Download "LangGraph:21 End-to-End Supervisor Multi-Agentic AI Project for Booking Doctors Appointment #aiagent"
%3Aformat(webp)%2Fi.ytimg.com%252Fvi%252F3ZauAhbTyBE%252Fmaxresdefault.jpg&w=1139&q=75)
Similar videos from our catalog
%3Aformat(webp)%2Fi.ytimg.com%252Fvi_webp%252F1Ff0G0Z6_nE%252Fmaxresdefault.webp&w=1139&q=75)
%3Aformat(webp)%2Fi.ytimg.com%252Fvi%252F1nleJxDdaj8%252Fmaxresdefault.jpg&w=1139&q=75)
%3Aformat(webp)%2Fi.ytimg.com%252Fvi%252FCRraHg4Ks_g%252Fmaxresdefault.jpg&w=1139&q=75)
%3Aformat(webp)%2Fi.ytimg.com%252Fvi_webp%252FIpOiRmAEle8%252Fmaxresdefault.webp&w=1139&q=75)
%3Aformat(webp)%2Fi.ytimg.com%252Fvi%252FN-HFI7kPIAQ%252Fmaxresdefault.jpg&w=1139&q=75)
%3Aformat(webp)%2Fi.ytimg.com%252Fvi_webp%252FPtETUYa3i2Q%252Fmaxresdefault.webp&w=1139&q=75)
%3Aformat(webp)%2Fi.ytimg.com%252Fvi%252FjRQaur0LdLE%252Fmaxresdefault.jpg&w=1139&q=75)
%3Aformat(webp)%2Fi.ytimg.com%252Fvi%252Fu1NMxMqoqtY%252Fmaxresdefault.jpg&w=1139&q=75)
Video tags
Subtitles
00:00:01
Hey. Hi everyone. Welcome back to my00:00:03
YouTube channel. My name is Sunonny00:00:05
Savvita and I'm back with another00:00:07
exciting and important video. So guys,00:00:10
in this particular video, we're going to00:00:12
implement one end to end project using00:00:15
multi- aenting system. Yes guys, so in00:00:18
my previous videos I explain you about00:00:20
the multi-agenting system. I discussed00:00:23
different different kind of00:00:24
multi-agentric system along with a00:00:26
complete theory. Now it's time to00:00:29
implement some project. So guys uh in00:00:31
this video I will show you how you can00:00:33
implement a project using supervisor00:00:37
multi- aentric system and this project00:00:40
is going to be complete end to end means00:00:43
I'm not going to write a code inside the00:00:45
IPV file instead of that I'll be writing00:00:48
complete modular coding and you will get00:00:50
a idea if we are getting any sort of a00:00:53
project related to the agentic AI or00:00:56
multi- aent system how we can code that00:00:59
end to end. So let's begin. Let's start00:01:02
without further delay. So uh here guys00:01:05
you can see the complete syllabus of00:01:08
this uh langraph course. Uh this entire00:01:11
syllabus I covered inside my course. Uh00:01:14
you can check out the playlist it is00:01:16
there over the YouTube. And now guys we00:01:19
are about to complete this course. So we00:01:21
are here at this particular point where00:01:23
we are going to understand the different00:01:26
different voltage index system along00:01:28
with a real world projects right so uh00:01:31
two to three more video I'll be00:01:33
uploading with respect to this course00:01:35
and then I'll be starting with a new00:01:37
playlist soon I will announce that a00:01:39
particular playlist so if you haven't00:01:42
seen my YouTube channel so far if you00:01:43
haven't subscribed guys please go and00:01:45
check please subscribe the channel hit00:01:48
the like button if you are liking the00:01:50
content, liking the video and leave your00:01:53
thoughts inside the comment section.00:01:55
Now, apart from that guys, you can00:01:57
connect with me over my LinkedIn uh and00:02:00
you can follow to me over the uh GitHub.00:02:03
There also I'm uploading uh my code base00:02:06
and all everything. Okay. And this00:02:08
entire code base of this particular uh00:02:10
course right this langraph course you00:02:12
will find out over my GitHub itself.00:02:13
Here you can see this langraph end to00:02:15
end. Now apart from this one if you have00:02:17
any any query right for me so uh you can00:02:20
click on this onetoone call and uh you00:02:23
can connect with me onetoone for sorting00:02:25
out any sort of a query whether it is00:02:28
regarding to the resumeumé career00:02:29
guidance interview preparation mock00:02:31
interview genative interview or if you00:02:33
want to do quick chat with me I'm00:02:36
available uh like for the given slot00:02:39
right so please uh check out this page00:02:41
and uh based on the like query okay you00:02:45
can running with me now. So, uh guys,00:02:48
first of all, let me clarify the00:02:49
complete problem statement. So, here I00:02:52
written the complete problem statement00:02:54
uh on my blackboard. Okay, step by step,00:02:56
we'll try to understand uh each and00:02:59
everything, right? And uh from in front00:03:01
of you only I'm going to code the entire00:03:03
project, right? A couple of pieces I00:03:05
kept in my system that I'll be copy and00:03:08
pasting at the specific place uh inside00:03:10
the project directory, right? But most00:03:12
of the thing I am going to write in00:03:14
front of you only. Now here guys, so uh00:03:16
first of all let's try to understand the00:03:19
problem statement. Uh I written in very00:03:21
uh clearcut way easily you can00:03:23
understand it. See we are going to build00:03:25
a doctor appointment system using00:03:27
langraph fast API and streamllet. So00:03:30
here guys we are going to use the lang00:03:32
graph for creating our agentic flow.00:03:34
We're going to use this fast API for00:03:36
creating our endpoint and this00:03:37
streamllet for the UI. Okay. Now you can00:03:40
read out the problem statement right.00:03:42
What is my problem statement? So this00:03:44
project is a multi-agentic system. Uh in00:03:47
multi-agentric system where which00:03:49
architecture it is following. So it is00:03:51
following a supervisor architecture00:03:53
right supervisor multi-agentic00:03:55
architecture. I'll show you in some time00:03:57
I cap the architecture as well and here00:03:59
you can read the further description. So00:04:01
this is a project this is a multi- aent00:04:03
project AI power doctor appointment00:04:04
booking system. Okay. So that's similar00:04:07
to realistic medical assistant capable00:04:09
of handling user query regarding doctor00:04:11
availability, specialization,00:04:13
appointment booking and all. So guys00:04:15
this is a a doctor appointment booking00:04:17
system which I developed using the00:04:19
multi- uh agent uh supervised multi-00:04:22
aenting system. How does it work? Okay,00:04:24
what are the what all functionality I00:04:26
have incorporate over here. Each and00:04:27
everything we are are trying to discuss.00:04:30
Right now here I'm talking about the00:04:32
tech ST right? What is the tech ST for00:04:34
this particular project? So we're going00:04:36
to use the langraph for the workflow00:04:37
automation right for creating the agents00:04:40
and all. Then we're going to use the00:04:41
langen for modeling model loading prop00:04:43
engineering okay for the tool creation00:04:45
and all. Then we're going to use the00:04:47
fast API here right for serving a API00:04:49
endpoint and all. Then we're going to00:04:51
use the streamllet for front end00:04:53
creation and then we're going to use the00:04:55
python panda CSV for the data headlight.00:04:58
So here I have I incorporate the data as00:05:00
well. I kept the data inside the CSV uh00:05:03
that that particular data I can leverage00:05:05
okay for building this particular00:05:07
project. I'll show you how the data00:05:08
looks like and what all components we00:05:10
have inside the data. Right now apart00:05:12
from this one guys here you can see uh00:05:14
so this is the sample use case. So how00:05:17
this application is going to work. So00:05:19
let's say if you are going to ask can00:05:21
you check if a dermatologist is00:05:23
available on 5th of August at 10:00 a.m.00:05:26
So according to that it is uh replying00:05:28
to me. Okay. So it is checking inside00:05:30
the database and all based on the tool00:05:32
calling based on that agented flow and00:05:34
it will provide me a answer right now. I00:05:37
hope this problem statement this tag is00:05:39
tag okay and this simple use case uh it00:05:42
is clear now coming to the architecture00:05:44
of this project. So guys this is a00:05:46
multi- this is a multi- aent00:05:48
architecture uh it's a supervisor00:05:50
multi-agent architecture. So uh over00:05:53
here what I'm doing please try to focus00:05:55
now you will get the clearcut idea.00:05:57
First we are passing the input. Okay,00:05:59
this input is going to the supervisor00:06:00
agent. Now based on the query it is00:06:03
going to be delegate, right? Let's say00:06:04
if I if I just want the information so I00:06:07
can pass my query to the information00:06:09
node. Let's say along with the00:06:10
information I want I I want the booking00:06:13
node. I want the booking also. So after00:06:15
collecting a information okay it is00:06:18
passing again cursor to the supervisor00:06:20
and then it is booking a slot for me.00:06:22
Let's say query does not belong to any00:06:24
information uh uh information node or to00:06:27
any booking node. In that case directly00:06:29
it's going to be finish the query. Okay,00:06:31
that mechanism is also there. So in00:06:33
short I can say it is a complete00:06:35
multi-agentic AI system supervisor00:06:37
multi-agent AI system where whatever00:06:40
query we are getting based on the query00:06:42
it is going to be delegate the task.00:06:43
First it is trying to retrieve the00:06:45
information from the given database and00:06:47
then what it is doing according to my uh00:06:49
convenient time or according to my date00:06:51
it's going to book the slot for me okay00:06:53
for the doctor I hope uh this entire00:06:56
thing is clear now guys so uh you you00:06:59
can uh take a pause or you can you can00:07:02
take a screenshot and all of this entire00:07:05
thing right because if you're going to00:07:06
mention this type of project in your00:07:08
resume so this entire information would00:07:10
be required right so you can take it up00:07:12
or else I can give you the P as a PDF uh00:07:14
in the description. You can download00:07:16
from there itself. Now guys, uh let's00:07:19
try to uh start with the implementation.00:07:21
Let's see how we can implement this00:07:23
entire project. How we can implement00:07:25
this entire use case guys. So here guys,00:07:28
I will show you from very scratch and00:07:31
what I'm doing. Uh first I'm going to00:07:34
create a folder and uh inside that00:07:36
folder I will open my VS code and then I00:07:38
will create my environment and all the00:07:40
complete folder structure. Okay. is00:07:43
saying enter character to delete up to00:07:45
right. Now here what I am doing first00:07:47
I'm going to create one directory uh00:07:49
inside this location see uh you can see00:07:52
my location over here see user sunny00:07:55
okay it could be any location in your00:07:57
system so I'm writing mktir means make00:07:59
directory and my project name is what00:08:03
doctor00:08:08
appointment multi- aent okay so I00:08:12
created this folder now I'm going inside00:08:14
this folder doctor00:08:17
appointment multi- aent00:08:22
okay with it now what I'm doing here I'm00:08:25
going to write this code dot so that I00:08:27
can open my VS code inside this location00:08:30
right so uh here I have opened this here00:08:33
I open the VS code inside the particular00:08:35
location now what I can do I can close00:08:36
the CM is not anymore required uh so00:08:40
first of all what I will do over here uh00:08:42
inside this VS code I'll be creating my00:08:44
environment ment. Okay. So for creating00:08:47
an environment guys, uh I I am going to00:08:49
open my terminal and over the00:08:52
terminal. So I'm not going to use the00:08:54
power cell here. I'm going to use the g00:08:56
bash. So here is my git bash and let me00:08:59
open the g00:09:00
bash. For the git bash actually you can00:09:03
uh execute the linux command. So my00:09:05
command is going to I'm going to use the00:09:07
cond for creating a virtual environment.00:09:09
You can use anything okay python or uv00:09:12
sorry not uv you can use directly python00:09:14
as well okay or whatever uh this00:09:17
environment manager you are using can00:09:20
use that so here conduct00:09:25
create00:09:27
penv and python which python version I'm00:09:30
going to I'm going to use python 3.900:09:33
here and then00:09:35
python guys uh you can see my00:09:37
environment is getting created and along00:09:39
with the environment what I will do I'm00:09:41
going to be initialize the uh g as well00:09:45
okay so I can give you this entire code00:09:47
through the github only so that you can00:09:49
practice so first I'm going to create00:09:51
this g ignore and inside I mentioned00:09:53
this v n okay now here I'm going to00:09:56
create one more file the file name is00:09:58
going to be requirement re00:10:01
ui m ns txt okay uh then guys uh I'm00:10:06
going to create one more file over here00:10:08
the file file name is00:10:10
setup. Okay. Uh now here I think so this00:10:14
is clear this all the files and folder.00:10:16
So let me do one thing. Let me00:10:18
initialize the gate for so for00:10:20
initializing the gate this repository.00:10:23
Okay. As of now it is my normal00:10:24
repository. Once I will initialize the g00:10:27
so it will become a local repos local00:10:29
get repository. I'm writing over here00:10:31
get in it and see uh my g has00:10:33
initialized. Uh now you can see it is in00:10:36
track. Okay. This file is being tracked00:10:38
by the G it okay under track.00:10:40
Unttracked. Okay. See here is it is also00:10:43
telling me unttracked. So once I will00:10:44
add it so you will find out this00:10:46
unttracked is00:10:47
going is like tracking that file and00:10:51
then I can commit that. I think you know00:10:53
the basic concept of the git. If you00:10:55
don't know I will create a separate00:10:56
video on it. Now uh here guys I think.00:10:59
So this is clear. Now let me create00:11:01
couple of more folder. So I'm going to00:11:02
create one more folder where I will be00:11:04
keeping my data. Okay. This data would00:11:07
be available in the form of CSV. But uh00:11:09
this data could be anything. It might00:11:11
come from anywhere from any database.00:11:13
I'm just keeping in the form of CSV.00:11:15
Okay. Now I'm going to create one more00:11:17
folder here. So my folder name is going00:11:19
data model. So inside this data models00:11:21
I'm going to keep all the pentic models.00:11:24
Okay. Then uh here is my one more00:11:26
folder. The folder name is notebook. So00:11:28
first I can do the experiment inside the00:11:30
notebook and then I can keep the code in00:11:32
a model. Right. uh then guys here I'm00:11:35
going to create one more folder the00:11:36
folder name is going to be prompt00:11:38
library so I'm not going to keep my00:11:40
prompt okay at a single place whatever00:11:43
prompt are coming so I will be keeping00:11:45
those particular prompt inside this00:11:47
prompt library is nothing it's a py file00:11:50
only but it could be any file in your00:11:52
case it could be a JSON file it could00:11:54
file in whatever file you want to00:11:55
configure your prompt you can do right00:11:58
now we have one more folder here so the00:12:00
folder name is going to be toolkit okay00:12:02
so inside this tool kit Guys, I'm going00:12:04
to keep my all the tools whatever tools00:12:07
is required for creating this00:12:09
multi-agent. Then I'm going to create a00:12:11
one I'm going to create one more folder.00:12:13
The folder name is going to be utils. So00:12:14
inside this utils folder I will keep all00:12:17
the utility function. Okay. Uh now guys00:12:20
uh there is couple of more file which is00:12:22
required. So here I'm going to create00:12:24
this agents py. Okay. So this agents py00:12:28
uh it will keep the main agent. Okay. it00:12:30
will keeping all the agents react agent00:12:32
and all whatever we are okay then I'm00:12:35
going to create one more file over here00:12:37
the file name is going to be main py so00:12:39
inside this main py I will define my all00:12:41
the endpoints okay then here I'm going00:12:43
to create one more file my file name is00:12:47
goingore ui so I will keep my uh I will00:12:51
keep my UI okay UI code inside00:12:55
this_i file I hope guys I have created00:12:58
each and every folder now I will keep00:13:00
the code inside this all the folder.00:13:02
Okay. So let's see step by step one by00:13:05
one what all type of code what all type00:13:07
of file files we have files basically00:13:10
okay we we required inside this00:13:11
particular folder and how to write our00:13:13
agents right and what all requirements00:13:16
are there for this particular project.00:13:18
Got00:13:20
it? Okay. So guys uh here I added all00:13:24
the requirements whatever is required00:13:27
for this specific projects. Uh before00:13:30
recording this video I was executing00:13:32
this project in my uh different folder.00:13:35
Okay in my different uh uh VS code and00:13:39
uh there actually I created an00:13:40
environment. I take I took all this00:13:42
requirement from there itself. So uh00:13:45
better if you can download if you can00:13:47
install this all the requirements so00:13:48
that you won't be you you won't face any00:13:51
such issue with respect to the uh00:13:53
requirements. Okay. Now let me highlight00:13:56
what all thing is required. So here guys00:13:59
uh this is required. Uh which one00:14:03
uh fast API is required. Then if you are00:14:06
executing uh if you are using the Google00:14:08
model then this Google libraries will00:14:10
required. If you are going to use the00:14:13
grog model then grock would be required.00:14:15
Okay. Now IPv kernel will required it is00:14:18
a by default. If you are going to use00:14:19
the if you're going to run your code00:14:21
inside the uh inside we inside your IPv00:14:25
file. Then apart from this one there is00:14:27
a complete ecosystem for the langen00:14:29
langen core00:14:31
experiment langen google generative ai00:14:33
gro hugging phase this all the libraries00:14:35
required. Okay. Uh then langraph would00:14:38
be required. Okay. Now apart from that00:14:41
you can see numpy would be required.00:14:42
Numpy pandas would be required. Okay00:14:44
here see I'll install this numpy and00:14:46
pandas. Then uh after that guys ubicon00:14:49
will be required because we're going to00:14:50
use the fast api. So we'll have to run00:14:52
the server using this ubicon command00:14:55
right. And then this stimulate langen00:14:59
community is required. The hyphen e dot00:15:01
is for what this hyphen e dot actually00:15:03
it is for installing the local packages.00:15:06
local packages means so let's say here I00:15:08
created this toolkit okay if I want to00:15:10
convert this toolkit into my local00:15:12
package so I will create one file over00:15:14
here unders00:15:17
init_init okay so if I have that file00:15:21
okay if I have that file uh inside any00:15:24
folder so that a particular folder will00:15:26
be installed inside this virtual00:15:28
environment okay what I'm what I'm uh00:15:31
see what I'm trying to explain you over00:15:33
here I'm just trying to explain you that00:15:35
how you can install So the local00:15:36
package. Okay. So if you are going to00:15:38
initiate uh this package as a local00:15:41
package, so you will have to define00:15:44
the_init file over here. Okay. And then00:15:47
you can install this particular folder00:15:50
inside your current virtual environment.00:15:52
So whenever you are going to be00:15:54
imported, whenever you are doing any00:15:56
crossimporting or any anything right,00:15:58
you won't face any such issue. Module00:15:59
issue or this uh module is not there,00:16:02
this module is not getting detected.00:16:04
That's why we generally follow this00:16:05
step. Now uh apart from this one guys, I00:16:08
have created a couple of more project.00:16:10
So if you will go and check with my do.00:16:13
So there only you will find out some end00:16:15
to end project. Okay. Long back I uh00:16:17
released one project with respect to00:16:19
this Google dialog flow e-commerce00:16:21
chatbot with deployment. Okay. So there00:16:23
also I have explained all these00:16:24
particular steps. Uh if you are00:16:27
following to me regularly then I think00:16:29
this step is nothing for you. So guys uh00:16:32
yeah if you are liking this content if00:16:34
this content is helping to can uh00:16:36
subscribe the channel okay because I'm00:16:39
trying to give you very premium content00:16:41
which you can uh directly utilize okay00:16:44
which will directly help you uh for00:16:46
landing your job right for getting your00:16:47
first job into this generative AI. Now I00:16:51
think requirement.txt is clear. Now what00:16:53
we'll have to do guys what would be the00:16:55
next step? So after creating the00:16:56
environment, we'll have to activate this00:16:58
environment and then we'll have to00:16:59
install this require.txe file. So no00:17:02
need to worry about it because I I00:17:04
already written all this command inside00:17:06
this read file. I will provide to all of00:17:08
you. You can take this entire commands00:17:10
from the GitHub directly, right? I00:17:12
already perform it just to save my time00:17:14
so that directly I can execute my code.00:17:16
Okay. So what you have to do guys? So00:17:18
first you have to create environment00:17:19
then activate the environment. Then00:17:21
mention the requirements over here.00:17:23
Okay. inside the recon.txt file and00:17:26
after that you have to install this00:17:28
requirement pip with this pip install00:17:30
recon.txt txt. One more thing you have00:17:32
to make sure over here. So inside the00:17:34
setup py file, this code should be00:17:36
present. Okay. So inside this code00:17:38
actually the setup function is the00:17:40
important one where you are going to be00:17:42
uh define your local package. What is00:17:44
the meaning of it? The meaning is very00:17:46
simple. Uh like this uh folder right I00:17:49
was telling to you know wherever you are00:17:50
going to be uh mention the init file. So00:17:52
those will be those will be like00:17:55
downloaded as a local package inside00:17:57
this particular environment. What would00:17:58
be the name for it? name for that00:18:00
particular package would be a doctor00:18:02
appointment agent. Okay. So in short uh00:18:05
you are going to be install the local00:18:07
package inside where inside your virtual00:18:09
environment and this would be a name.00:18:11
Here is a version here is my name here00:18:14
is a mail id. Find packages means uh00:18:16
like it is finding out the package means00:18:18
wherever the init file is available is00:18:20
downloading only those folder. Okay with00:18:22
under this particular name and uh here00:18:25
is what called requirement. So yes, I00:18:28
can install this requirement also while00:18:30
I'm downloading the packages. Right? So00:18:32
it won't give me any such issues. And00:18:34
here is what at the last you will find00:18:36
out the Python requirement. So at least00:18:39
3.10 is required for this particular00:18:41
project. Okay. So this is the basic uh00:18:43
setup for executing the project. Uh00:18:46
still we haven't uh started our main00:18:48
coding and all but this was very much00:18:50
required for running the project. Okay.00:18:52
So I hope this entire thing is clear.00:18:54
Now let's come back to the data. So guys00:18:57
uh here I kept the data inside the CSV00:18:59
file. As you can see the data. Okay. So00:19:01
inside this data I have around 4,00000:19:03
rows. If I will scroll down now. So here00:19:06
you will get the entire post. So in00:19:08
total how many rows are there? So in00:19:11
total guys 4,000 rows are there?00:19:14
4,281. Now what it what this data about00:19:18
right? So to understand about this data00:19:20
you will have to read the column of okay00:19:23
column name of this data. So here I have00:19:26
this data slot then specialization that00:19:28
doctor name then is available or here00:19:30
you can see the patient attend. Okay. So00:19:33
uh here you can see the date and the00:19:36
time then uh here you can see the00:19:39
specialization this doctor this John Doy00:19:41
okay he is expert in the general dentist00:19:44
he's expert of the general dentist. Then00:19:46
is available true? Okay if it if is he00:19:48
available that case it it like this00:19:51
doctor don't have any patient to attend.00:19:54
Okay, it is not having any patient ID.00:19:56
Now just uh let's look into the next00:19:58
data point. So here is a date slot.00:20:01
Okay, here you can see general dentist00:20:03
for this John is available false. And00:20:06
here is a patient ID. So this John Day00:20:09
which is a like expert in this general00:20:11
dentist. He's going to be attended to00:20:13
this particular patient. Okay, at this00:20:15
particular date and time, right? So you00:20:17
can see the availability of this John00:20:19
Day at which date it is available. Right00:20:22
now, apart from this one, we have other00:20:24
doctors as well. Okay, in the different00:20:26
different other expertise and00:20:27
specialization. Let me uh show you uh00:20:30
those doctors. So, there is one more00:20:32
doctor Emily Jensen. Okay, he she is a00:20:35
expert in general dentist again. Okay.00:20:38
So, she is not available at this00:20:40
particular date and time. And here is00:20:42
the ID of the patient which is going to00:20:44
be attended by this John Emily Johnson.00:20:47
Right? Now, uh see the other entries.00:20:49
Let's say if I want to check with the00:20:51
Emily Johnson itself. So Emily Johnson00:20:54
like who is a general dentist she is not00:20:57
she is a available at this particular00:20:59
time. Why? Because is inside this is00:21:01
available we have mentioned true and00:21:03
there is no patient to attend. Okay. Uh00:21:05
this Emily Johnson is not going to be00:21:07
assign any sort of a patient at this00:21:09
particular time. Okay. Date and time. I00:21:11
hope you understood this particular00:21:12
data. Now how we are going to be use00:21:14
this particular data. Okay. for our00:21:16
requirement for our multi creating a00:21:18
multi- agent for you will get in some00:21:20
time but guys this data could be00:21:22
anything this data could available00:21:23
anywhere inside your case let's say this00:21:26
data is available in in it is00:21:28
available in some uh private databases00:21:30
in your in-house databases okay let's00:21:32
say it is available it over any server00:21:35
any API okay wherever guys I'm just00:21:37
assuming this data is available anywhere00:21:39
in real time but in my case this data is00:21:42
there inside this data folder I hope00:21:44
this thing00:21:45
Now coming to the rest of the file. So00:21:47
guys, now here I'm going to create a00:21:49
rest of the file. So inside this data00:21:51
models, I will keep my all the pinetic00:21:54
models. So the file name which I'm going00:21:55
to create inside this data models, the00:21:58
file name is going to be models py.00:22:02
Okay. Uh now inside this notebook guys,00:22:04
I will keep my uh IPv file so that I can00:22:07
do the experiment. So here I can write00:22:10
multi-00:22:14
agent dot ipy. Okay. Now uh then I have00:22:19
next file prompt dot prompt library00:22:22
prompt library. Okay. This is the00:22:23
folder. Now inside this particular00:22:25
folder I'm going to create a file. The00:22:27
file name is prompts. Okay.00:22:30
Prompt. Then what I will do? So inside00:22:32
the toolkit guys again uh I required one00:22:35
file. So inside the toolkit guys I'm00:22:37
going to create a file. The file name is00:22:39
going00:22:40
toolkits.p okay00:22:43
toolkits py. I will keep my all the00:22:45
tools inside this particular file. Then00:22:47
inside the utils guys I required one00:22:49
file. So uh for loading the llm and all.00:22:52
So here I'm going to write llm py. Okay.00:22:55
So this is one more file. Now agents is00:22:58
there main.py is there. Rest of the file00:23:00
is here. Okay. So I think no need to00:23:02
worry about. I hope this entire thing is00:23:06
clear to all of you. Now coming to the00:23:08
next part. So agents py is here. Did I00:23:12
do any mistake? Okay, it is a folder00:23:14
guys. Not it should not be a folder00:23:16
actually. It should be a file only. Let00:23:18
me give00:23:21
file00:23:24
ag. Okay. So my file name is what?00:23:27
Agents py or it should be agent00:23:33
only. Perfect guys. So I hope this all00:23:36
the file and folder everything is clear.00:23:40
It is not very much difficult. Now let00:23:42
me define let me write the init file as00:23:44
well. So here what I can do I can me the00:23:47
init00:23:49
file 4 py. Now inside this toolkit also00:23:54
I'm going to mention this00:23:58
4 py. Okay then the utils. Okay. So00:24:10
unus py so whatever folder is there00:24:13
wherever you have defined the py file or00:24:16
in inside all the folder actually you00:24:18
can00:24:21
create_init p okay so see inside the00:24:24
data it is not required inside a data00:24:26
model it is required I created then uh00:24:29
inside notebook it is not required00:24:31
prompt library yes required Toolkit00:24:34
required, utils required. Okay. And rest00:24:36
of the file is a pile a py file itself00:24:39
at this workspace. Right? This is my00:24:41
whole sort of a project structure for00:24:44
doing the for creating a like end to end00:24:47
flow. Okay. End to end multi- agentic00:24:49
flow. Uh okay. So this particular folder00:24:51
structure it could be present or it00:24:53
could be a part of any sort of00:24:54
application. Let's say one multigenic00:24:57
system you are going to be append inside00:24:59
your website or one multigenic system00:25:01
you are going to be append inside your00:25:03
app and running software. So you can00:25:05
create a folder structure like this you00:25:07
can do all the modular coding and all00:25:09
and you can expose your endpoint from00:25:11
here okay from this man. py and then you00:25:14
can make a connectivity but again I'm00:25:16
saying guys this could be a in little00:25:18
more detailed way. So in the upcoming00:25:20
session uh definitely I'll try to00:25:22
discuss in that manner as well. But yeah00:25:24
this is enough for creating any end to00:25:27
end flow. Okay all this f files and00:25:30
folder you can add on more file and00:25:32
folder according to your like00:25:34
understanding. Okay some yaml file00:25:37
config file and all. I hope till here00:25:39
everything is fine. Now let's do one00:25:41
thing. Let's try to start a coding00:25:45
guys. Okay. So guys first let's start00:25:48
with the data models. So if I'm saying00:25:51
data model that mean uh I'm talking00:25:53
about the parentic model model for the00:25:56
validation classes for the validation.00:25:59
So here I created one file inside this00:26:01
data_m models. So the file name is00:26:04
models py. Let's try to write a code00:26:06
inside this models py. So first of all00:26:10
we'll have to import this regular00:26:12
expression. I'll show you why we are00:26:14
doing it. Why we are importing this00:26:15
regular expression. Then uh from pentic00:26:19
we are going to import a base model. So00:26:23
let me import this base model and using00:26:26
this particular class we can convert our00:26:28
class into the pyic uh model itself.00:26:31
Then the next one is a field just to00:26:33
showcase the description just to write a00:26:34
description of a given parameter. Uh00:26:37
then we have a next one which is called00:26:39
field validator. So let me write this00:26:41
field validator. It is very amazing uh00:26:44
function which I can use as a decorator.00:26:46
I'll show you what is the use of it. So00:26:48
here is my field validator. Now, so let00:26:51
me create a classes. Uh the first class00:26:54
is going to be datetime model. So I'm00:26:58
going to create a a class. The class00:27:00
name is date time model. So what it will00:27:02
do guys, it will check the format of the00:27:04
date and time. Uh it will check whether00:27:07
whatever date and time we are getting00:27:09
from the user end or whatever date and00:27:12
time we are getting after generating00:27:13
from the tools. Okay. from the model it00:27:16
is correct or not. So here I'm going to00:27:18
inherit my base model. So let me inherit00:27:20
my base model and inside this one I'm00:27:23
going to write my parameter which is00:27:26
date and the type of this date is str.00:27:28
Now here I will mention the field. So00:27:30
field it will keep the detail of this00:27:32
date parameter and let me mention the00:27:36
detail of this field. So here guys I'm00:27:38
going to mention the description. Now00:27:40
along with the description I'm going to00:27:42
mention the uh pattern also. So here you00:27:45
can see the complete code. So the00:27:48
description is properly formatted date00:27:50
and the pattern would be like this. So00:27:52
what I'm saying here what I what I try00:27:55
to return in the pattern. So D in the00:27:58
curly presses we have this two. So it is00:28:00
saying that the date will be up to two00:28:02
digit month up to two digit and here the00:28:05
year up to four digit and here is my00:28:08
hour and this is the minute. Okay. So it00:28:10
is representing a time. So if you will00:28:12
look into the data guys, so here is a00:28:15
data dr availability csv. Just look into00:28:18
this date slot column date slot. You00:28:22
will get a clearcut idea that why we are00:28:24
writing that type of pattern. So uh here00:28:26
is my date is day mm and here is my year00:28:30
and you can see the time. Okay. So just00:28:32
to validate this date slot we are going00:28:34
to write this date time model. Now uh00:28:37
let me write one more thing over here00:28:39
that's going to be a field validator. So00:28:41
here is a code for the field validator.00:28:43
Let me do one thing. Let me copy and00:28:45
paste and uh see guys this is the code.00:28:48
So what we have here inside the field00:28:50
validator we are passing a date. Okay00:28:52
this is just one more level of00:28:54
validation inside this datetime model.00:28:57
Right. So whenever we are getting a00:28:58
date. So here itself inside this00:29:00
particular class we can validate whether00:29:02
it is correct or not whether it is00:29:04
following this particular pattern or00:29:06
not. So here we are passing a date. So00:29:08
and to this check format date we have to00:29:10
pass the cls reference along with this00:29:12
v. V is nothing it is a date itself.00:29:14
Then we are checking whether this v00:29:17
means my date is following this00:29:18
particular pattern. If it is not00:29:19
following in that case I'm going to00:29:21
raising this value error. Otherwise I'm00:29:23
going to return this date as it is. Now00:29:25
apart from this one I have I have two00:29:27
more date models. So let me uh take00:29:29
those model as well. One is for date.00:29:32
Okay, only for the date excluding the00:29:34
time and the next one is for the00:29:36
identification number. So guys, this is00:29:38
only for the date. I think you can00:29:40
clearly understand this date model. We00:29:42
don't have time over here. If someone00:29:43
giving a date only, so we can validate00:29:46
this uh we can validate the date using00:29:48
this particular class. Okay. Now the00:29:51
second one is what identification00:29:52
number. So if you will look into the00:29:54
data, so the last column is a patient to00:29:56
ID which is having one identification00:29:58
number which is having one ID number.00:30:00
Just count the numbers how many digits00:30:02
we have over here. So 1 2 3 4 5 6 700:30:06
digit we have. So just look into this00:30:08
model py. So here we are saying if it is00:30:11
not going to be matched means if inside00:30:13
the ID we don't have 7 to 8 digit. In00:30:15
that case I will raise this value error.00:30:17
The ID number should be seven or 8digit00:30:19
number otherwise we are going to be00:30:21
return is as it is. Okay. So one more00:30:23
level of validation we are going to do00:30:25
over here. So this is my simple uh data00:30:27
model. I hope you understood this data00:30:29
model and all. Don't worry guys, I'll00:30:31
show you the output of it because I00:30:32
created one notebook over here. Inside00:30:34
this I kept the entire code see over00:30:36
here and one by one step by step we'll00:30:38
try to execute this entire one. But00:30:40
first of all let me place the code at a00:30:42
appropriate folder at appropriate file00:30:44
so that later on we can execute it00:30:46
directly. Now the next file we have00:30:48
which is a prompt or py. So uh let me00:30:52
show you what actually I'm going to keep00:30:53
in my prompt. py file. So here I'll be00:30:56
keeping the system prompt. Okay, I'm00:30:58
going to copy and paste this particular00:30:59
prompt. See here uh is a prompt and00:31:01
along with the prompt I'm giving some00:31:02
additional information. So try to read00:31:05
out this particular information what I00:31:06
have mentioned over here. Uh so first of00:31:08
all let me do the word rap for you so00:31:10
that you can easily read out the00:31:11
information. So the first note this is00:31:14
my first note first agent. It is a00:31:15
information agent. So it is a detail of00:31:17
the information agent what it is doing.00:31:19
Then this is my second agent which is a00:31:21
booking agent. Okay. And here is a00:31:22
detail of this particular agent. So00:31:24
information agent is just to fetch out00:31:26
the information from the given data okay00:31:28
database and booking node it is for the00:31:31
booking right now here guys you can see00:31:34
we have a option so how many options we00:31:35
have we have a two option one is00:31:37
information node second is a working00:31:38
node and uh here third one will be a00:31:40
finish right so if you will look into00:31:42
the architecture which I shown you I00:31:45
think in my uh previous uh okay just a00:31:48
second guys let me show you that00:31:50
architecture00:31:52
uh give me a moment00:31:56
Here you can see the architecture guys.00:31:59
So inside this architecture we have two00:32:01
agent. One is a information agent.00:32:02
Second is a booking agent and third is a00:32:05
end. Okay. So the finish finish which I00:32:08
shown you over here. This finish00:32:09
actually it is representing to the end00:32:12
uh end node itself. And here what I'm00:32:14
doing I'm compiling the complete00:32:16
information inside this particular00:32:18
variable working info and I'm appending00:32:20
inside this system prompt. So you you00:32:22
can read out the system prompt. Uh you00:32:24
are a supervisor tasked with a managing00:32:26
a conversation between following worker.00:32:28
So here is a complete information of00:32:30
this particular agents. Then your00:32:32
primary role to help user make an00:32:34
appointment with a doctor and provided00:32:35
FAQ. Okay. If customer requested to know00:32:38
availability of the doctor or to book00:32:40
reschedule or cancel appointment you00:32:42
have you have to be delegate the task00:32:44
according to that. Okay. Respond with00:32:46
the worker act. Next, each note each00:32:48
worker will be perform a task and00:32:51
respond with their result and status00:32:53
when finished. Okay, response response00:32:56
respond with finish. Okay, whenever the00:32:57
task is going to be finished. So here00:32:59
you can see this is a complete prompt.00:33:01
You can pause the video and you can read00:33:03
out this complete prompt. Now coming to00:33:05
the next part. So here after the prompt00:33:07
guys, the main thing which is coming00:33:09
which is a toolkit. So we have to write00:33:11
a toolkit. Okay, inside this toolkit I'm00:33:13
going to define my four to five tools.00:33:15
Now this tool00:33:16
basically it is interacting with the LLM00:33:20
okay and it is providing a information00:33:22
based on the logic. So inside the tool00:33:24
guys see how many tools we have. First00:33:26
of all let's take a walk through of it.00:33:28
So we have this we have this tool class00:33:30
okay uh this we are getting from that00:33:32
len itself. So we are using as a00:33:34
decorator on top of the function and we00:33:36
are creating a custom tool over here. So00:33:38
what this tool is taking let's try to go00:33:40
through with it and I will show you the00:33:42
complete execution inside the notebook.00:33:44
First of all, let me put out the code.00:33:46
Okay. Uh as I told you in a appropriate00:33:48
file. So here the we have a first rule00:33:51
which is check availability by the00:33:52
doctor. Then the second is a check00:33:54
availability by the specialization. Then00:33:56
we have a third one which is a site00:33:58
appointment. Okay. Then fourth one is a00:33:59
cancel appointment. Uh then we have a00:34:02
fifth reschedule appointment. So in00:34:04
total we have five tools. And here you00:34:06
can see what we are passing to the tool.00:34:08
We are passing to the desired date and00:34:09
the doctor name. Okay. To this tag00:34:11
availability by the doctor. And here is00:34:13
a complete logic behind uh the inside00:34:15
this tool. Okay. Now what it is going to00:34:17
be written we'll look into some time.00:34:19
Now here we have next tool check00:34:21
availability by the specialization. So00:34:22
we are giving a desired date. Okay. And00:34:24
we are validating this date using this00:34:25
date model as I already told you. So00:34:27
this user user input date we are going00:34:29
to be validated. Here is a00:34:31
specialization right according to the00:34:32
specialization we have written a logic00:34:34
over here. Then we have a next one which00:34:36
is a site appointment. So again we are00:34:37
giving a date time model and here you00:34:40
can see identification number and we are00:34:41
giving a doctor name and we are setting00:34:43
the appointment. If you want to cancel00:34:44
the appointment so again we are giving a00:34:46
date time and here we are giving a ID00:34:49
number and we are giving a doctor name00:34:51
and we are going to be cancel the00:34:53
appointment. Now reschedule appointment00:34:54
is also there right. So this tool is00:34:56
helping to me for performing a several00:34:59
task and whatever agents we have created00:35:02
it will interact like the LLM model00:35:04
frequently is going to be interact to00:35:06
the appropriate tool and it is providing00:35:07
me a result. I hope this toolkit is00:35:09
clear. Now coming to the next part. So00:35:11
guys we are going to define our LLM00:35:13
model as well. So instead the utils I00:35:15
will keep my LLM. Okay. So here let me00:35:18
keep my lm. Now see so I created one00:35:22
class for the llm model. The class name00:35:23
is LM model itself. And here we are00:35:26
passing a model name. Okay, it could be00:35:27
a configurable model means uh it this00:35:30
model name could be could come from any00:35:32
configurable file. You can create any00:35:34
JSON file or any AML file from there you00:35:36
can pick it up instead of passing it00:35:38
over here. But just for the sake of00:35:40
simplicity here only I pass this00:35:41
particular model name. Now I'm checking00:35:43
whether model name is there or not. If00:35:45
it is not there I will raise the error00:35:47
otherwise I'm taking a model name and00:35:49
I'm initializing the model name and00:35:50
using this particular method I'm going00:35:52
to be return it. Okay, this openai00:35:53
model. Now here is what here is my LLM00:35:56
model. Okay, this is just a testing00:35:57
file. Uh you can run this file00:35:59
individually and you can test whether00:36:00
your model is working fine or not. I00:36:02
already tested my model is working fine00:36:04
and here guys what I have done I00:36:06
mentioned the API keys where I mentioned00:36:08
the API key. I mentioned the API key00:36:10
inside this env. Okay. So in the same00:36:12
way you have to mention it. See I have a00:36:14
so many API key lengchen Google tably00:36:17
but these all the API key is not00:36:18
required for this particular project.00:36:20
only this API open API key is required00:36:22
for this particular project. Okay. Now00:36:25
coming to the next part. So I think so00:36:27
this API key is fine. Uh now along with00:36:30
the API key right? So if we're talking00:36:32
about the API key guys. So along with00:36:36
the API key what all information is00:36:37
required? Uh I think uh okay so what uh00:36:41
what all thing is required guys this00:36:43
exception handling and all for that also00:36:44
we can create a separate folder over00:36:46
here. The exception handling py and we00:36:49
can create a logger. PY. So later on I00:36:51
will show you I I was thinking about it00:36:52
but I couldn't create that because of00:36:54
the time constant but uh for sure in the00:36:56
next video or in the next like tutorial00:36:58
I will show you that exception handling00:37:00
and the logger as well. Okay then we00:37:02
have a next one which is a agent. py00:37:04
okay this agent. py is containing all00:37:06
the agents right so let me show you this00:37:09
particular code as well regarding the00:37:10
agents py and here guys this is a00:37:12
complete code and you can you can see00:37:15
guys I I created what I created this uh00:37:17
like the custom class what is my custom00:37:20
class name my custom class name is a00:37:22
doctor appointment agent now inside this00:37:24
doctor appointment agent we have three00:37:25
main agent okay let me show you the00:37:27
agent the first agent is a supervisor00:37:29
node the second agent itself is a00:37:32
information node okay and here the third00:37:34
one is a booking node. So this00:37:36
supervisor node what it is doing it is00:37:38
going to be delegate the task from one00:37:40
uh from like whenever it is taking a00:37:42
input it's going to be delegate either00:37:43
to the information agent or to the00:37:45
booking agent. Okay. So to get a better00:37:48
understanding so you can look into this00:37:50
particular architecture your doubt will00:37:52
be clarified then guys here what you can00:37:54
see. Okay let me check it is giving me00:37:56
error. Is there any issue with the input00:37:59
statement toolkits is there? Okay it is00:38:01
a toolkit. Uh it's a tool toolkit00:38:06
toolkits. Okay. So I missed this as and00:38:09
because of that I was guessing getting00:38:10
getting a issue. So let me check out the00:38:13
spelling now. So is it is tool kits not00:38:17
tools kit. Now it is perfect. Now see00:38:20
here guys we are not getting any sort of00:38:21
error and everything is fine over here.00:38:24
So guys I'll give you the quick overview00:38:26
of this particular supervised node and00:38:29
all. I'll show you this code inside the00:38:30
IP bind file. As of now just uh you can00:38:33
pause the video and you can take a look00:38:35
on top of it what I have written and00:38:37
here is my information node and the00:38:39
booking node as well. I I'll show you00:38:40
each and every line of code. Don't00:38:41
worry. Now inside this particular class00:38:43
I have created one more class one more00:38:45
function. The function name is a00:38:46
workflow and inside this workflow we are00:38:48
going to mention this nodes and ages and00:38:51
all. Okay. So what will be my node and00:38:53
what is my ages? So I I mention over00:38:55
here and this notes means it is00:38:58
representing the function right and uh00:39:00
and here the ages is nothing it's00:39:02
representing a connectivity between the00:39:04
function I no need to write all the00:39:06
conditional statement and all why00:39:07
because here I mentioned this command in00:39:09
return a statement and uh according to00:39:12
that it is all it is like it's going to00:39:15
be map okay from node to node the00:39:17
conditional and all everything so we00:39:19
just need required this command module00:39:21
and even I have explained you this thing00:39:22
in my previous tutorial as how this00:39:24
command module is working. If you want00:39:26
to get a detail of this command module,00:39:28
you can check out with my previous00:39:29
video. Now coming to the next part. So00:39:31
guys, this is done. Okay, I think my00:39:33
back end code is completed. Now what I00:39:35
have to do, I have to write my front end00:39:37
code. Okay, so I'm going to write my00:39:38
front end not with the HTML or not with00:39:40
the CSS. Okay, so here we are going to00:39:43
be write that front end code using the00:39:44
HTML only. So let me show you the front00:39:47
end code how it looks like. So here is00:39:49
my front end code guys. I'll give you00:39:50
the walk through of the entire code. No00:39:52
need to worry. So here see what we have00:39:54
we have a title okay this title will be00:39:56
represented then we have a user ID and00:39:58
query which we are taking from the inh00:40:01
from the user only and then here is my00:40:02
button right so whenever we'll submit00:40:04
this button this will hit the API okay00:40:07
this API we are going to be create using00:40:08
the fast API itself so after that I'll00:40:11
show you the API as well how we can00:40:12
create a API so this will hit to the API00:40:15
and here is my API URL this one and00:40:18
after that if it is getting response00:40:20
code is 200 from that particular API In00:40:22
that case we are getting a response back00:40:24
and we are going to be represent this00:40:26
response back we are going to be write00:40:27
it over here. Yeah like this is very00:40:29
simple code I think you can00:40:30
understanding just by reading this00:40:32
entire code and now what we are doing.00:40:34
So here we are going to be create a API.00:40:36
So inside this main py file I'm going to00:40:38
be write a code for the API for the API00:40:40
creation. So guys uh here let me show00:40:42
you the code for the API creation how it00:40:44
looks like. So this API we are going to00:40:46
be create using the fast API itself. So00:40:48
we are going to be import the fast API.00:40:50
Here we are going to be import this base00:40:52
model. Okay, this is for the this is for00:40:54
the py object. Here is my doctor00:40:56
appointment agent and here you can see00:40:58
my human message and OS. Now uh this is00:41:01
uh okay so here guys I mentioned this00:41:03
SSL certificate file. If you're getting00:41:05
any SSL related issue you can mention00:41:07
this particular file because what we are00:41:08
doing it guys we are running it over the00:41:11
we are exposing the API. So it sometime00:41:14
it shows this particular issue if you00:41:16
have a antivirus or something inside00:41:17
your system but for the production00:41:19
system it is pretty much required this00:41:21
SSL certificate okay for the security00:41:23
reason so we will see if we are going to00:41:26
be deploy any sort of a project over the00:41:28
cloud now what I'm doing here I'm going00:41:30
to be create a object of the fast API00:41:32
class and here is my class guys user00:41:34
query and here I'm importing the base00:41:36
model you can see we we have ID number00:41:38
and message so this two thing we are00:41:40
going to be validate whatever user is00:41:42
passing right in a in a text form sorry00:41:45
inside a text field right so those thing00:41:47
we are going to be validate using this00:41:48
pentic model now here is my agent okay00:41:51
this agent needs to be called so where00:41:53
we are calling this particular agent so00:41:54
we are calling inside this particular uh00:41:57
route okay so let's execute so inside00:41:59
this let's execute we are getting a00:42:01
query here we are going to be initialize00:42:02
a workflow okay as see we got a workflow00:42:05
so this is my all the input data which00:42:07
we need to pass to the workflow okay and00:42:09
here guys you can see we have a input00:42:11
this is a message from the user. Here is00:42:13
my query data. So we are going to be00:42:15
form this JSON object sorry this uh00:42:17
dictionary basically. Why it is00:42:18
required? Because this all the key is00:42:20
available in the pyic class which we00:42:22
have defined message ID next query00:42:25
current reasoning and all. So in this00:42:27
particular format only we'll have to00:42:28
provide a data. So here is my message00:42:30
which is a input only. Here is my ID00:42:32
number. Here you can see the ID number.00:42:34
We are validating using this uh00:42:36
basically user query only. And uh here00:42:38
you can see we are going to be uh00:42:40
validate this user input using this00:42:43
message key. Right? So I hope so this00:42:45
everything is perfect. Now we are00:42:47
passing this particular thing to the00:42:48
application00:42:50
app.graph.invoke and whatever response00:42:52
we are getting from our workflow we are00:42:54
going to return it back to the00:42:56
streamllet UI over here. Okay. Now we'll00:42:58
be getting a response and we are going00:43:00
to be write it down over the UI. Okay.00:43:03
Using this st.right. I hope so guys this00:43:06
entire code is clear. I kept the code in00:43:08
the appropriate files and folder. Now00:43:11
let's take a look on top of this00:43:13
notebook. Let's try to understand this00:43:15
code in a little depth and then we'll00:43:17
try to execute it and we'll see the00:43:19
response. So where is my notebook guys?00:43:21
First of all, let me close this entire00:43:23
file and let me open the notebook and00:43:26
here you will get a nitty-gritty detail00:43:28
of each and00:43:29
everything each and everything of this00:43:32
code. So yeah, I think I selected the00:43:34
environment also. So let's try to00:43:36
understand this code. So first guys,00:43:39
let's import all the uh modules. Uh here00:43:43
I'm going to be loaded. Then uh here I'm00:43:45
going to load my environment variable00:43:48
using this load env. Uh then guys I here00:43:51
I'm going to setting up the openi key.00:43:54
Okay. And then after setting up see here00:43:56
is my open key. Let me delete it. Then00:44:00
I'm going to set it up as a environment00:44:02
variable. Now here I'm going to00:44:04
initialize my open air model which is00:44:06
GPD40. And now here I'm going to test it00:44:08
whether it is working fine or not. So00:44:10
everything is working fine uh so far.00:44:12
Now let's try to understand about this00:44:14
data time model. Uh so this I already00:44:16
explained you. Now let me show you how00:44:18
it is validating the date. So this is00:44:22
fine. Uh this date model is also fine.00:44:24
Now identification number is also there.00:44:26
This is also perfect. Now here is my00:44:28
tool. Uh so let's try to check out this00:44:30
tool. uh let's try to call this tool and00:44:33
then we'll get a better understanding.00:44:34
So what we need to pass over here we00:44:36
need to pass the desired date whenever00:44:38
we have to check the availability of the00:44:40
doctor and here is a doctor name. So00:44:42
literal means what literal is just00:44:44
representing to the type hinting. Now00:44:46
here you can see we have the DF dot read00:44:51
CSV we are going to read the data. Now00:44:53
here what we are going to do we are00:44:54
going to be segregate the date and time.00:44:56
Okay. And here this is my logic based on00:44:59
this particular logic what we are doing.00:45:00
we are going to be check the00:45:02
availability. So uh what I have written00:45:04
over here. So here we are going to be00:45:06
split the date first. Okay. Uh then we00:45:09
are matching it with the desired date00:45:11
whether the date is matching or not. If00:45:13
the date is available along with this we00:45:15
are going to be check the doctor name.00:45:17
See here is a doctor name. Then we are00:45:20
checking doctor is available or not. If00:45:22
it is available then what we will give?00:45:23
We'll give the time slot. Okay. So this00:45:26
is my filter condition which I'm going00:45:27
to be apply. You can go through with it00:45:30
and you can understand but whatever I00:45:32
explain you I think you got the idea00:45:34
that how it is working. So we are00:45:36
combining multiple condition here you00:45:37
can see desired date with the available00:45:39
date whatever date is there then here is00:45:41
my doctor name and here is my00:45:43
availability here is availability of the00:45:45
doctor and according to that I'm00:45:46
providing the date and time slot so if00:45:49
row is equal to zero in that case there00:45:51
is no availability in the entire day00:45:52
with respect to the doctor if it is00:45:54
there then here we are returning a00:45:56
output. So let's see how the output00:45:58
looks like. So let me pass the date over00:46:00
here. So I'm passing this particular00:46:01
date and see uh see guys. Uh here I'm00:46:04
checking how the date looks like. Okay,00:46:06
we are checking with the date model00:46:07
only. Then uh here I'm going to call it.00:46:11
So once I will call it guys, so see this00:46:13
is all the time slot when the doctor is00:46:15
available. So 8:00, 8:30, 11:30, 12:00,00:46:18
12:30. So this doctor is available at00:46:20
this particular time slot at the given00:46:22
date. I hope this is clear. Now let's00:46:25
try to check by the specialization or00:46:28
let's say there is some person which00:46:29
don't know the doctor name but they are00:46:31
they want they want to know their00:46:33
problems. So let's say someone want to00:46:34
search dentist, cosmetics, dentist or00:46:37
the uh00:46:39
proist. Okay. Now here is a pediatric.00:46:42
So there is a different different00:46:43
occupation which I mentioned over here00:46:45
inside the uh list itself right inside00:46:48
the specialization. So uh let's see what00:46:51
is happening. So again it is reading a00:46:52
data. Now here again we are going to be00:46:54
segregate the date and time right and00:46:56
then what we are doing so we are00:46:57
checking. So here is a condition. So if00:47:00
the specialization is getting match if00:47:01
the date is getting match if the uh00:47:04
availability is there in that case we00:47:06
are going to be group all this thing00:47:08
right and then we are giving it back. So00:47:11
here we have written one more additional00:47:13
thing. So here is what here is a convert00:47:15
to A.M. to PM. Okay, because you can see00:47:18
in some time basically uh the this that00:47:21
is from 0 to 24, right? So in that case00:47:24
we have to convert from A.M. to P.M. So00:47:26
let's say we are talking about the P.M.00:47:28
right? So it is from 12 night uh sorry00:47:31
12 noon to 12 night. And if we are00:47:33
talking about AM so it is from 12 night00:47:34
to 12 noon right? So this is what we are00:47:37
going to be perform over here. Now how00:47:38
you will get your output? Let me show00:47:40
you. So here is my date. Again I'm00:47:42
checking with my patient object. Now00:47:44
here I'm checking with the availability.00:47:45
Okay. by the specialization. So see here00:47:47
you are getting availability. So you're00:47:49
getting 8:00 a.m. uh 11:30 p.m. then 200:47:52
p.m. Okay, because inside the data they00:47:54
haven't given this particular thing. If00:47:55
you will go and check with the data00:47:56
itself. So see there there is no A.M. PM00:47:59
something like this. So from where from00:48:02
which time uh like which is representing00:48:05
to AM from and which time it is00:48:07
representing to A.M. sorry PM. So00:48:10
everything detail I have added over here00:48:12
programmatically only. So where it is00:48:14
inside this IP1B file. I hope you00:48:16
understood this entire thing. Now let's00:48:18
try to test the next tool. So the next00:48:20
tool is re reschedule the appointment.00:48:22
So before rescheduling the appointment,00:48:24
you will have to understand two more00:48:25
tool which is cancel appointment and set00:48:27
appointment. Okay. So let's understand00:48:29
this cancel appointment how it is00:48:30
working. So if I want to cancel the00:48:32
appointment for any sort of a doctor. So00:48:34
here what I'm doing I'm passing my date00:48:36
and time model. Okay. Inside a date and00:48:38
time model I would be having date and00:48:39
time both. Then here is my00:48:42
identification number. Okay. Uh00:48:44
identification number you will get when00:48:46
when you are going to be set the00:48:47
appointment. So here I want to cancel00:48:48
the appointment. So identification00:48:50
number would be required. So this is00:48:51
also I'm going to validate using this00:48:53
identification number model. Then here00:48:55
is my doc doctor name to which doctor00:48:57
I'm going to be cancel the appointment.00:48:58
Okay. Now uh I gathering all the00:49:01
information. Now here is my data. Now00:49:03
case to move. Right. So if I want to00:49:05
cancel the appointment. So first I'm00:49:07
checking with the date patient to00:49:09
attend. Okay. this id and the doctor00:49:11
name. So if it is going to be match this00:49:14
all the thing is going to be match. So00:49:15
what we are doing see if is not going to00:49:17
be match we are getting zero. So here I00:49:19
return you don't have any specific00:49:20
appointment. If this thing is going to00:49:22
be match in that case see here what I'm00:49:24
doing uh so patient to attend. Okay so00:49:28
instead of true see uh here you can see00:49:32
uh just just follow from here df lock df00:49:35
date slot date is date. Okay. Then here00:49:38
is a patient to attend is equal to what?00:49:40
ID number. This is a one condition. This00:49:42
one. Now the second is what? So sorry.00:49:45
Third is what? Doctor number. So what00:49:47
I'm doing instead of this is available.00:49:49
Okay. Or this patient to attend what I'm00:49:51
passing true and none. So what is the00:49:53
meaning of this true and none inside00:49:55
this particular file? See if there is00:49:58
true. True means uh basically doctor is00:50:01
available and here is don't no patient00:50:03
ID means my doctor is available at this00:50:05
particular time slot means we are going00:50:07
to be cancel our appointment. I hope you00:50:09
understood this logic. Okay. So if there00:50:13
is a doctor appointment in that case it00:50:14
would be false and there will be some ID00:50:16
but if we are going to cancel the00:50:18
appointment now the doctor will be00:50:19
available. So it will be representing00:50:21
false and the ID over here. That's it.00:50:23
Now we are again giving back to the CSV00:50:26
only. So, DF2 CSV availability dot CSV.00:50:31
Okay, let me check out the specific name00:50:32
with which name we have to create a00:50:34
file. Uh, I think it is there. So, we00:50:37
have to create a file with this specific00:50:39
name. Uh, just a second guys. Let me00:50:43
mention the file00:50:45
itself. So, here this would be the00:50:50
file. I hope this is clear. Just a00:50:54
second. Let me take a specific path only00:50:57
and let me paste it00:51:12
on. Okay, perfect. Now let me execute it00:51:15
and see uh I'm checking with my daytime00:51:18
model. Here is my notification number.00:51:19
Now cancel the appointment. So if there00:51:22
would be any appointment, so it's going00:51:23
to be cancelled that. See successfully00:51:24
cancelled. Okay. Now set appointment00:51:26
again it's going to do a same thing. So00:51:29
set appointment means if the appointment00:51:30
is not there first of all I will take00:51:33
appointment and then basically I'm going00:51:35
to be define as a set means I'm going to00:51:37
put the put my value over there. Let me00:51:39
show you what all value basically we are00:51:41
keeping we are going to be set it. Let's00:51:43
try to trace a logic from scratch. So00:51:45
first we are going to read the data. Now00:51:46
here convert date time format. So here00:51:48
we are going to be convert the date time00:51:50
format. Okay, means we are checking the00:51:52
date date and the time it is in a00:51:54
specific format or not. Okay, if it is00:51:57
not we will be converting that. Now00:51:58
after that we are we have a condition00:52:00
over here. So after the converting and00:52:03
all see here we have a case then what is00:52:06
happening? So date time format guys why00:52:08
we are going to be converted because we00:52:10
want to be match okay with the same date00:52:13
time format which is available inside00:52:14
this particular file. If it is matching00:52:17
with this date time format means the00:52:19
appointment is there means what is the00:52:22
meaning of it the appointment is there00:52:24
okay and we can set that particular00:52:25
appointment see over here what we are00:52:26
doing so we are passing it to this00:52:28
condition and if it is not matching so00:52:30
I'm saying no available appointment for00:52:31
the particular case00:52:33
okay otherwise here you can see all the00:52:36
thing is going to be match okay in that00:52:38
case what we are going to do we are00:52:40
saying okay this appointment is00:52:42
available and we are going to be update00:52:44
inside the file also now I Hope this00:52:46
thing is clear. See here I'm passing is00:52:48
available and patient to attend. False00:52:51
and ID number. So false and ID number00:52:53
means what? To that particular position00:52:55
we are going to be passed this false and00:52:56
ID number means position is there. Okay.00:52:59
Sorry. Appointment is there. So that's00:53:01
why it will be representing false and00:53:02
that ID number. I hope so this thing is00:53:05
also clear. Now let me execute it. So00:53:07
guys once you will go through with it00:53:08
now one by one step by step you will be00:53:10
getting each and everything. So this00:53:12
tool by design by me only. And here you00:53:14
can see we have a reset appointment. So00:53:15
based on the date and time we are going00:53:17
to be rescheduled the appointment and00:53:19
here no available slot for the desired00:53:21
period. If the condition is not going to00:53:22
be matched if the condition is going to00:53:24
be matched then first we'll cancel the00:53:26
appointment and then again we'll set the00:53:27
appointment. So this two method cancel00:53:29
appointment and send appointment00:53:31
basically it belong to this rescheduling00:53:33
appointment. Okay let me show you how it00:53:35
works. So here I'm giving my date and00:53:36
time and now see here is my new time old00:53:39
time. Now here is my identification00:53:41
number. Okay this one and I'm checking00:53:43
with the rescheduling. So it is saying00:53:45
no available slot is desired period. So00:53:47
at this particular time the new time the00:53:50
available slot is not available. I hope00:53:52
this thing is clear. Now coming to the00:53:55
next part. So guys I was talking about00:53:58
this particular information the member00:54:00
dict. So here you can see the member00:54:01
dict. Okay. Uh let me execute it and let00:54:04
me show you how this member dict looks00:54:06
like. So just a second guys. Let me00:54:09
print this particular variable members00:54:12
and see. So we have two node and this is00:54:14
the information of this two node. Now00:54:15
here let me run it and see the all the00:54:17
option. This is all the option00:54:19
information node booking node and00:54:20
finish. Now here let me show you the00:54:22
working information. So the this is the00:54:23
working information. Now this particular00:54:25
information we are going to append00:54:26
inside the system prompt and I hope so00:54:28
this is fine right? This is also perfect00:54:31
to you. Now after this one guys what we00:54:33
are doing we are going to be uh define00:54:35
my agents. Okay the main agents. Now00:54:37
let's try to understand this supervisor00:54:40
agent. Okay. So to the supervisor agent00:54:43
what we are doing guys we are passing a00:54:44
agent state. Uh here you can see we have00:54:48
already defined the agent state this00:54:49
one. So what all thing uh what all keys00:54:52
are flowing throughout the flow00:54:55
throughout the uh this agentic flow. So00:54:58
message would be there ID number next00:55:00
query and reasoning with respect to that00:55:02
query. Okay. Now uh here uh let me show00:55:06
you what will happen. So the for see00:55:08
here uh command is going to be written00:55:11
this three thing information either00:55:12
information node revoking node or this00:55:14
finish right so we are going to print00:55:16
the state this print statement just to00:55:18
understand the flow that's it what is00:55:20
being passed what is not being passed00:55:21
that's it so here is my message which is00:55:23
coming so in the message I'm appending00:55:25
the system prompt I'm appending the user00:55:27
ID and here is my system message right00:55:29
so this is a combined message I printed00:55:32
the message as well now here I'm00:55:33
checking a query so if uh system message00:55:36
is equal to one in that case I am taking00:55:38
a query from there. Okay, means I'm00:55:40
taking a question. I'm keeping it inside00:55:42
the query and then I'm passing it to00:55:45
this uh see what I'm passing to this one00:55:47
open AI dot openi model with structure00:55:50
output uh we are passing this message00:55:53
okay this particular message this entire00:55:55
message and here I pass the router as00:55:58
well see what is my router just look00:56:00
over here look into the router so the00:56:02
router actually is providing either this00:56:04
next or this reasoning okay based on the00:56:06
given message so what it will what it00:56:09
will what it will sorry what it will00:56:11
provide. So that particular thing will00:56:13
go will get inside this go to variable.00:56:15
Okay, from this response uh next from00:56:17
this response uh and this next and here00:56:20
I'm going to print this go to and then00:56:22
here is I'm going to print my reasoning.00:56:24
If go to is finished in that case we are00:56:26
going to uh stop the process. Now below00:56:29
is my state. So I'm printing the state00:56:31
as well. Okay. And here guys you can see00:56:33
what we are doing. So if query is there00:56:36
means if it if we are running it first00:56:38
time in that case what we are doing? So00:56:40
in that case we are passing the go to00:56:43
means where we have to go right from00:56:45
this particular stage what will be the00:56:47
next agent here is my query means my00:56:49
question here is a reasoning okay and00:56:51
here is my human message. So why I have00:56:53
written this query and this why I made00:56:55
this particular logic because if I'm00:56:57
passing it first time then only this00:56:59
identification number this particular00:57:01
information is required. Okay. Now if00:57:03
you will look into the second return00:57:04
statement if query is not there when the00:57:06
query will not be here see over here if00:57:08
the message length is more than one00:57:10
means in the next iteration the message00:57:12
length will be more than one in that00:57:13
case the query will not be here. Okay.00:57:15
So based on this particular message what00:57:17
I'm getting I will be getting my next00:57:18
and here whatever reasoning would be00:57:20
there. So I'm passing this next. Okay,00:57:22
next means what? To which agent we have00:57:25
to go and this current reasoning. So at00:57:27
first time I'm passing this uh00:57:28
information along with this user ID.00:57:30
Okay, and this query and then in every00:57:33
iteration I'm passing the reasoning.00:57:35
Whatever reasoning is happening based on00:57:36
the given message and the next node.00:57:39
That's it. I hope you understood this00:57:41
logic. Now let me execute it. This is00:57:43
fine. Information node. Now let me show00:57:45
you the information node. What is there?00:57:47
The information node. We have the system00:57:49
prompt which we have designed according00:57:50
to the information node. You can pause00:57:53
this video and you can read out this00:57:54
particular prompt. Okay. What is there?00:57:57
So here uh you are a specialized00:57:59
assistant. Let me do the water app. Let00:58:01
me see the complete detail. Okay. So00:58:03
here see you can you can pause the video00:58:05
and you can read the complete prompt. So00:58:07
we are passing it to the s we are00:58:09
passing it to the chat prompt template.00:58:10
We are creating a prompt template and00:58:12
then we are creating this react agent.00:58:14
Okay. It's an inbuilt agent which we are00:58:16
creating. So we are passing a model open00:58:18
AAI means this is my open AI model and00:58:20
here is my check sorry tools. So what00:58:22
are tools we are passing over here? We00:58:24
are just passing two tools check00:58:25
availability by the doctor and check00:58:26
availability by the specialization and00:58:28
we are passing the prompt as well. So00:58:30
whatever information we'll be getting00:58:32
after this reasoning. Okay this is00:58:34
individual agent only guys. Okay this00:58:35
information node. So after the00:58:37
individual uh sorry after the uh00:58:39
execution whatever information we'll be00:58:41
getting. So we are going to be uh return00:58:43
it using this command. So where it is00:58:45
going again? It is going to the00:58:46
supervisor. Now supervisor will take a00:58:48
decision what needs to be done. And this00:58:50
is a updated message guys. Okay. Which I00:58:52
am going to be mention which I mention00:58:54
over here. Now this is my first thing my00:58:57
first uh note the child node. This is my00:58:59
second child node. Okay. So you can read00:59:01
this about this child node. It is all00:59:03
about the booking. Just look into this00:59:04
react agent. Create react agent and you00:59:07
will get the clearcut understanding over00:59:08
here what is happening. So see uh first00:59:11
you can read out the complete prompt.00:59:13
Okay. I don't know why it is not00:59:15
working. So you can read out the00:59:16
complete prompt guys. Here you can see00:59:19
let me do the right swipe. See uh you00:59:22
can pause the video you can read it. And00:59:23
here is I'm creating a prompt and then00:59:25
this is my react agent. Now to this00:59:28
react agent we are passing a set of00:59:30
tools. So set appointment, cancel00:59:32
appointment and reschedule appointment00:59:34
because it is my booking in order right00:59:35
and here it is doing a reasoning and00:59:37
based on that based on the result we are00:59:39
updating the agent. Okay that's it. Uh00:59:42
now uh this is fine. Uh here I'm going00:59:44
to create my graph. So first is my00:59:46
supervisor node. Then my information00:59:48
node then here is my booking node. Okay.00:59:50
And here I'm going to be start the00:59:52
process. Here I'm going to compilation00:59:53
it compile it. Then here is my app. So00:59:56
how the app will look like? It will look00:59:57
like this only. Now here is my input00:59:59
message guys. Uh and here along with the01:00:01
input see what will be the my input01:00:03
message. My input message going to be01:00:05
include this input as well as this id01:00:07
number. So this is going to become my01:00:09
state. Now this particular state I'm01:00:10
going to pass to my invoke app01:00:12
do.invoke. Okay. Now see guys, it is01:00:15
working and see the entire uh thing is01:00:18
going to be execute and once it will be01:00:19
done I'll be getting my result. So01:00:21
simply I can print my result. So for01:00:24
printing a result just a second. Uh let01:00:27
me give you the01:00:30
logic. So after running it how you will01:00:33
be getting a result. So guys sometimes01:00:35
you might get a accurate information01:00:37
sometime you might not get. If you're01:00:38
checking with other model like grog and01:00:40
all you might get a issue okay because I01:00:44
seen uh the reasoning capability of the01:00:46
forro is quite good compared to the01:00:47
other model. So please be stick with a01:00:49
good model if you want a good result and01:00:51
this scenario you can fit your uh you01:00:53
can fit in your any use case. Okay let01:00:56
me show you the result. So see here is a01:00:57
result. So currently there is no01:00:59
available for the general dentist01:01:00
whatever question I have asked. Okay.01:01:02
And here it is giving me alternative01:01:04
option also. So here yes guys we are01:01:07
able to achieve our result but this01:01:10
result basically we are able to achieve01:01:12
based on this question where we are01:01:13
checking a booking with the general01:01:14
dentist on 8 p.m. and 24 2024 at 8:0001:01:18
p.m. right 8 sorry 8th August 2024 at01:01:21
8:00 p.m. And this is the complete01:01:23
result. Now let's see how we can get the01:01:25
same result over the UI itself. And if01:01:28
you want guys uh if if I should provide01:01:30
you the complete uh basically chatbot01:01:33
kind of UI where you can do the chatting01:01:35
and all. So I'll provide you that as01:01:36
well. Now first of all what we have to01:01:38
do guys we have to execute this entire01:01:40
code. Let me show you how we can execute01:01:42
it. So open the terminal and here the01:01:45
terminal let uh here you have to do the01:01:48
split terminal. Okay just click on this01:01:50
split terminal get bash only and the new01:01:52
terminal is getting open. Let me clean01:01:54
it clear it first of all. So here I'm01:01:56
going to be up my streamlit UI first.01:01:59
Okay. So let me up my streamlit UI. How01:02:04
to do that? Let me show you. And here01:02:06
guys, I will up my uh fast API. Okay. So01:02:11
here I'm going to write my uh command01:02:13
source activate01:02:16
dot s o ur source activate dot / venb.01:02:23
Okay. Now yes my environment is ready.01:02:27
Now what I will do guys I'll be writing01:02:29
a command over01:02:31
here. Okay so guys here I written a01:02:34
command you can see streamlit run01:02:36
streamlit ui.py. This is a command for01:02:39
upping the server for running the01:02:41
server. This is streaml and here is a01:02:43
command for running the server of this01:02:44
fast API. So ubicon main is my file01:02:47
name. App is a app inside this main. py01:02:50
file. reload it will automatically01:02:52
reload it if it is going to be detect01:02:54
any sort of a change and here is a port01:02:56
okay on which port I'm running it so by01:02:58
default it will take 8,000 port so if I01:03:00
I already executed my code on top of01:03:02
that it was running over there so that's01:03:03
why I'm mentioning the other port right01:03:05
so guys uh here whenever you are going01:03:07
to be write any sort of a port over here01:03:08
so please make sure that you are going01:03:10
to be write a same port inside your API01:03:11
URL as well okay otherwise it will give01:03:14
you the issue it will not hit the same01:03:16
API let me execute and let me see01:03:18
whether it is working fine or not so01:03:20
first I'm going to up my streamlet UI.01:03:22
So here uh it is getting execute. Let's01:03:25
see whether it is working or not. And01:03:28
then uh here I'm going to execute my01:03:31
fast API also. Okay. So yes guys, uh I01:03:36
think it will up just a second. Let me01:03:38
show you01:03:39
that. Uh give me a moment guys.01:03:43
Okay. Yeah. So it is asking for the01:03:45
allow. Uh let me share that UI on top of01:03:49
this01:03:50
screen. Give me a01:03:52
moment. Yeah guys.01:03:57
So here is a UI. Uh wait where it is01:04:08
representing. Yeah perfect guys. So I01:04:10
got a UI over here and see this is my01:04:16
UI. Okay. Now what I have to do I have01:04:20
to execute this fast API also. So let me01:04:23
execute this fast API. And uh here it is01:04:27
running. It will only up sorry it will01:04:30
also up. Just a01:04:32
second. So the code is loading.01:04:38
It takes time guys if you're running it01:04:40
first time now. It takes time.01:04:52
Okay. Okay. So, perfect. My fast API is01:04:56
also up. It is on a different port. So,01:04:59
here is a URL. Once you will click on01:05:01
this URL, you will get that fast API01:05:05
also. Okay. So just a second let me01:05:08
check it out whether it is up or not. So01:05:10
yeah it is up. Now guys here what you01:05:12
have to do you have to write this /01:05:14
docs. So you will get your uh swagger01:05:17
UI. Okay. Now see this is your swagger01:05:19
UI and here you can see your endpoint01:05:22
this one. So you you have to hit uh this01:05:25
particular endpoint this uh 127.0.0.101:05:30
0.1 col 80002 /01:05:34
docs/ this one execute only. Okay, docs01:05:37
is not required. No need to write a01:05:38
docs. It is just to check with a swagger01:05:40
UI only. Okay, if you're going to be01:05:42
write slcute only. So it will01:05:45
automatically hit this particular01:05:46
endpoint. Now let me show you how to hit01:05:48
that. So first of all uh let me take a01:05:51
same question. So I'm going to take a01:05:53
question guys. Let me check whether I'm01:05:55
getting a same response or not. So where01:05:58
I kept the question in my IPB file. So01:06:01
here is my IPB and what question I took.01:06:06
This was my question. Okay. No need to01:06:08
mention this thing. Uh you just need to01:06:10
mention the ID and this particular01:06:12
question. Everything will work for you01:06:14
as expected. If you want to me to create01:06:17
a uh basically a separate video with a a01:06:22
proper UI uh definitely I will do that.01:06:24
So here I will put this question and01:06:27
here I will put my uh ID. Okay, you can01:06:30
put anything. Uh I I'm just testing with01:06:32
the same question whether it is working01:06:35
or not. So let me take the ID also.01:06:40
Okay, but feel free to check out with01:06:42
any sort of a uh ID or a question. Uh01:06:45
for sure it's going to work if you have01:06:47
write a code according to me. Now what01:06:49
I'm doing guys, I'm submitting the01:06:50
query. So it is going to the fast API.01:06:53
Let's see whether everything is working01:06:55
fine or not. Over the console itself you01:06:56
will you can check it. See over the01:06:59
console everything is going to be01:07:00
execute guys in a back end. Now what I01:07:02
will get if everything is perfect. So01:07:04
here I will get my final answer. So01:07:06
let's wait for the final answer and uh01:07:10
yeah let's wait for01:07:13
that. All right guys, so our code01:07:16
execution is completed and uh here it is01:07:19
returning some output over the UI. Let's01:07:21
see what it is going to return. Uh so01:07:24
here is the output guys. See it is01:07:25
returning a complete uh uh JSON itself01:07:29
and I think the specific slot is not01:07:31
available because of that. So just look01:07:33
into the add answer the last answer. See01:07:36
it it is saying I'm sorry for the01:07:37
inconvenient but this there is no01:07:39
accessing availability information at01:07:40
this moment. Okay means with respect to01:07:43
the doctor there is no available slot01:07:45
because of that it is giving me this01:07:46
particular information. It is adding01:07:48
some more information inside the hand01:07:49
message. So you can tweak that according01:07:51
to your prompt and all. Okay. Now uh01:07:54
guys if you want the answer only this01:07:56
particular answer. So what changes you01:07:58
will have to make inside the code. So uh01:08:01
just uh look out look into the code and01:08:05
inside the code go with your streamllet01:08:06
dot01:08:08
streamllet_UI. And here instead of this01:08:10
message write message one. So it will01:08:12
only give you the last message. Okay.01:08:14
And you can print the content and all01:08:15
something like that. so that you will be01:08:18
able to get your message only right over01:08:20
here. So I'm leaving it up to you. You01:08:22
can test it out again with a different01:08:23
different query. And one more thing here01:08:26
inside this particular application along01:08:28
with the exception and logging you can01:08:30
mention one more concept that is human01:08:33
in loop. I already shown this thing in01:08:35
my previous video. So if you will go and01:08:37
check with my YouTube channel there you01:08:38
will find out the human in loop concept.01:08:41
Okay, you can incorporate that human in01:08:43
loop inside. So just look into this01:08:45
particular uh uh this particular01:08:48
basically playlist. Okay, you'll find01:08:51
out all the videos. So yeah, this is it01:08:53
for this particular video guys. I hope01:08:55
you like this session and I created this01:08:58
project by myself. There are lots of01:09:00
effort. So if you're liking this content01:09:02
then you can hit a like button. You can01:09:04
subscribe the channel and uh soon I'll01:09:07
be uploading couple of more video with01:09:08
respect to the multi-agenting and all. I01:09:10
will try to explore couple of more01:09:11
framework with a different different01:09:13
concept of the generative AI. So yeah,01:09:16
bye-bye. Thank you. Take care. I'll see01:09:18
you in the next video.Description:
In this video, we demonstrate a fully automated, end-to-end multi-agent AI system for booking doctor appointments. Using agentic AI, each agent handles a specific part of the process — from understanding the user's request to finding available doctors and scheduling the appointment. This showcases the power of AI agents working together to complete complex tasks without human intervention. What You'll Learn: ✅ Understanding Multi-Agent Systems: How multiple AI agents can collaborate, specialize, and communicate efficiently. ✅ Role of a Supervisor Agent: How a central supervisory agent can oversee and coordinate multiple agents, ensuring task delegation, monitoring, and execution. ✅ LangGraph Implementation: Using LangGraph to design, connect, and control multiple agents with structured workflows. #langchain LangGraph Don't miss out; learn with me! P.S. Don't forget to like and subscribe for more AI content! PDF: https://github.com/sunnysavita10/langgraph-end-to-end/tree/main/multiagent_system End-to-End-Langgraph-Project: https://github.com/sunnysavita10/doctor-appoitment-multiagent Multimodel RAG Playlis: https://www.youtube.com/watch?v=7CXJWnHI05w&list=PLQxDHpeGU14D6dm0rmAXhdLeLYlX2zk7p&pp=gAQBiAQB RAG detailed Playlist: https://www.youtube.com/watch?v=wTVTkOb3SZc&list=PLQxDHpeGU14Blorx3Ps1eZJ4XvKET1_vx&pp=gAQBiAQB GenAI Foundation Playlist: https://www.youtube.com/watch?v=ajWheP8ZD70&list=PLQxDHpeGU14D7NiPgqxC9qhKkx4jMQcDk&pp=gAQBiAQB Connect with me on Social Media- LinkedIn : https://www.linkedin.com/in/sunny-savita/ One to One Call: https://topmate.io/sunny_savita10 GitHub : https://github.com/sunnysavita10 Telegram : https://t.me/aimldlds
Mediafile available in formats
* — If the video is playing in a new tab, go to it, then right-click on the video and select "Save video as..."
** — Link intended for online playback in specialized players
Questions about downloading video
How can I download "LangGraph:21 End-to-End Supervisor Multi-Agentic AI Project for Booking Doctors Appointment #aiagent" video?
http://univideos.ru/ website is the best way to download a video or a separate audio track if you want to do without installing programs and extensions.
The UDL Helper extension is a convenient button that is seamlessly integrated into YouTube, Instagram and OK.ru sites for fast content download.
UDL Client program (for Windows) is the most powerful solution that supports more than 900 websites, social networks and video hosting sites, as well as any video quality that is available in the source.
UDL Lite is a really convenient way to access a website from your mobile device. With its help, you can easily download videos directly to your smartphone.
Which format of "LangGraph:21 End-to-End Supervisor Multi-Agentic AI Project for Booking Doctors Appointment #aiagent" video should I choose?
The best quality formats are FullHD (1080p), 2K (1440p), 4K (2160p) and 8K (4320p). The higher the resolution of your screen, the higher the video quality should be. However, there are other factors to consider: download speed, amount of free space, and device performance during playback.
Why does my computer freeze when loading a "LangGraph:21 End-to-End Supervisor Multi-Agentic AI Project for Booking Doctors Appointment #aiagent" video?
The browser/computer should not freeze completely! If this happens, please report it with a link to the video. Sometimes videos cannot be downloaded directly in a suitable format, so we have added the ability to convert the file to the desired format. In some cases, this process may actively use computer resources.
How can I download "LangGraph:21 End-to-End Supervisor Multi-Agentic AI Project for Booking Doctors Appointment #aiagent" video to my phone?
You can download a video to your smartphone using the website or the PWA application UDL Lite. It is also possible to send a download link via QR code using the UDL Helper extension.
How can I download an audio track (music) to MP3 "LangGraph:21 End-to-End Supervisor Multi-Agentic AI Project for Booking Doctors Appointment #aiagent"?
The most convenient way is to use the UDL Client program, which supports converting video to MP3 format. In some cases, MP3 can also be downloaded through the UDL Helper extension.
How can I save a frame from a video "LangGraph:21 End-to-End Supervisor Multi-Agentic AI Project for Booking Doctors Appointment #aiagent"?
This feature is available in the UDL Helper extension. Make sure that "Show the video snapshot button" is checked in the settings. A camera icon should appear in the lower right corner of the player to the left of the "Settings" icon. When you click on it, the current frame from the video will be saved to your computer in JPEG format.
How do I play and download streaming video?
- in video formats, hover your mouse over "Streaming Video**";
- right-click on "Copy link";
- open VLC-player;
- select Media - Open Network Stream - Network in the menu;
- paste the copied link into the input field;
- click "Play".
- copy the video address (URL);
- select "Open Network Stream" in the "Media" item of VLC player and paste the link to the video into the input field;
- click on the arrow on the "Play" button and select "Convert" in the list;
- select "Video - H.264 + MP3 (MP4)" in the "Profile" line;
- click the "Browse" button to select a folder to save the converted video and click the "Start" button;
- conversion speed depends on the resolution and duration of the video.
For this purpose you need VLC-player, which can be downloaded for free from the official website https://www.videolan.org/vlc/.
How to play streaming video through VLC player:
To download streaming video via VLC player, you need to convert it:
Warning: this download method no longer works with most YouTube videos.
What's the price of all this stuff?
It costs nothing. Our services are absolutely free for all users. There are no PRO subscriptions, no restrictions on the number or maximum length of downloaded videos.