the cloud code is a

product with that actually we can

develop a application. uh this code uh

basically is mainly created for the

developers who can uh directly configure

this cloud code inside their system and

they can do their development work.

Okay. And today I will show you how you

can configure it and how you can develop

a project using this cloud code. The

second uh product and the very important

and the like the recent product is a

cloud co-work. Now this cloud co-work

basically it's a AI collaborator. So

whether you are a developer or you are a

non-developer anyone can collaborate

with this uh AI with this cloud co-work

uh you can configure this cloud co-work

in your system you can give a access of

your file and folder and uh it will work

in a similar way like any human is

working and in a more efficient way

basically so this is a way where you can

collaborate with the AI and you can work

along with the AI so uh cloud co-work I

will show you in some other video today

we'll discuss about the cloud a code.

Now they are offering you different

different other feature. You can

configure this cloud a in a different

browser in a different services and they

are offering you the different model as

I told you the advanced model is a cloud

opus the mediocre model is a cloud sonet

and the cheapest model is a kaiku okay

so you can explore the different

different model and you can understand

their capability now uh platform wise

guys so they are giving you the

different different platform where you

can check out the pricing you can check

out the documentation and all everything

so if I will click on this developer

docs guys so you to get a documentation.

Now this is the complete documentation

of the cloud API. Uh it is similar to

the openi API uh like we are using uh

open API right uh and the different uh

services around to that. So in the same

way they have given you the cloud API

and you can configure it in a various

way. Okay. And you can use the different

different services uh just by reading

this particular documentation. Now apart

from this one guys you can check out the

pricing and all everything. So they are

mentioning the pricing of their model

and all. Now solution wise you can check

out the different different use cases.

They are giving you the several use

cases which is belonging to the

different different industry means they

are saying how they are helping to a

different industry and uh just uh by

navigating to this use case you can

understand about their work. Now pricing

wise again they have given you one like

one more page you can check out over

here. Resources wise they are providing

you the various uh platform from there

you can learn about the cloud a you can

understand about the cloud a you can

explore about the cloud they are

providing their own blog website they

are providing some uh uh like recorded

courses okay some tools connect uh

plugins and all everything they are

providing you over here even the

community is also there you can directly

connect with their community. So this is

the homepage guys of the cloud a if you

want to really understand this cloud a

product then please [clears throat] uh

go and check out with this homepage now

guys one more thing we are talking about

the anthropic if we talking about the

cloud a right so they are mainly

focusing to the developer community so

they are coming up with a product where

they can help as much as to the

developer community and cloud code is

one of the product which a developer can

directly use. If you will look into this

cloud a chatbot also uh there also if

you are going to be asked a general

purpose question maybe it is not not

very good in terms of reasoning like

chat GPT chat GPT actually it's a

general purpose uh platform but guys it

is very good uh in terms of coding and

the development if you will ask uh this

to develop anything right uh to create

any application it will create in a very

professional way it will uh give you the

very uh like a good code right and uh

even basically you can do a certain

level modification. So their model are

the uh developer focus and it is like

very good in terms of development and

all. Uh maybe it is not good uh in terms

of the general knowledge but yeah they

are helping a lot to the developer

community. Uh now guys I hope you got

the clearcut understanding of this

entire platform. Now if I will uh show

you this cloud a code. So just click on

this cloud a code and uh after clicking

on this cloud code guys you will get the

complete detail about them. So who is

using the cloud code what is the purpose

of it how you can configure that right

and uh you will get everything over this

page itself. Now uh if you want to go

step by step one by one right if you

want to explore everything about this

cloud a code then just click out on this

documentation. Now once you will open

this documentation guys. Now over here

uh itself you will get the complete

detail about the cloud code. So this

specific documentation they have created

for the cloud code only. Now how to

configure the cloud code, how to run

this cloud code and how to use this

cloud code. Everything they are uh

telling you over here right. So just

navigate to this documentation and you

will understand everything about the

cloud code. uh even uh they are talking

about the uh different ID right where at

what all places you can configure this

cloud code. Uh even they are talking

about the GitHub action right how you

can utilize it in a CI/CD pipeline. Uh

apart from that they are helping us in

uh in a GitLab also. So various uh

integration they have provided to us.

Okay. So, GitLab is there, GitHub is

there, uh, and the integration with the

various ID is there, you can directly

configure it over any uh, uh, over any

like operating system, whether it's a

Windows or Mac operating system, Linux

operating system, anywhere you can

configure it. Now, I hope guys you

understood everything about the cloud A

and the cloud A code. Now, guys, uh, to

read out this documentation, it's like a

bit challenging in a starting if someone

is a fresher or developer, right? So uh

maybe uh they might confuse by after

reading this documentation and all. So

for the beginner guys what I have done I

have created a summary of this entire

documentation. So this is the entire

summary guys uh which I will show you

which I will uh uh like which which I

will use to configure this cloud a code

inside my system. Okay. So I written at

least five to 10 steps over here and

each and every step we are going to

follow one by one in the step by step.

So first let me show you in uh how many

ways actually you can configure this

cloud code inside your system. So now

now once guys you will go ahead with

this cloud code documentation and here

uh if you will uh click on this overview

right uh so after that guys you will get

you will get this page. Now just uh

check out with this page uh here only

you will get the complete uh detail

about the configuration. So you can

configure it via terminal, you can

configure it via VS code,

you can uh configure it via desktop

application, uh you can configure it as

a web web, right? And the Jet Brands

also JetBrans actually it's a company

which is providing us a PyCharm ID,

right? Uh it is a owner of the PyCharm

ID. Uh now guys uh inside this video I

will show you how you can configure it

via terminal uh via CLI and I will show

you how you can uh configure it even

inside the VS code. Okay. So I will show

you this both way uh even you can

install the desktop application. So they

are providing you for all the operating

system like Mac OS, Windows, Windows ARM

64 for various uh for various operating

system for the different operating

system they are providing their ready to

go installation. So directly you can

utilize this as well and you can

configure the cloud a inside your

system. Okay just for the development.

Now uh here guys I will show you the

easiest way where you can uh use this uh

uh simple command and using the simple

command you can configure the uh you can

configure the uh cloud inside your

system. Okay. Uh now apart from this one

guys if you are using any other system

like I am using Windows system so I'm

going to use the Windows command. If you

are using other system like Mac OS,

Linux, so for that also they have given

the specific command right. So whatever

system you are using, whatever operating

system you are using, uh doesn't matter

throughout the operating system,

throughout the different operating

system, they are supporting to us. Got

it? Now let's do one thing guys. Let's

try to read out this uh do let's try to

read out this step one by one and uh

let's try to set up this cloudy code

inside our uh inside our system.

So guys uh first and the foremost thing

you should have a g inside your windows

system. Uh if you are running this thing

inside your windows system then g is

mandatory. Now why it is mandatory?

Because in a back end it is using git

bash to execute a command. So uh if you

will ask to generate anything to cloud a

code it will generate but where it will

execute that it's not going to be used

the terminal directly instead of that in

a background process it's going to be

use your g bash so g is a important one

even in the documentation also they have

mentioned the same thing if you are

working with the windows system then

install the git the second is the cloud

code install the cloud code after that

uh you just have to run this command run

cloud A then uh choose the login method

there is a various login method uh

mainly they are providing you three

login method so I will show you all the

three login method and then we are

selecting any one method uh then uh

setup billing so pro API whatever right

so pro with the pro plan you can go with

the API you can go so I will go with the

API uh right now okay and even I will

show you the pro plan if you are willing

to go with the Pro plan you can go with

that. Uh open a project folder ask cloud

to inspect the project work module by

module verify the verify and generate

the code. Okay. And then run it and test

it. So this is all the command this is

all the steps we are going to follow

throughout this video. So uh in step one

guys we'll install the git and we'll

verify whether we have git or not. So

how you can install the git? So for

installing a git guys you just need to

go through with this git scmm.com. Now

once you will go through with this git

scmm.com here they have given you the

various uh option to install the git.

You can install the git for uh window

for the Mac, Linux and for any platform.

So every sort of a way you will get over

here to install the git and yes uh you

just need to search this gitcm.com over

the browser you will get this page. uh

even if you're working with the Mac OS

and Linux but guys uh please keep out

keep this git inside your system

configure this git inside your system it

is very easy uh now what is the next one

so after that guys uh here you can see

so the next is going to be install cloud

code so we'll install the cloud code

here is a command for installing the

cloud code so we are going to be install

this cloud code using what using the cmd

or powershell I'm working in a windows I

can download it from here or else what I

can do I can directly install the

desktop version of the cloud a code and

with that also I can configure it inside

my system. Uh now after the installation

guys you will have to start the cloud

code then just write the cloud a okay or

after the installation itself they will

give you the couple of command to set up

the uh set up the subscription and all

you just need to follow those specific

command. Uh now after writing a cloud

guys uh you will your cloud will be

started. Now you have to choose the

login method. So after writing a cloud

you will get this login method option or

else while you are installing at that

time only after the installation you

will get this login method okay I'm

telling about the first time if you are

running it inside the first time you

will get in a two uh you will get uh

alongside this installation okay uh or

else what you can do if the installation

is completed if you're not getting the

login method then once you will write

the cloud it will give you the login

method uh then guys what you have to do

you have to decide the uh method uh with

which method you are going whether you

are going with the API one or you are

going with the subscription okay then uh

you can choose out the different

different method they are giving you

this two method okay cloud subscription

and the second is what uh the second is

the API key now apart from that they are

giving you the third method also you can

uh set up the cloud code using the uh

AWS or the Azure credential so if you

have access of the Azure bedrock if you

have access of the Azure openai in that

way also you can uh configure this cloud

A because uh anyways you are going to be

read a model right uh whether it's a pro

plan or API key you are just going to be

read a model so uh you can read a model

from the Azure from the Azure co-oundary

uh cloud foundry and from the AWS

bedrock okay so that option also they

are providing you over here uh then

after configuration guys you just need

to check out from the terminal itself

you can write a prompt you can uh check

out uh you you can write a prompt guys

you can check whether it is working or

not and uh here you can use it in a

various way. So you can understand the

codebase, you can generate a new

project, you can fixes the work, you can

refactor the code, you can write the

test cases, anything regarding the

development you can do using this cloudy

code. It's going to be a very good tool.

Okay. Now uh guys uh here is a way right

to set up inside the window. Either you

can open the PowerShell or the cmd. So

let me show you the step-by-step

process. Uh actually I already installed

it inside my system. So what I can do I

can uh simply open the cmd and after

that guys I can uh write over here this

cloud A. Okay. So once I'll write a

cloud A. So this is my cmd either you

can open the power cell. Now guys

spelling should be correct otherwise it

will give you the error. So once you

will write cloud A guys see my cloud A

is already in uh my cloud A is already

installed. Okay. So uh here it is asking

to me uh do you trust to this folder

means uh can I uh take access of this

folder if I will say yes then it will

take. If I will say no it will not take.

So I'm just hitting the enter. Okay

after keeping the arrow over here. So if

I will hit enter see the cloud code is

activate over here. Now uh here what I

can do guys I can log out first. Right.

So let me log out first and let me show

you how it's going to be work. How you

are going to be logged in. Okay. So for

log out guys what I will do? So I will

simply write slash and then here I will

write log out. So once I'll write this

log out see I will be logging out from

the system. So my cloud a is going to be

log out right. So let it log out. Yeah.

So it is log out now. So if I'm going to

be routed now so it will ask me sign in

again. Sign in again. So here I written

a cloud. Now see it is asking to me on

which mode you want to to keep. So dark

mode, light mode, right? It is asking

for this particular terminal. So if I

will click on the dark mode, it will

keep on the dark mode. Now this is very

important thing and this is what I was

talking about. So it is giving you three

way to access the cloud a guys to cloud

a model. The one is the subscription

waste. Uh you are uh you can access the

pro subscription, Mac subscription,

teams or enterprise. You can uh keep the

money inside your API account or else

you can use the third party uh platform

also. Okay,

you can use the third party platform

also. Now, let me uh show you guys uh if

uh we are talking about this pro

subscription. So, how you will get that?

So, once you will open this cloud,

right? Where is a cloud? Here's a cloud

guide. Now, here you can see right now

I'm I am over the free plan. So, if I

will upgrade my plan, right, it it is

similar to the chat GPT. You can access

the pro of the chat GPT with some $20,

right? With some $100. So in a similar

way they are also offering you the pro

version. So you can uh uh you can access

it with this $70 of amount you can uh

use this $100 dollar of account as well

right uh you $100 of plan as well you

can where you can uh like use the

maximum capacity of the cloud. So this

is the one way if you have a pro access

or the or the max access you can uh

utilize this cloud code. Now the another

way via the API. Okay. So let me show

you the way of uh how basically we can

access it via API. So for that what you

have to do this is cloud code document.

Now let me go through the API. So guys

see uh how I can navigate to the API.

Just a second they are giving you the

option over here itself or simply I can

write over here cloud a API. Okay cloud

api key.

So once I'll write this cloud API see

they are giving you the uh

platform where I can generate the API

key. So it is similar to the open API

key right. So see they are giving the uh

cloud console. Yeah. So I think once you

will click on the cloud console now

console login this one. Uh yeah so once

you will click on this cloud console

correctly. So from there also we can

navigate. So once you will click on the

cloud console you will be able to get a

option to this API key. So you can

generate this API key. But again guys,

here the free uh means uh you will have

to put some money inside your API

account. So let me show you the my let

me show you my account. So if you will

check my account, I already I already

kept at least $5. Okay, at least you

need to keep $5. See uh on 6th of March,

I kept $5. It completely gone. Okay, I

build a pro project. It completely took

my money. Uh then again uh today I kept

$7 and maybe after this video this money

is going to be vanish. So it takes money

guys it is money oriented. Uh don't

think you can uh use it freely. No it is

not like that. Uh crowded code will take

your money and uh even in an enterprise

also it will it will it will be taking a

very huge amount of money. If the entire

team team is using then uh maybe5 to 10

lakh of budget just for the cloud code

you will have to keep. So uh cloud here

I just kept5 to $6 guys just to

demonstrate you the thing. uh you can

void some credit at least $5 500 to 600

rupees only and then you can practice

with the cloud code if you just want to

practice otherwise you can ask to your

uh project manager or your uh like the

this manager right director to give me

the cloud code access I can try out the

project with the cloud code so they will

provide you the enterprise account okay

I hope this thing is clear now guys why

this cloud code is important the cloud

code is important because it is being

used inside the enterprises. Even in my

company also we are using it. So

recently we have configured this cloud

code and the result of the cloud code

was very amazing. Uh definitely it's

going to be take your money but again uh

the result of it the result out of this

cloud code is going to be very amazing.

So we have uh we already have

GitHub copilot we have ID like cursor

and the integraity where also we can

utilize this co-pilot. Okay. instead of

this cloud code but again if you're

going to be use this cloud e code or

openai GP codeex guys right open also

release the same similar kind of

platform which is a codeex so yeah the

result is going to be little different

but uh I did a coding with a github

copilot also that is also pretty amazing

and uh even I tried the cloud codeex it

is a money oriented but it is also

amazing right is still I'm exploring a

depth of it right now uh uh I hope this

thing is clear now coming to the next

card. So, uh I hope guys you understood

how to set up the cloud accord and all.

Let me again give you the walk through.

So, which cloud I am, which uh account

I'm going to be choose over here. So,

I'm going to be choose this Anthropic

console account. So, I will uh go down

and then I will hit enter. Now, once I

will hit enter, so it will redirecting

to me to the different uh page. Okay,

this is the page. Now, once I'll click

over here authorize, so yeah, I am

authorized to use. If you don't have

money over here then it won't allow to

you. So please keep some money inside

your account otherwise you will not be

able to access that. Now see login is

successful. Now if I will hit enter now

see uh I can access the I can access the

cloud. Okay. So see guys I'm able to

access the cloud A and it is configured

inside my system. It is very easy right?

Nothing is there. Now it is having

several command many commands even you

can give the simple prompt also. So

let's try out inside any folder. Okay.

Then I'll show you how you can configure

it inside the VS code and then we'll

talk about the project. So uh first let

me exit from here and let me go inside

any project. So once I'll write the

diir. See I have a various project over

here inside this uh directory. So I'm uh

going inside one of the project. Let's

say this is my project. I'm just going

inside this project. I'll write cd and

then I'll give the project name. So what

I can do guys? I can copy the project

folder name or I can simply write that I

think fishing website. So this is the

project which I created uh using the uh

AI only and u today I'm not going to

show you this project. I will show you

something else. So for that also guys I

kept the entire node. See this is the

entire nodes and definitely we are going

to discuss this entire thing. Okay. Now

uh here what I'm doing so uh I'm uh just

initializing the cloud now. So once I'll

write cloud a now see uh it is asking me

uh access right of this folder I am

saying yes you can uh I trust this

folder so yeah now it is having access

of this folder now here I can ask

anything uh hi can you explain

explain this project can you explain

this project now see it is uh initialize

so it is it is trying to read the entire

file and all everything So yeah after

that I it will give me the complete

explanation of this entire code uh just

by giving some prompt and guys see it is

reading the entire code. It is reading

the entire code, right? And uh see what

it does. A machine learning system that

detects what whether a URL is a fishing

website or is a legitimate.

This is the architecture here is a key

component. Right? Uh this is a tech ST.

Very good. Then flow summary. Uh okay,

it did not uh tell me the flow summary.

Maybe it told or not. No. Yeah, here is

a flow summary. Right. If I will like

press the down arrow then I'll get that.

So guys it is able to explain me the

entire project and even I can ask

anything to this and it is going to be

work in the next level. Uh definitely I

can show you that but not here I will

configure it in properly in a VS code

and there only we'll discuss about the

project. Now which project we are going

to develop. Uh that is very important

thing. So project wise guys I kept the

entire uh uh I kept this entire use case

over the blackboard itself. Uh now uh

step by step uh one by one let's uh

discuss this entire use case and let's

try to develop that. So guys uh here I

return a very important line. Now before

starting the project uh let's uh read it

out. So I written over here anyone can

generate a code with AI today but a real

engineer understand the architecture

behind the system. So uh the meaning of

this line is uh don't become a bio coder

become a real engineer guys. uh what I

have seen nowadays everyone is

generating a code everyone is uh

creating a project or everyone is uh

developing their product but most of the

people they don't even know the

architecture or the fundamental behind

that code. So guys until and unless you

don't know the real fundamental right

until and unless you don't know the

exact meaning of that application where

it's going to be used what should be the

correct architecture of it it's not

going to work in a production at all so

what I would suggest you I would suggest

you generate a code but before that uh

clear your fundamentals until and unless

you don't know the real meaning of the

programming you don't understand the

python you don't understand the

fundamental of AI that application may

be it might be a disaster. Okay, because

see as a individual learner right as a

individual learner or as a individual

builder you can develop anything right

but whenever you are working for any

company whenever you are working for a

real enterprises there the product the

application which you are going to build

it is having some meaning. Now before

starting of any project you will have to

understand the problem statement of it.

Okay. So the second line is same you

will have to understand the real world

problem statement before starting any

use case and that is a requirement of

the industry. Then you will have to

understand the user flow right. So how

the data is flowing in how the input and

output is flowing what will be the input

what will be the output how the user is

going to be use that then uh you will

have to understand the complete backend

flow then you will design the UI right

then you will design the DB and the

knowledgebased flow right so if you are

going in this specific order if you're

going in this specific order then only

your application is going to be

successful if you are randomly doing any

W coding or anything guys In that case

your application might be a disaster or

it might fail okay in a real time in the

production. So please be careful. First

clear your fundamental then only start

writing a code. Uh now uh guys uh here I

mentioned some of the more point. Let's

understand uh what I have written over

here. So these lines also very

important. So do not do this mistake

while using copilots. The first uh let's

say we are going to create a project.

Okay. Okay. So in today's video guys, I

will show you how you can create this

customer support email system and this

we are doing with the help of AI agent.

So guys, if you're going to be create

any system, if you're going to be create

any application using the copilots, then

don't directly give this kind of a

statement, create an AI agent that can

handle customer support email

automatically or build a full production

grade multi-agent customer support uh

system or build an AI email assistant,

right? That reads incoming mail,

understand intent and generate

appropriate response, right? So if

you're giving these kind of a statement

to any co-pilot okay so it might fail

right where it it might fail. So these

are the point which I highlighted. Now

let's try to read it out. So if you're

directly asking to the cloud echo to

build everything right in that case what

will happen you know uh the design of

the system is going to be inconsistent.

Okay, files may be generated randomly.

Whatever files you wanted to generate,

it might generate here and there and

because of that the imports may break.

Okay, the architecture may become weak

and the production readiness may be

slow. Yeah. So guys, if you are giving a

single prompt, right, if you are giving

these kind of prompt or direct prompt to

the co-pilot system, in that case your

co-pilot system might be a pro might it

might create a problem. Okay. And these

all are the problem which I have

highlighted over here. Now how to use a

co-pilot? So here I written couple of

points. Let's try to read it out this

point. So if you are using a co-pilot

then use it in a smart way. Now what

could be that smart way? Do not generate

the entire project at once. Use work

module by module. Okay. Use an exact

prompt for each module. Verify the

generated code. Perform testing and then

only do the integration. Understood

guys? So we have to go step by step. If

you are not going step by step, if

you're not deciding what will be your

backend flow, what will be your UI, what

is going to be a data flow, in that case

your system might fail. So that's why

I'm that's what I'm saying guys. Uh

whoever don't know the programming that

person also is going to be generate the

application that is fine. This is good

for the individual learning. But guys if

you are working on an enterprise level

then guys your fundamentals should be

clear right and in future you will see

people companies are hiring those people

only who knows the fundamental okay the

system uh their uh system architecture

concept is very much clear. If your

architectural concept is clear, if you

know the fundamental, if you have worked

on some project, right? If you worked on

some project, then you can easily do the

W coding. And then in that case, W

coding is going to be your uh then in

that case guys, W coding is going to be

a very uh efficient productive tool.

Okay, you can achieve a highest

productivity using the co-pilots and

all. And even I felt the same when I was

working on some projects. So I hope guys

this fundamental thing is clear. Now

coming to the next part. So what we are

going to develop inside this video. So

guys inside this video we are going to

develop AI powered customer support

email agent. Okay. Now you know guys

email communication is a very important

part of almost all the industry right.

Uh this agentic application can be used

across many industry. In every industry

we are going to be use the email

communication right. Uh without email

communication none of the industry is

going to be work right. So we can use

this particular system across the

industry and how how it will look like

we'll see at the end. But first of all

let me tell you the complete flow of it.

Okay. How this entire thing is going to

be work. Now what is the goal of it?

What is the goal of this uh system? So

goal basically it will generate a faster

response than human. Right? Human is

thinking then he's typing. Now here uh

it is going to be generate a bit faster

than human then lower support cost. Yes.

Uh like if you're going to be hire 10 to

15 people right for this customer

support there the single system is going

to fulfill your requirement. So here it

will lower your support support cost.

Okay. Scalable support operation. Let's

say guys you wanted to reply for uh you

wanted to reply for uh thousand mails

right 1,000 to 2,000 mail. Now in that

case uh how many more people will hire

or let's say your mail flow is very uh

very huge in your organization in your

company you you are going to be uh you

are going to be generate like let's say

lakhs to two lakhs mail okay per uh

months or maybe per quarter and you have

to handle that by a customer support.

how you will do that? Okay. So, how many

people how many like teams you are going

to be build for that? So, it's going to

be a difficult for you. And now here

this system basically it is for the

scalable support which can help you even

for the uh lacks of lacks of email. Uh

now I hope guys the goal and the uh idea

of the project is pretty much clear over

here. Now coming to the next part. So uh

what is going to be a input and what is

going to be output of this application.

So input wise uh you will get any sort

of a input. Let's say here is the

example of the input. Uh in my email I'm

getting I am I was not I was charged

twice for my subscription. Okay. Now

here customer uh support will help this

people this like person. Okay. Uh they

will tell to this guy how much you

charge. Okay. when you paid and all this

thing they are going to be asked right

now but instead of the uh the real human

now this uh thing we are going to do

with the help of AI agent okay so we

have a so we are getting a mail once

we'll get a mail okay my agentic flow is

going to be triggered so first it will

understand the uh incoming mail it will

read the incoming mail it will classify

that so uh I might get a mail regarding

the different different like uh query so

The query with respect to the product,

the query with respect to the bug

report, right? The query with respect to

the billing issue. So this particular

query, the example query basically it's

going to be a blinking related query.

Then feature related request complex

issue. Okay, we have a complex issue.

Then we are going to redirect that to

the human. Okay, then what we are doing

guys, we are going to be search or we

are going to be search over the

knowledge base. If any sort of a

information is going to be over the

knowledge base over the uh documentation

or uh in any sort of a SOP right so we

can pick that information from there

okay we can pick the information from

that specific documentation now

according to that we can draft the

responses okay for the customer if any

complex issue is there if any urgent

case is there right if it is not going

to be solved by the AI then human can

intervention over there okay and then it

can send the final reply to the customer

and even it can maintain the foundation

for scheduling the follow-ups. Getting

my point? So this is the work this is

going to be a complete flow of this

entire application which we are going to

build using the cloud code. Okay. And

I'm not going to be write a single line

of code. Uh the all the thing will be

done by the cloud code itself. I hope

guys this thing is clear. Now the com

next thing is the design flow. So uh

this is going to be a complete

technology ST guys. So here we are going

to use Python, Langraph, Langchain,

Langchain, OpenAI, fast API, pyic or

vector database. So this is going to be

a complete technology stack okay for

this entire workflow and uh the same

thing will provide to the cloud A and

behalf on this technology stack my cloud

A is going to be generate the code.

Okay. Uh then guys uh here you can see

this is my folder structure. Now again

guys uh this is just a normal folder

structure which I kept whenever we are

working in the industry right so there

is no predefined folder structure okay

uh if we are going to build a micros

service right if we are going to be

deploy our application in the form of

microservices then the uh folder

structure could be different uh if we

are going to be like deploy our

application in the monolithic right so

the folder structure could be different

uh there could be server worker service

right then API then UI layer right so

the multiple layer could be so here I

created one uh generic folder structure

but guys more and less the folder

structure will be like this whether you

are using the microservices or whether

you are creating a monolithic

application right in all the situation

uh the core technical files and folder

right it will be arranged like this now

apart from this you can create as many

as file and folder according to your

requirement now what whatever you wanted

to keep whatever uh middleware, whatever

workflow or whatever like configuration,

notification you wanted to keep, you can

create a files and folder accordingly.

But guys, more or less the core

architecture, okay, the core code uh you

can keep it inside this file and folder.

So this file uh and folder this

basically complete folder structure I

just created just to show you how my

application may look like. Okay. And

even the same kind of instruction we can

provide to the cloud A also because as I

told you if you are going to generate

anything using the cloud A right uh in

that case cloud A may generate anything

any uh uncertain thing right so if you

wanted to certain if you want to take a

uh like the complete control of the

project in your hand in that case you

have to define everything guys and

that's what I'm saying from the starting

of this video okay even if you're a B

coder but guys if your fundamental

system architecture knowledge is clear

then your application is going to be

very robust and it's not going to be

break inside the production. Okay. So I

hope this is also clear. So same kind of

architecture maybe we can utilize for

this application also and more or less

it might be different. Okay. Now over

here guys uh one more thing I have

written over here. See whenever we

talking about the back end right. So in

back end what we design. So in back end

guys we uh will be having the detail of

all the module we are going to be decide

the connectivity between the module okay

connectivity between the module then

guys we are going to be discuss the API

design then we'll be having the API

layer then in the back end only we'll be

having the data database architects

knowledge based architect uh knowledge

based architecture then a logging design

exception design okay and the low-level

class design which class pattern you are

going to be use okay So according to

that only we are going to be decide the

class design. So these all the things

actually comes under the backend design.

Okay. The low-level code where we are

going to be write for the main

functionality for the main business

logic. So this entire thing actually is

going to be recite over here. Module is

nothing uh so the python file we we are

talking about the module. So python file

itself is called the module. So if I'm

saying P module in a Python code in a

Python project that doesn't mean it is a

Python file only. Right? Now over here

guys we're talking about the UI. So

again UI is a very subjective thing. It

might vary according to the requirement.

So here also we'll build one UI and here

I return one uh uh basically uh

statement. Okay for that and yes

according to this only we are going to

build a UI. Now if we're talking about

the deployment guys so yes I will come

to the deployment that how we can handle

the deployment using this kind of

project means if we are going to build a

project using the cloud code right so

how we can handle the deployment now see

guys a deployment side means cloud a

cannot log my AWS console it cannot

create a service like EC2 ECS lambda

okay from the console itself but it can

do via cloud formation it can do using

the terraform it can do using the CDK.

Okay. So guys, uh cloud A cannot

directly log to my AWS console but it

can handle thing at certain level using

the script. Getting my point? So the

same kind of thing basically I have

written over here. Just try to read it

out your uh all the doubts related to

the deployment is going to be clarified.

So uh here you can see cloud does not

automatically create or configure AWS

services by itself. Means uh it can not

open your browser okay over the browser

it's not going to be search over the AWS

then it's not going to be uh write your

email id over there or it's not going to

be write your password there okay then

uh it's not going to be configure the

service like EC2 ECS okay so all this

thing all this thing matters a lot while

we are doing a deployment but manually

this thing is not going to be happen and

even in the industry we don't do such

thing manually right So how this thing

is going to be done? So cloud a code

cannot directly configure AWS services

by itself but it can generate the

infrastructure as a code and this is

what we are following in the industry.

But again guys it needs to be followed

under the supervision of the DevOps

engineer. If it is not going to be

followed under the supervision of the

DevOps engineer in that case it might be

a very big disaster for the company

because it might generate anything.

Okay. And we even don't know like what's

going to be generate and what kind of

service is going to be create in a back

end right means whatever services we are

going to be write maybe it might

multiply that because just like AI right

so it's not a guarantee that it's going

to be follow a similar kind of

instruction which we are giving there's

90.9999% chance it's going to be follow

the same but there could be mistake 0.1%

could be a mistake so we have to do this

thing under the supervision of the

DevOps engineer right so cloud basically

it is capable to write a config

configuration it can perform

infrastructure as a code when the code

runs through the CI/CD pipeline that

requires AWS are automatically created

and configured. So we just need to

configure the AWS credential inside the

uh uh GitHub if we are using GitHub

action genkins uh if we are using a

genkins pipeline. Okay. And how we can

do that? So uh we we can do that

manually configuration of the credential

and all and then we can write a script

in the terraform cloud formation AWS CDK

right this is all going to be a

deployment script right uh

infrastructure as a code means we are

going to be build a infrastructure using

the code now we can create these many

services even any services using that

configuration okay so here I I have

written one very important line just try

to read it out it does not directly log

into the AWS console but it generate

infrastructure as a code that instruct

AWS to create them. I hope guys you

understood. Now let me tell you the

final capability of the cloud A. So with

that guys your understanding is going to

be very clear. So cloud A can write

a code. Okay. Cloud A can push it to the

GitHub. Okay. Cloud A can trigger the

GitHub action

pipeline.

Okay. Cloud A can help us to write

infrastructure

as a code and using this infrastructure

as a code using this configuration

right infrastructure infrastructure as a

code configuration right we can create a

infrastructure we can configure the

services over the AWS

right I hope you are getting my point so

using the configuration ally using the

infrastructure the service only right

which we can do using the terraform or

using the GitHub action configuration

right uh we can uh basically create a

infrastructure over the cloud and there

this cloud code can help us I hope you

understood it now inside this project

how much we are going to do so guys we

are doing a development level of work

right inside this particular project and

in uh rest other video maybe in some

other video we're going to see the

deployment as well. Okay, because as of

now it might be little advanced and it

might be the expensive as we are already

paying to the cloud code. So the

deployment side we can look uh in the

another video. Okay so yeah I hope guys

you understood the entire thing. Now

let's uh start with the cloud code.

Let's start with the uh building of this

project. Now first guys uh open your cmd

and create a directory at any location.

You can directly create a directory or

folder from the file of explorer also.

Okay just go inside any drive and create

a folder there. Now here I'm going to

create a folder at this location c user

sunny. It could be anywhere inside your

system. So what is a command for

creating our directory? The command is

mkdir.

And then I'll write a folder name. My

folder name is customer

support

email

agent. Then I what I'll do guys I will

uh write a cd then I will provide a name

of my folder customer support email

agent. And now here at those location I

will open the uh VS code. So how to open

the VS code guys? At this location so I

will simply write code

dot

Got it. Now see I open the VS code and

uh I open the VS code at at the same

location. You can verify it. You can

open the terminal. See here is my

terminal and you can clearly see I am at

this location. Okay. And this is the

this is my workspace and now here only

I'm going to code my entire project. Now

guys how I will code my entire project.

I'll code my entire project with the

help of cloud A. So first of all what

you have to do you have to configure the

cloud A inside your VS code. So you can

do one thing you can uh initi initialize

the cloud A over here. Once you will

write the cloud A now see the cloud A is

here. Okay, you can uh give any sort of

a command to the cloud here. Let's say

you can write you can say uh create a

requirement txt file.

requirement

dot txt file

with all the necessary

requirement

okay for the agentic

project.

So once you will give this uh command to

the uh cloud a then see what it will do

guys it will create the require.txt with

all the requirements. So it will take

some time it is thinking and uh the file

will be created over here.

Okay. Now see guys it has created a

file. It is asking to me whether you

would like to proceed with that or you

would like to uh see here it is giving

three option. One is yes. Second yes

allow all the edits during this session

and then no. So what is the meaning of

it? It is asking to me to confirm this

file. Okay. The meaning of yes is the

same. So it has generated a file. I just

need to confirm that. The second

basically I need to confirm it. But

yeah, I am giving access to edit the

other file also and even add it to this

file also during this entire session.

Right? Not only for this command for the

upcoming command also it will

automatically edit the file. Third is

no. Okay. Simply I'm denying this thing.

So if I will say yes. Now see my file is

created over here.

That much easy guys. That much easy.

See.

Okay. Now if you if you want to install

this fi file recon txt uh let me

increase the size just a second size of

my uh I think now it is visible

perfectly so guys how easy and how

simple it is just just think about it

right now see I'm working through the

terminal now if I want to work with the

cloud copilot how I can do that so for

that guys uh see I I configure the cloud

over the terminal right now and I'm able

to work. Okay, but for the cloud a uh

copilot I will have to install the

extension. So just go and search the

cloud a over here.

Cloud A. Okay. Now once you will search

the cloud A. Now here guys see this is

the cloud A. Open this cloud A and

install this cloud A. It is already

there in my system. I already installed

this uh particular plug-in inside this

VS code. Now how you will confirm that

you have this uh cloud extension or not.

Now right hand side guys just open any

file. Okay let's say this file. Now

right hand side you will get this this

symbol. Okay. I hope it is visible. It's

not going to be hide. Uh I think so but

yeah it is visible right. So you will

find out this icon. Okay. Now just click

on this icon. Now once you will click on

this icon see here is a cloud a code.

You can chat with the cloud A. You can

do anything whatever you want to do.

Right? So I'm uh pressing Ctrl C. I'm

not going to handle this cloud a code

using the terminal right now. Instead of

that what I'll do guys I will use

instead of that guys what I'll do I'll

use this particular chat window. Okay.

From here itself I'm going to write any

sort of a command and I'm going to build

a application. So uh let me delete this

kind of txt. I will I will uh create it

later on. So I I'm going to be delete

that. Okay. Now what I will do guys?

Just a second. Where is a cloud code?

This is the code right

and here is a cloud code. This is the

GitHub copilot. The by default GitHub

copilot. If you will click over here now

you will get the by default GitHub

copilot. Here is a cloud code. Okay. Now

over the cloud code also

uh

over the cloud code also guys what you

will see you will see a different

different option okay over this chat

window I hope it is visible let me keep

my webcam left hand side for just for

right now sorry just for a couple of

second uh yeah so now see guys this is

the entire screen of the cloud record so

here you can see the past uh here you

can see what here you can see the past

basically

uh conversation right so this is the

past chat which I initiated now apart

from this one you can see uh the

different option ask before edit then

edit automatically then here plan mode

okay now here guys you will select your

file in whichever file you want it to

work right you will select that file so

that cloud code can take a context of it

then uh here is one of the option you

can select a file and folder means you

can add that file and folder as a

context so here uh guys That file and

folder will be visible and cloud will be

using as a context. Then uh here guys

you can see so the different uh option

you can attach a file same one file and

folder then mention file from this

project you can mention the direct file

from this project you can clear the

conversation you can switch the model

you can think okay account and uses you

can check everything from here itself

toggle fast out mode opus 4.6 six only

then output a style agent hopes memory

permission everything you can control

from here and then here you can see some

commands as well. So using this uh

commands guys you can check out the loop

inside init headup debug cost right

everything you can check out from these

particular command. So they are giving

you every sort of a option to this chat

window. So instead of uh like uh instead

of managing everything from the terminal

you can take a help of this cloud code

window okay in a VS code the same is

available in the cursor and in a other

ID also and it is a much more easier

right it is similar to the GitHub

copilot now what I will do guys so I

will develop my application so I have

selected the haiku model it is a cheaper

compared to the other model I don't want

to waste much money right now just to

showcase you this uh example. Okay. Uh

otherwise I could have used the other

model also. Let me show you the other

model. If I want to switch then you can

check over here. Sonet is there. Opus is

there. Oppus 1 million context is there.

Hiko is there. Okay. So these all are

model we have. If I'm going to select

the sonet or opus. So I'm going to like

spend my token in a much faster way and

it will charge me very high cost. Okay.

The uh cred credits is going to be

reduced very fast. the five or $10 of

credit right so for the practice guys

you can uh use this ha only uh now

everything is set so I will provide some

prompt to this cloud a and let's see

what's going to be generate so I return

uh couple of prompt already uh let me

keep it over here so this is going to be

a prompt guys this is going to be a

prompt this my first prompt okay so I

return it before the session itself and

I'm just going to be paste it over here

so what is my prompt guys. So I want to

build a langraph based customer support

email agent in Python. Okay. Create only

an initial project uh create only the

initial project is scaffold. Uh what is

the meaning of this is scaffold? Is

scaffold means the project architecture

not the full implementation yet. Okay. I

don't want to create a full

implementation right now. I just wanted

to create a uh folder structure.

Okay. Then this is the requirement guys.

So Python 3.11 plus fast API lang graph

lang chain and lang chain openi. Now

here guys I will give the specific

requirement. Uh I don't want to use uh

python 3.11. Okay or uh the latest

version 3.14. I want I just wanted to

use python 3.12 for this project. So I

will write over here python 3.12. Okay.

Use specifically

use

specifically

specifically

Python 3.12 past API langraph lang lang

openi pyntic okay also generate

require.txt txt.nb

readme. Now this is going to be a folder

structure. So I can uh say over here uh

you can take a reference

of the above

folder structure

but feel free

feel free

to add any other

files and folder according to the

requirement.

Okay, according to the requirement. So

this is what guy this is my prompt. Now

here see uh I given the clearcut

instruction. So I return do not heavy

code yet. First show me the plan then

create the file. Okay I can remove this

one. This is fine. Now after that guys

what I will do? I will install the

require.txt. First let this thing is let

this thing be create. Now guys just see

the magic of the cloud. Okay. Now here

uh basically I have selected a option

where uh cloud is asking me about every

step. Okay, it will not do anything

without my permission. Why? Why it will

not do anything without my permission?

Because I selected that option. Okay, if

you will I don't know you have seen or

not but uh below there was a option ask

before edit right or automatically edit

or plan the thing right. So I kept the I

I I opt the first option. So it is

asking me for every sort of a it is

asking me for every sort of what every

sort of a action. Now here guys you will

see that it is going to be used the

batch terminal in a background process.

Okay. So here it's in the inside this

workspace it's going to be create all

the file folder and all everything. But

guys how it is going to be create that

how it is interacting with my system for

creating this file and folder. So guys

here it is going to be interact with my

system for every file and folder using

the bash terminal. Uh you will not see

anything over here over this terminal.

Okay it's going to be use the best

terminal in a back end. So yes it is

asking to me do I need to create all

this file and fold or something. I will

say yes create that. So I said yes. Uh

now guys see data is there. Now src is

there. Now here is a test. Okay. Now it

is again asking to me sunny uh can we

write a requ.t txt can we create that uh

according to this 3.2 two. So I would

say yes. Okay. Now see it is going to be

create. Okay. So this is all the

require.txt

uh which I can see over here. Even it

will show you right what is being

generated. You can simply check that

what is being generated. See this is the

file. Uh this is the file guys. Uh just

a second what it is.

Okay. It's av file right? It is av file

and open AI key, open AI model

everything I can keep over here right.

So yes uh we can generate as of now it

has created the environment.xt is

already there. So it is asking to me

should I generate that? I would say yes

it will create that. I will say yes

allow all this session allow edit all

the session. So it will not ask me again

and again until this chat is going on

now. It will not ask me again and again.

It will be creating all the file. So

here I will I would say yes. So see this

file is created. Env is also created.

Okay. Then it is thinking and it is uh

creating some other file. Let's see what

is going to be create next. Now guys I'm

not doing anything. I'm just sitting.

Okay. My AI is working for me. Now see

it has created the readme file also. So

if I'm going to push this entire thing

over the the GitHub then uh it is going

to be a very good representation of the

project via this readme file. So if I

will say yes now see the readmi file is

also there. Redmi file is also created.

Okay. And uh after that guys see what is

going to be happen. Let's see. So here

it is creating a toml file also. Okay.

Create the toml file. This is also

perfect. And now what it is doing guys?

It's going to be create the main py.

Okay. Let's uh let it be create. Ben py

is also going to be create. Okay. Now

after that guys here it's going to be

create the config. py. Okay. So this is

also fine guys. It's going to be create

a config. py great and it will create

guys it will create every sort of file.

I'm not doing anything. See it has

created a logger for me. Okay go ahead

for that. So I can see I can uh verify

that means it it is giving me option to

verify all the file and all first then

only it will proceed further otherwise

it will not proceed. Okay. If I have to

do any any man manual changes inside

this particular file I can do that.

Right? But here guys I'm not putting my

mind right now. I am just sitting okay

holding with my coffee or with my iced

tea or with my beer whatever right so

yeah I am going to be create a file so

here is app py is also created now let's

see what all other file is being created

so email py so it is saying sunny we can

create this also email py as of now just

it it is just creating a folder

structure and it's just doing some

highle coding and all not the complete

in-depth okay so it is creating the

graph py Y also. So yes, this is also

created. Later on we'll give the proper

use case to it and then it will go

accordingly. Okay, it will make a

changes accordingly. So init. py is also

created. This is there. Now after that

guys, what is going to happen? So it is

checking everything. Allow this command.

Okay, this is fine. So this is also

perfect.

Okay, perfect. I think it is uh just

cross-checking right that all the file

is created correctly or not. It is

running everything in a bash terminal in

back end. See the bash terminal is going

on. Sorry the all the command is being

executed over the bash terminal. Okay.

So allow this command. What is this

command guys? Okay is checking this uh

pent classes.

Okay. So this is in it. Perfect. Now

here is my complete folder structure. So

it is created a template py also this is

great prompt is created I don't know

guys what kind of okay so guys it has

detected the flow also how it is doing

that with the simple like uh prompt only

I think it is taking uh because I built

uh earlier actually uh two days back

just to practice so maybe it's taking

context from there but yeah we'll we'll

modify that Okay, it's not stopping

guys. It's not stopping. Uh let it

create. We'll do the changes later on.

Okay, so for the health also it has

created a file and folder guys.

literally this much of coding basically

I would have taken around uh maybe one

or two week just to write this much of

code and project but uh it has created I

think within a 5 to 10 minute only and

that's the power of the AI right now and

I have used the simplest model actually

see this is uh so customer support

development MD okay one more MD file it

has created that's great now what will

happen okay this is what guys envam

example. Great guys. So one more file is

there. This is the example which I can

share with anyone so that someone can

refer that. Great. Helper py inside the

util there is a helper. py. Okay. Then

envamp example. Okay. This is fine. Now

it is testing that display directory

structure. Great. Output customer. Okay.

It is checking whether everything is

fine or not. The complete structure and

all. This is the project directory

structure. Great. Oh, wow. Uh key file

generated file config txt. Oh, now it is

[laughter]

guys. Uh it's really amazing. Really

really amazing guys. So it has done I

think most of the thing and uh yeah this

is like really awesome. Great. So what I

can do guys now I can create one UV

environment over here. Even though I can

give a command to this to create a UV

environment but I think it's not going

to be interact with this terminal. Uh

GitHub copilot can do the interaction

with this terminal but uh cloud code I

don't know I tested but it was not able

to do uh instead of that in a back end

it was using the batch terminal. Okay.

So let's do one thing guys. Let's create

one environment over here. It is giving

me a command for creating an

environment. I can create a UV

environment or normal Python

environment. Both is fine. Then I can uh

activate this environment. Okay. So the

environment is created. I can activate

this environment.

Source is not applicable for this. Okay.

So live

script

and this is my environment. So copy the

path

and paste it over here. Okay. Now it is

done. Now pip install.txt.

So this is the file which I have to

install. Great guys. So it has

installed. It is saying Python imp is

not there. So if I will give this error

to this cloud only it will

fix that.

Uh let me check the Python version. Oh

guys it is 3.14.

So it's taking a latest Python version.

Because of that only it is giving me

this particular issue. So what I'll do

guys? Uh I will create a virtual

environment. Let me delete this virtual

environment. So here I will delete it

first. Okay. Now here I will give ask to

the cloud only. Uh give me a command.

Give me a command

to create

using

UV. Okay.

and give me a command to create a

virtual environment

virtual

environment using UV and

with Python 3.12 only.

So let's see what it is going to be

generate. Maybe it will give me a

command so that I can run it over here.

Otherwise I can ask to this only uh to

create a basically environment in a back

end. it will run the back end process.

It it will create it through the batch

only. So, ubnv python dot okay this is

the command I don't think so it is

correct

uh you know python let me check it out

so first guys maybe

okay so this is the command let me check

it out that it has created great now I

can

use that okay great so guys my

environment has activated now this is

the environment customer support agent

Now what I'll do I'll clear it and then

here I will write the command ub pip

install

- r requirement dot txt. Okay. So it is

saying langra depend the langent core.

So just a second I can again ask to it.

Ah now it is correct. The python

environment is correct. Okay. Now I can

give this error to this cloud only.

Let's see what's going to be generate.

Uh okay. So it has detected it. Let's

see. H it has ver verified.

And uh yeah it's doing that guys. It's

doing that. Pi projectl. It has updated

this file also. Great. Even we can

verify. Okay. So don't like do it uh uh

right after the uh generation. Okay.

false verify properly and then only do

it. Now let's run this command. UB

uh okay this is again saying something.

Let's see what it is saying.

This dependency issue comes uh with a

specific Python environment.

So let it be how it's going to resolve

that. I'm not searching anything guys. I

left everything over the AI itself.

Great. Now let's see what is going to

happen.

Uh again guys, it is not going to be

solved

here. Maybe the AI is going to be fail

or let's see.

Okay, this is perfect.

Uh uh this is fine. First, let me clear

the screen and let's see.

Okay,

guys. We're stuck. It's stuck. It's

stuck.

It's trying very hard [snorts]

otherwise I will have to change the

model basically.

Uh let's see now first of all let me

delete this one. I'm going to open a new

terminal

and then I'm writing uvp pip install

- r requirement txt.

Okay guys, so now it has done. Uh, okay.

It is up install.

No sh resolve dependency because only

version of pyic are available.

I think some issue with the pyic. Now

let me check it out what they are giving

to me. Pyic setting. Okay, no worries

guys. I'll copy and paste this.

Otherwise I will ask to cloud only to

resolve it in a back end. Okay, means it

will create an environment in a back

end. It will do that. So let it be

uh okay take this is fine now.

Okay now it is saying do it.

No sorry founder of python imp.

Oh very blunder guys. I think uh this

model haiku is not uh like up to the

mark.

>> [laughter]

>> It's taking many iterations.

I think my entire token is going to be

empty over here itself. I think the best

way is to ask uh this cloud only to

check in a back end. See guys, I'm not

cutting the video and all. Uh I will

show you everything whatever challenges

and all I'm facing. Okay. Uh maybe now

it is done. So it has solved that issue

[laughter]

how does so many iterations. So even I

see guys if you're getting stuck now at

this kind of situation and all. So one

thing is keep trying right AI will give

you the answer at the end or else the

second thing what you can do you can ask

the this cloud copilot itself to create

the environment first to install the

requirement if it's going to be suffice

right if it's going to be fulfilled if

it's not generating any sort of error

then give me the yes and go ahead with

the final version of the with the final

version of the requiretxt. So this is

the idle way of doing it. So yeah fine I

think my environment is ready. My

environment is set up. Now it's time to

give the uh another instruction to my

co-pilot. Now how I can do that guys. So

here I return the complete flow of my

application. So this is the complete

flow guys. Uh see uh first uh let me

give the instruction. So this is a

complete flow

of my

application.

Okay,

read it

carefully

and build accordingly.

Okay,

build accordingly. Right now here uh see

what is the flow. So read the incoming

customer email. Classify the intent.

Search into from required document if it

is required. Draft an appropriate

response. Human review is required if

needed. Then send reply. Okay. So let's

see how it's going to be code. You can

write much more uh like effective

prompt. Okay. You can give like so many

with so many condition. It will be able

to take a very huge prompt as well. But

as of now I'm just uh uh I'm just giving

a basic prompt and the stepbystep uh

prompt right but you can give a very

huge and the uh very technical prompt as

well it will be able to understand right

it is just a dummy project and how to

use the cloud code how to build a

product project it is just about that so

I'm like just showing you on that level

right now so yeah I given this prompt

let's see how it is going to be deal now

I am just sitting okay [laughter]

let's see guys how it's going to be

built now. So yeah,

how much time? I think it is uh it will

take some time guys. Let it be.

Okay,

it has added more packages something in

a readme file.

I hope so guys it will be able to do

that. I'm not pausing a video. I'm not

doing anything guys. Uh, let it be.

I'm just sitting in front of you

checking my phone in between. Let my AI

work.

Yeah, my AI is working. So, what they

are doing right now? Let me check that.

So here guys, it's going to be added.

See guys, it will give you the uh see

here it is giving you the option where

you can accept that you can revert that.

Okay. And here you can eval that in a uh

with the current uh workflow and all so

many options they're giving you over

here. And even you can run it you can

run that you can run it run it in a

debug mode. Many more thing you can do.

Okay. So if I'll say yes so it's going

to be proceed.

Got it. So now I think it came in a

different file. So it is working with

this database py. Okay. So see if I'll

click on yes. So yeah this is accepted.

Okay. So here also you can click this

one. This this bar you are getting now

right. This bar you are getting. So just

uh work with this also or else uh here

also it will provide the option. Now see

ask before edit. So it is asking me

before editing anything. It is asking me

to before editing sorry it is asking me

editing uh before anything. Now see

again yes and uh model py is there but

guys see for uh like uh the enterprise

code for the realtime project you will

have to understand this entire code by

yourself what is happening inside. Okay.

Now again, yes, as of now, I'm not doing

it. As of now, I'm just building. If I'm

going to be verified line by line, it

will be challenging. But again, guys,

you have to do it, right? You no need to

approve it without verifying the code.

You should have the highlevel

understanding of this entire code. Okay.

So, you can do the come to the previous

changes, next changes. Great. I think so

many things. uh ser LM service py. Now

one more thing I required the openi key.

So I kept in one of the file let me take

from there and I will paste it guys

before running before running now before

running this application I'll keep my

open a key and the open a model name

which model I'm going to use. So DB

service is also created

and with whatever architecture you are

going to be work guys just give that

exact flow exact name exact file name

exact class architecture whatever you

will give according to that is going to

be build that okay if you don't know

anything if you just know the

fundamental still you want to do still

you wanted to build the application then

give anything [laughter] okay let it be

on AI itself see this is the complete

code it has created this review service

and something like that. Okay, great.

Now, let me click on yes.

It's going to be create a complete

workflow guys. Don't worry. As of now, I

think it is working on the service side.

So, it is following this architecture,

this uh worker server and the service

one. So, schedule service is fine.

Great. And guys, as of now, I'm not

going to be integrate my Gmail. What I

will do, I will test it with a dummy

email. Okay? I'll show you how you can

test that. So I'll keep some dummy email

over the UI itself. We'll keep it and

from there only we are going to test it.

So llm template py is going to be added.

Okay,

it failed now. Yes. Okay, this template

is also changed.

I hope guys everything is visible on my

screen. Let me

keep me in a small one.

Yeah, this is fine, I think.

Yes,

even you can uh like uh allow to this uh

cloud a to edit everything in a single

go, right? Just click on this edit allow

all edits in this session. Right? So if

I'm going to click on this now see it

won't ask me again and again

it's going to do every sort of edit by

itself. But this might be harmful.

Right?

See here is a uses 55% uses of my credit

limit. [laughter]

I hope so guys within like this 50% uh I

think two to two to like uh $2.5 maybe

within that I'm going to be complete

this application [laughter]

otherwise I will have to add more money

oh it is using log guru great log guru

is a good package for generating a logs

structure log even struct log is a good

one 57%

okay I hope so guys it won't be reach up

to 100%.

Uh, too much code. Too much code.

Followup scheduuling. Great.

Yeah. See guys how it has created the

state the completed state. This is

workflow. py. See this is my langra

workflow. Great guys. Great. Very good.

Email retrieval classification context

analysis review check and all

everything. Everything is being done. Oh

that's great guys.

60% users limit. See here you can see

this one

40 remaining until auto compact. Okay.

Click to compact now.

Okay. It is using the autoco compact

feature I think. So yeah we'll have to

deep dive more to optimize this one

actually.

Now it is not asking me to edit

anything. I kept it on a auto auto

copilot mode and it's doing that. It's

every if anything is going to be like uh

break it will like uh check by itself.

[laughter]

Okay.

Memory is also created memory MD. Why it

has created a memory MD? You can check

with this thinking. Perfect. Now I've

created a complete mean memory. Oh, it

has created by itself guys for for

himself right see here see file document

all the you you can check like what it

is thinking let me read and file get

thinking completed implementary summary

DB layer I think guys service layer is

done database layer is done langra

workflow is done email will come

response generation review routing human

review response great guys it has

created it has created okay great great

see this is a complete folder structure

it has generated in a very amazing way

that's really good guys now what I'm

doing database model service okay API

routes is not completed it needs to be

done it needs to be done okay great guys

so what I'll do I'll give the next

prompt to it so what is going to be a

next prompt

uh can you implement files right so I

will keep every data and all everything

in a file itself and yeah I'm giving you

that I'm giving the same prompt now Can

you implement files as a knowledge base?

create

some dummy

document

store it in a file

so that I can retrieve

the information

info from there and can utilize it

in my workflow. Okay. So guys uh see

this document this data could be

anything in a real time as of now I'm

asking to cloud a only to create some

document to keep some information over

there and out of that document create a

file index but in a real time this

document could be anything getting my

point this document basically it could

be a real data real data right from the

variety of the uh task. Okay. So I'm

giving this prompt. Let's see how it's

going to be deal with this particular

prompt and whether it's going to be keep

the data or not. And it's going to be

get a files on top of it or not. Let's

see. So first it's going to be addit uh

the

require.txt.

So edit automatically. I I selected this

now. See it is taking workflow. This let

me hide this one. See here you can

select this file as a context or you can

not select okay just uh deselect this

file as a context means uh if if you're

going to be select this file now cloud a

code is taking a context out of this

file as of now now we as of now we are

not giving a context from any specific

file so I just deselect it over here see

this is the deselect one if I will click

on it again it will be selected as a

context okay

so recxe let me check uh it has added

the files is uh related.

Yeah, it has added. Okay, great. This is

great, guys. Now, what it is doing, it's

like very good. See, even though I'm

using the Haiku model, if I'm going to

use Opus now, guys, it's going to be a

blaster.

But guys, it will uh suck your money for

sure. Uh I spend uh just like 1,200

rupees to show you this de

demonstration. Okay. [laughter]

Buth if you are going to be like uh do

it in a real time then it's going to be

suck your money that's for sure.

So AI is not affordable right now. The

premium AI I'm saying

74% used.

I believe guys I'm not going to be lose

my all the money

and before that only I will be able to

build this application.

76% used.

Okay. HighQ is there.

Thinking mode is there. Target. I think

if I'm going to be like stop this

thinking mode, maybe it might not be

used this much of tokens.

77%.

Great.

Okay guys, so it is creating five minute

step. Okay, this is fine. Let me check

where it kept the data. DB is here.

Database.py

uh

src script is there.

service service API

graph schema

service. Okay, I think still it is doing

that.

Oh. Oh, this is a like

this is a command it is asking to me run

that.

Okay,

what where is the data? App. py is

there. Compu db is there. Database.py

graph

nodes template and it has created the

complete one in a blank graph itself.

Okay, this is fine. And every time guys,

it will not create a same code. Okay,

see guys, this is the uh okay, F

implementation, F knowledge base. Great.

This 2 MD it has created. But I don't

know which data we are going to be

retrieve the information.

Uh maybe I will have to give some

specific prompt for that.

It's going to be keep the data

DB models schema nodes.

Yeah, this is fine. Let let's wait guys.

Okay, let it be complete.

Perfect. Now let me create a final

summary showing you everything was

implemented.

Great.

Now guys after that we'll create a UI as

well. Okay. So API and UI is the next uh

target. It has used 85% of my limit.

Let's see how much is going to be expand

now.

Okay. Five is not completed. Customers

of final vector similarity. What you

have now five similarity search semantic

uh 20 sample support document product

information

uh where it has created this product

information

uh setup helper is there this is fine

compressive document of five minute

guide this is fine

create file index of 20 document I think

it has created inside only maybe let me

check a script

so where is a script script.

This is the API API DB

uh schema. Here is a script. Inside the

script, what we have guys? Just a

second. Inside the script, populate

knowledgebase. py. Yeah. So, see guys,

this is a sample document. I think uh

see it it it haven't created any MD file

or something to create the uh index.

Okay. So it has created this dummy data

and it is using that

fine. So this is clear. Now let's do one

thing. So let's say yes and I think this

is done. But guys I can give the

specific instruction where it can keep

the data right where it can keep the

data inside the uh data folder

specifically and then on top of it it

can build a index. Okay. So your

fundamental of the rag the agent should

be clear here. Now I'm giving the final

uh command. So and final prompt uh I

kept over here.

So here is my final prompt guys. This

one. So can you also create a UI API

and UI where we can test the system

using the mock mail and also please

create another page where we can view

all the email inbox email along with

their email details and the

corresponding agent output. Okay, let's

see guys how it's going to be behave on

this particular uh basically prompt. Now

compact now what is this? Uh okay so

compacting let's see I think it will

optimize it maybe

uh great so let me ask now what it is

going to be generated let's see

[snorts]

but guys it's like really uh amazing see

how much line of code it just created

within like 10 to 20 minutes

and even we are able to do the same

thing using the copilot as well but yeah

cloud aim it might be much more

efficient as I told you their models are

more focused towards the coding and the

development so if you are doing an

enterprise level of coding then this

would be useful right

okay guys so it is going to be generate

Compact compact chat manual uh token

freed

okay compact thinking the user is asking

the API info there great this is done

this is being done okay guys it is

working on that main py then models py

it's working on the workflow py

okay as of now it is reading that

according to that maybe it's going to be

create the API and all everything.

So yes, it is working

models workflow email. Guys, don't leave

this video, okay? Just to check what is

going to be generated by the end. Wait,

[laughter]

we'll test the entire application with

the UI itself.

Now after this one, what I'll do, I'll

keep my open AI key and all everything.

Then only we are we going to test that.

Okay, APA route is created. Now it is

thinking about that first. It has read

all the files and folder. Great.

SRC

in the SRC static it has created static

or not so far.

Okay, here you are full content. Can I

read the specific file

existing schemas and models? Great.

As of now, it has not created a stat

static and the other one. You can create

UI in any other services also in

whatever services you want.

Guys, I'm not going to be stop it. I'm

not going to be cut this part. Let it be

generated. Okay. See how AI is working.

Okay. Summary of what to build. Great.

It is thinking that

I think it is working in a compact mode

that's why it is taking this much of

time to think and the process.

Oh, it is in a plan mode. Oh,

okay. It is showing me a complete plan

right now. Maybe when I switch to the

compact now first it is planning with

the best and possible summary then only

it will do the editing.

Oh now it is saying it I have

everything. Now let me write the

implementation plan. Right

thinking mastering I don't know where

when it is going to be generated next

one

okay maybe

except this cloud Oh yes and auto

accept. Yes and manually approve edits.

No keep planning. Yes and auto accept.

Great. So I accepted all the thing. Now

maybe

it is starting the implementation.

Okay guys. Great. So it has created a

complete checklist and it's going to be

build that one by one.

Mhm.

So this is a complete checklist. I think

by end of this uh session it's going to

be complete all this one. So guys be

careful read everything whatever is

going to be generate and according to

that you need to take your decision.

Okay.

Great. So I got my first file. This is

base.html.

Yeah, it is having a code. Alloy open

JDK.

Okay, it is asking me some permission.

Okay, test email. HTML is also there.

Okay, I think it will take time.

Allow this command. Fine. App loaded

successfully.

Great guys. I think it is about to done

you. Okay. It is checking in a back end

whether everything is working fine or

not.

Okay.

So I am just giving a access. Okay.

Okay. So guys, see it is creating an

environment just for checking the entire

application in a back end. Read it. I

need to think. Oh, files was not

working. It is updating that. Now it

went to the edit automatically. So fine

guys, click to compact now. 45. Great.

Great. Great.

So when I did the compact now guys in

that case it was optimizing everything

means it was generate a planning in a

summarized form. Okay. So I think you

should try over here when you have a

like lower limit then

uh compact right your generation and all

everything and then uh basically see how

this copilot is behaving right

so it is asking for one more command

maybe it is checking in a back end

everything is working fine or not

okay but guys I will test it manually

over here otherwise it will uh

okay this is fine print successfully

from src import loaded successfully. It

is checking everything. If uh something

is going to be like a break then maybe

it will uh look into that code that

part.

So

test app loading and build UV run

import uploaded successfully.

Okay, I think it is trying to check in a

back end what is happening. Everything

is working fine or not. Maybe UV and

Python main. py. Now it is checking with

all the routes in main. py I think we

have all the routes. Oh, it has created

the customer support DB also.

Main. py. This is my all the routes

inside the route only. Yeah, this is the

routes guys. See email routes, UI

routes, everything is there. You see it

has created different different uh

routes over here. We can verify the

entire code at the end but yeah let it

be complete right now.

So customer support DB equal database

file deleted.

Okay.

GRP test

the code has become very huge now guys

and uh

to check manually everything over here.

It will take a lot of effort.

Okay great human review is there. This

is fine

after fixing indexes.

It's running everything over the

terminal and it's checking each and

every module. See, you can read over

here what it is doing.

Dabase initialize. It has created

everything. Okay, this is done, done,

done, done, done. Uh, then open. Great.

It has given me the command also, guys.

See even I can test that now but for

that I required the open AI key maybe

let's read what the what it is saying it

is saying quick start install this all

the dependency then this run this one u

run python main py then open local host

in your browser great test with the UI

okay fine guys so it has it has done

everything okay now let me do one thing

let me basically add the open a key

inside my env. So just a second I'm

doing that

so that I can generate the response okay

if it is required.

So this would be my openi key and I will

have to keep my openi model also over

here. So I kept the key in my notepad.

I'm just going to be copy from there. So

let me copy that. Just a second guys. So

here is the model name as well as the

open AI key. Okay guys, so I kept the

both. Now

I can execute the command to check

whether everything is working fine or

not. So I already installed all the

thing. Now if I'm going to be run this

main py

uh it is saying validation error. Okay,

let me provide this error to this

cloud only. Let's see what it is saying.

So here I am providing. Now see guys, I

kept this edit automatically. That's why

I'm editing. That's why it is editing

automatically everything. Okay.

Open a model. Uh great. Open a model.

Open a API key.

Why it is changing that? Uh

model name line number nine. Uh but the

config open model. Okay. Model. Great.

Got it. Got it guys. I think the name

was different basically.

Uh oh great. I think I just have to

write the model. I just have to keep the

value of that maybe. So that is a issue.

Let it complete then I will keep that.

Okay.

Excellent. The config fix the file app

is running. What I can correct? Uh okay.

Now it has done that. So what I can do

guys I can keep the model only the model

name. I'm not going to be remove

now. I will keep my open API key. I'm

not going to be write the different uh

key name. Okay, I'm just uh editing the

value. Now here I will keep my key.

That's it. Uh let's see whether it's

going to be work or not. So here guys,

what I'll do? I will run this command

again. Okay, now it is running. My

application is running. Let's see

how it is looking like. Okay. So, let me

run it over the local host 127.

Yeah, guys. So, this is the application.

Great. [laughter]

Uh test email inbox is here. Total

pending processing respond failed. Great

guys. It has like really created a great

application. Oh, this is my all the API.

I can test over here. I can check my all

the routes and all everything. That's

really amazing. That's that's really

amazing guys. This is a test email. Even

I can tell the different category

regarding the testing and all. So this

is really amazing. If I'm going to be

test this now submit test email. Let's

see what is happening over the oh so

this is generating now.

Ah great. So what is happening? Oh it is

loading a model.

Which model it is loading? I think it

has done something else in a back end.

uh embedding

H let me check what is the status file

category low confidence zero great guys

so it is giving me a thing now now let

me fix that I think I have to generate

everything from the open itself so

okay let me give the command can you

config the open AI model and open AI

embedding ing OpenAI embedding only I

think it has configured something else

some different model

please correct that in the code okay so

let it be let it create let's see what

is happening now

okay it's doing that maybe

ah it has it was using sentence

transformer that's what I'm thing now

see what is happening I don't know in

the back end I think something was

fishing it was using a different model

now it is doing something config py is

getting changed

great let's see now what will happen

guys uh I hope you like this video if

you liked it then Please hit the like

button and even you can check my YouTube

channel also there I am uploading this

kind of videos and all everything. So

yeah this is going to be run again. Auto

startup completed. This is fixed I

think. So it should be

let it be guys completely. I think much

more enhancement are there. Reading

reading reading sentence transformer is

going to be removed. That's great.

Thinking

perfect. Now clean all the files.

Oh, the file index is going to be

delete.

It is having access of your terminal now

because of that it is able to do like

everything right it has configured the

best terminal in the back end that's why

see this much of code guys within hour

within hour this much of code and the

complete working application test email

this is just a mock email this is an

inbox and here is the API docs now guys

see what we can do we can connect it

with a real email okay with the real

system and yeah I can check like How

many email is being replied? How many is

not being replied? Everything we can do

over here. Everything we can check. So

let it complete. Maybe it is done.

Everything is configured. Again I'm

going to be check my email. So this is

the sender email. The subject and submit

test. Let's see guys what is going to

happen.

Uh it is saying something.

Let me give this error to this cloud A.

Cloud do something now.

Maybe it's going to be break somewhere.

Okay. I think it the key is not going to

be load.

Uh in ENB they have uh just a second

open AI key. Again they have edited this

again they have edited this. Uh that's a

problem. Now guys if we are giving a

full autonomy to the system this is the

issue.

This is the issue guys. This is the

issue.

Here is my key. Now what I'll do guys,

this is my

model. I wanted to use GPT4 model. So

I'm giving a model over here. Oh guys,

I'm just crossing my finger and let's

see whether it's going to be work or

not. Now okay, again I will run it.

Okay, great. So, my host is running. If

I can reload it or I just have to turn

27.

Okay, now guys, I'm just crossing my

finger and running it. Let's see.

It has failed. Now, why it is failed?

Let's see.

No, no response. This time I think I

given the key.

What is the issue?

Email classification failed.

Okay, let's see guys. Again I'm giving

this particular thing to the cloud A.

Let's see now

the config accept turbo and not

I noticed.

Okay. GP4 Turbo preview the it is saying

the model name should be this.

Okay

guys, it is editing this key again and

again. Uh

okay, you should render it immediately.

Formatted the quit around value might

cause parsing issue.

Okay, it has fixed the thing. Now guys,

this time I'm not going to be change

anything except the open key. Okay, I

think we'll have to give the specific uh

like the uh instruction to use the GPD4

model. I think they are using a

different model GP4 turbo preview. Let's

see. Let's execute now. And guys, again

I'm crossing my finger. Let's see.

Everything is going to work. I think the

application is fine. Maybe the key and

all everything is giving giving a issue.

So this time it should work. Failed or

what? So failed. Why? Oh loop tight it's

not able to classify

uh

cloud why what you are doing uh

again let's see

I think the system the NV file is still

the API the export Same one from there's

a major security issue that needs to be

okay. It is saying there needs to be

addition looking at the logs. The app is

running fine. Open a embedding is

successfully but the classification

failing because objective the API key

seems to be causing issue even though it

is in the env. The problem with the is

the old expose key. I need to redirect

it immediately before the user see. But

more importantly notice the log show the

API key is still not working properly.

Uh key is not working properly. I don't

think so. Critical security issue APK

expert again. Okay, let me run it

without this API key. I think so. It has

kept some mock replies and all. We'll

have to check that part guys. Okay, key

part. Maybe my key got exposed and

because of that it is not allowing or

maybe I will have to keep the key again.

So, submit email. Uh it has failed guys.

Why it is failed?

because key is not there. Okay.

Uh

let me ask to it.

You

Okay. So guys, I written this prompt.

Let's see how it is going to be deal

with it. The user is asking a very valid

question. They're confused because I

keep saying APK and redirect

redirecting. But the need real AP is the

system that you have under APK use AP

service. Sorry, they need a get a real

API from the open AI. The need to keep

it safely not in a g report, not shared,

not in a lock. They should a file for

the local developer. There should never

be committed to the version control. You

need the API key, open APK without it.

Open a call. This is normal.

Keep in normal. Okay, your key is here.

This correct perfect. Okay. Get env

commit it don't push it when I keep

warning about exposure then issue is in

a chat history or accidentally past the

actual key this is expo for testing only

just a new one if public github you are

safe it was user production

okay great

I'm writing over here I am using it for

the testing only and it is in my local

so Please accept

my key

and I will not keep it on a edit mode

and I'll keep it in this

uh

I will keep it over the ask before edit

mode. So open a API key. Now let me keep

this thing over here. system is ready

and let's see now guys if I am going to

be run it

okay my workflow is initialized

my application should be ready and if

I'm going to submit

it has failed again

no response spend for 6 No response

available.

Oh, something is happening in between.

Maybe it is a issue of the LLM only.

Uh

looking like the logs I can see. So

clearly email is successful.

Classification attempt what fail because

the email defaults to category other