1. Evolution from LLMs to Agentic AI: Complete Agentic AI Course

Added: 2026-05-09

[00:00:01]Hello everyone.

[00:00:02]My name is Bappy Ahmed and you are welcome to my YouTube channel.

[00:00:07]So guys, from this video I'm going to start one amazing playlist on my channel called end-to-end agentic AI engineering.

[00:00:17]Because if you explore my channel, there are very less content I created so far on top of this agentic AI. That's why you can also say this is the most requested playlist on my channel.

[00:00:30]My audience uh told me why not you can also bring a complete agentic AI playlist on your channel and you can cover the major part of agentic AI like how agentic AI system works and how we can implement end-to-end agentic AI applications with different different frameworks, protocols, and tools.

[00:00:49]Nowadays, if you see the market, this is the most demandable skills. Okay, especially if you're looking for any kinds of job in generative AI.

[00:00:59]Now, company wants uh you should implement a complete agentic AI system with different different tools, framework, protocols, etc. That's why I'm going to start from very basics itself. Uh first of all, we'll try to understand how agentic AI system works.

[00:01:15]What are the characteristic components, okay? Then what are the concepts are available behind this agentic AI? How AI agents works exactly, we'll try to understand each and everything.

[00:01:27]Then slowly we'll try to move towards the ad- vance part. There we'll try to cover different different frameworks like uh LangGraph, CrewAI, AutoGen, then some no-code platform as well like n8n, okay? And we'll be also learning different different tools and technology which we can uh integrate inside our AI agents to make our AI agents like more powerful and more efficient.

[00:01:52]So make sure you complete the entire playlist, guys, if If to master the agentic AI.

[00:01:57]so in this video, first of all, we'll try to understand and see the complete evaluation of this agentic AI, like how agentic AI came.

[00:02:07]And uh what we used to do in our traditional generative AI application.

[00:02:12]Uh first of all, I will give you the entire understanding how agentic AI came, what are the things they have introduced.

[00:02:18]Then uh I'm going to discuss about uh the detailed understanding of agentic AI, uh the characteristic of agentic AI, different component of agentic AI we'll try to understand with a good example. So, before I start today's agenda, I want you to uh support my channel.

[00:02:38]So, if you haven't subscribed yet uh on my channel, guys, please try to uh do subscribe and uh share this video with your friends and family.

[00:02:46]And if you like this content, please uh hit the like button, okay? And if you have any kinds of questions, you can feel free to comments in the comment section.

[00:02:56]So, guys, you can see on my screen, here I have already written the definition, like what is agentic AI.

[00:03:04]So, if you see here, uh agentic AI is nothing but uh it's a type of artificial intelligence that can take up a task or goal from a user and uh then work towards completing it on its own with minimal uh human guidance, okay? And it plans, takes at actions, adapts to change, and seeks helps only when necessary.

[00:03:33]So, by this definition itself, I think uh you are getting little bit of understanding what I'm trying to say.

[00:03:41]Um those who are already familiar with uh ChatGPT or any other uh agentic AI system, uh if you have already used uh like uh um VS code then anti-gravity, cursor AI, Cloudy Desktop, right? So, these kinds of application if you have already used, so there you will see that whenever user gives any kinds of prompt, right?

[00:04:07]Based on the prompt, that application decides what to do. Let's say if you're asking a very simple questions. Let's say you're asking, "Tell me about Python."

[00:04:17]So, most of the large language model have been trained with lots of data, okay?

[00:04:24]Especially whatever data we are having on the internet, so they have used those data and they have trained those at the model.

[00:04:32]And every model is having a knowledge cut-off, okay? Every model is having a knowledge cut-off. Knowledge cut-off means a specific date till they have trained the model. Let's say if I'm talking about uh chat GPT or let's say GPT 3.5 turbo or let's say GPT-4, you will see that those model probably they have trained uh uh on the year 2022 or 2023 around, okay?

[00:05:00]Uh till the date they have taken all of the data from the internet and they have trained those at the model.

[00:05:06]So, whenever I'm asking about the Python, so definitely in 2022 or 2023, this information was available on the internet and definitely our large language having these kinds of knowledge, right? So, it will be able to give you the answer in short or directly.

[00:05:24]But whenever I'm asking anything which is latest, let's say I'm asking um I'm asking a latest information. I'm asking like "Tell me about like the latest news uh of Iran and USA, okay?

[00:05:39]War in 2026." So, that time definitely if you are using a single large language model, okay? This kinds of language model won't be able to give you the response, okay? It will tell, "I don't have enough context, okay? After 2022 or 2003, so I can't answer your questions, okay?" This kinds of I think you will get the answer. If you have used the older ChatGPT, I think you are getting what I'm trying to say.

[00:06:08]So, but if I'm talking about nowadays, whatever application we are using like Claude index stuff, then antigravity, cursor ID, if I'm giving any kinds of prompt, let's say I'm telling just try to implement application for me. Let's say implement Python game for me. So, what it it will do, it will try to take that prompt as a command, and it will automatically, let's say, plan for a task like what to do, okay? How it can implement the entire game for you. So, to implement a game, first of all, it has to create the environment, it has to add the requirements, it has to create the user interface, it has to make the character, okay? So, one by one, all of the plan would be sorted, then once all the plan is ready, okay? It will execute the plan one by one, and it will complete the entire system, and definitely, in between, it will try to test that particular let's say application, okay? If it doesn't get any kinds of bugs, it will continue, and it will complete that particular work for you, okay? So, that means everything is happening automatically. And sometimes you will see that in between it will ask a human interaction, it will ask for a human input. Let's say, whenever it will try to implement a game, that time it might ask you what kinds of color you want for this particular environment, how many character you want in this particular game, what would be the, let's say, car color, what would be the car speed, okay? So, sometimes it will ask some kinds of questions to the human. Okay, for the guidance and once we'll try to provide the feedback or let's say our input, it will take that input again. It will try to continue the workflow. Okay, so that's why here you can see it is telling with minimal human guidance. Okay, not not complete human guidance. We give like very minimal human guidance here and it try to plans takes action. Okay, adapt to changes. Let's say it has to let's say it has to do one particular changes in the environment or let's say color or let's say any character, it will automatically do that. Okay?

[00:08:17]And it seeks help only when necessary.

[00:08:20]So guys, before I give you the entire discussion on this agentic AI, first of all, I want to walk you through the fundamental concept of generative AI application. Like so far whatever application uh we usually uh create. Okay, and how these agentic uh AI or let's say AI agents came into market. Then we'll try to understand this agentic AI concept. So for this guys, I'm going to take you on my whiteboard and there we'll try to discuss each and everything.

[00:08:49]So guys, I'm inside my board. So here I'm going to write down each and everything.

[00:08:54]So see, whenever I'm talking about uh AI agents, right?

[00:09:02]AI agent.

[00:09:07]So this is the application of generative AI.

[00:09:14]Okay?

[00:09:15]This falls into generative AI domain.

[00:09:18]And those who are already working with generative AI, so I am having a dedicated course on my channel. The complete generative AI course. So there I have already discussed the foundation of generative AI. So in that course I have already taught you all the concept regarding generative AI large language model, okay. Retrieval augmented generations. So, each and everything I've already covered there.

[00:09:42]So, if you are not familiar with generative AI, first of all, try to complete that particular course, then it would be easy for you to understand, okay?

[00:09:50]So, in generative AI, the main component we usually work with large language model.

[00:09:57]Okay? Large language model. So, there are different different large language model nowadays. I think you will know um there are some organization, there are some company they have launched different different models. If I'm talking about Meta, okay, Meta AI, so they have launched something called Llama.

[00:10:19]Okay?

[00:10:20]Llama.

[00:10:21]Then, if I'm talking about OpenAI, they have launched GPT.

[00:10:28]Okay?

[00:10:29]Then, we are having Mistral.

[00:10:34]We are having Gemini.

[00:10:39]Okay, this is from Google.

[00:10:41]So, that's how we are having different different large language model nowadays we usually use.

[00:10:47]So, previously, whenever we started generative AI, that time um we used to uh only use a fine-tune fine-tune based or let's say pre-trained based large language model. Let's say, here I may having a large language model.

[00:11:05]So, we used to provide a prompt.

[00:11:10]Okay, prompt.

[00:11:11]And it used to give a response.

[00:11:16]Okay, response. So, basically, we used to use this large language model for text generation, okay? For very good quality text generation.

[00:11:25]Or for some other task also we used to use like for language translation.

[00:11:36]Okay.

[00:11:38]Then for text summarization.

[00:11:45]Then definitely for chat operation.

[00:11:49]I used to ask different kinds of question and we used to get the response.

[00:11:54]Then definitely for like some other NLP task like NER and so on. Okay.

[00:12:04]So initially those who have already used this chat GPT, I think you are trying to relate the concept what I'm trying to say. It was like a very basic application. Okay, we used to use for these kinds of text generation task.

[00:12:17]But slowly what they did, they actually introduced also image generation.

[00:12:23]Okay.

[00:12:25]Image generation.

[00:12:30]So image generation happens whenever they introduce something called multimodal system. So in the multimodal, you not only generate the text, that you can also generate the image. Okay.

[00:12:41]But what was the problem with these kinds of application? As I already told you, let's say if I'm talking about any kinds of large language model, it is having a knowledge cut off. Okay.

[00:12:52]This is having a knowledge cut off.

[00:13:00]So what is knowledge cut off? Let's try to understand. So for this let's go to the Google.

[00:13:05]And here if I'm searching for any kinds of model, let's say I'm searching for OpenAI models.

[00:13:11]Let's open up the models.

[00:13:15]Now let's pick any kinds of model from this OpenAI. So let's say if I'm talking about the GPT-4 5.4 mini or let's say if I'm taking any older model.

[00:13:25]Older model. Yeah, so I think Here some models that are available.

[00:13:30]Let's see if I'm talking about this um the P1.

[00:13:37]Let's see if I'm talking about this GPT-4.1. So, if I click on this model, you will see that this model having a configuration. Configuration means the context window. Like uh how much context it can take. Then maximum output tokens, how much token it can generate. And there is a section [clears throat] called knowledge cut off, okay?

[00:13:58]So, here the knowledge cut off you can see January 1, 2024. That means this model uh has been trained uh till January 1, 2024 uh internet data, okay? So, if you're asking anything after that, let's say if you're asking February 1, 2024, definitely um this model is not going to give you the response because this model doesn't have uh the knowledge after uh January 1, 2024, okay?

[00:14:28]Whatever, let's say uh recent update uh we are having on the internet, this this model doesn't know about that. So, the main problem, I think you can understand. Let's say if my prompt is uh before, okay? Before this particular knowledge cut off, the information I'm looking for on uh from my large language model, definitely this model uh can give you the response. But if it is after the knowledge cut off, that time it will not able to give you the response, okay? It will tell, I don't have the um context, I don't have the informations, okay? After this particular knowledge cut off, so I'm extremely sorry for that. So, that that is the uh things actually uh uh happened uh whenever ChatGPT came, okay? Uh initially in the market, and I think you remember, okay? Uh ChatGPT used to give these kinds of response.

[00:15:18]Then uh what uh they have introduced, they have introduced a concept called rag, okay?

[00:15:25]Why they have introduced the concept called rag? Because now let's say if I want to add some other information, let's say this is 2026.

[00:15:33]So now I have to I want to add some more information, okay? Inside my large language model. So what I have to do? I have to fine-tune this model, right? I have to fine-tune this model.

[00:15:47]And fine-tuning means we are taking the pre-trained model, okay? And on top of that we are adding some new data.

[00:15:57]Adding new latest data and we are training few parameters here, okay? And whenever I'm talking about the LLM parameters, it will count like from million, right?

[00:16:07]Million to billion.

[00:16:10]Okay, this is the issue. So fine-tuning is not an easy task. For this you need a good resources, then good budget, okay? Then you you should have also time.

[00:16:20]If you're having these kinds of things, then you can easily fine-tune one large language model, okay? There is no issue with that.

[00:16:28]So for the company, this fine-tuning task was easy because they are having a good resources. They are having like very heavy investment, okay? They are having lots of time, so they can do that. But what about for the developers? Let's say if I'm creating a application, okay? For my client. And if any new data is coming and I want my application to be aware front of of this new data. So for me, for for me as a developer, this is this is going to be like very hectic task, right? For fine-tuning a model. Because I don't have these kinds of super computer with me. I don't have this much of budget, okay? So that I can purchase a good cloud for the training. I don't have that much of time time so that my client will wait for me because they has to also do the business, right? If I'm running my business also, this should be continuously running and I should have handled all of the client with latest information and everything. Okay?

[00:17:22]So, that time researcher introduced something called rag concept. Okay, this is called retrieval Okay, retrieval augmented generation.

[00:17:40]Okay? RAG rag component. In the rag component, concept what we used to do, let's say we are having a large language model. This is completely fine.

[00:17:50]Let's say we are having a large language model.

[00:17:57]Okay?

[00:17:59]So, it will be connected to a knowledge base.

[00:18:04]So, knowledge base is basically a database.

[00:18:08]Okay, it's a vector database.

[00:18:11]So, this is called knowledge base. So, it is having all the latest information.

[00:18:20]Latest data, I can say.

[00:18:25]Okay? So, this data you have to store in the knowledge base, and you have to connect with your large language model.

[00:18:33]Okay? And for this kind connection, we use our orchestration framework. Some orchestration framework, I think you will know inside generative AI. We are having LangChain, we are having LlamaIndex, okay? So, this is called orchestration framework. So, we use these kinds of orchestration framework to make the connection with our LLM.

[00:18:51]Okay?

[00:18:52]Now, if user is asking anything, Okay, let's say user is giving input.

[00:18:57]Okay? First of all, this input would be verified in the pre-trained model. That means the large language model itself. First of all, it will try to check whether this information he's asking, what kinds of question they are asking, it is available in the LLM itself or not. It is available in this knowledge cut off or not. If it is having, okay, in this knowledge cut off, this will give you the response.

[00:19:25]Directly, this will give you the response.

[00:19:27]Okay? This will give you the response directly.

[00:19:30]But, what about this information is not available? That time, it will go to the knowledge base. It will go to the knowledge base. Okay, it will do something called semantic search, similarity search. This will get the relevant uh result about the questions user is asking.

[00:19:48]Then, this particular relevant answer again, your large language model will take. It will try to analyze.

[00:19:55]It will try to um it will try to clean up. It will try to rearrange the uh response. Then, it will try to send it to the user again. Okay? That's how the entire RAG system works.

[00:20:09]Okay? RAG system works. So, basically, the major component we have added this knowledge base.

[00:20:14]And adding data in the knowledge base, it is super easy because only you just need to uh fetch the latest information and add in the knowledge base.

[00:20:24]And uh your LLM is already connected to the knowledge base. So, anytime if you're asking any kinds of question, it will uh bring that particular latest information and uh it will do the refiling operation, then it will pass to the human. Okay? So, this will uh work like that.

[00:20:39]Okay? And this was the like uh very famous technique uh that time. Even nowadays also use the same technique we we create the RAG application.

[00:20:50]And this actually helps us uh from this fine-tuning operation.

[00:20:54]Because here we are not doing the fine-tuning, okay, on our LLM. Only we're just working on the knowledge base. We are adding the data in our vector database. This is the things, right?

[00:21:04]But, there are some problem with this vector database or this RAG system. What is the problem?

[00:21:10]Whenever I'm talking about the real-time data. Real-time data means the data is continuously changing. Let's say if I'm talking about weather information. If I'm talking about temperature, if I'm talking about the latest news. Okay, it is continuously changing. That time it is not possible for me to sit down whole day and take all of the latest information and like add in my knowledge base. Okay, that that kinds of things we can't ever do that. So that that is why this RAG system fails. Let's say if we're asking uh questions to the RAG system, let's say tell me about latest news. Okay, right now in the morning. So definitely, this information is not available in the knowledge base. Okay, let's say a morning news you have added, but what about the afternoon news? What about after 1 hour news? Okay, so this kinds of things you don't have. Okay, so that time your application won't be able to give you the response. So what you have to do that time? You have to uh you have to think about a different approach. So that's why researchers thought, why not we can create a agent?

[00:22:16]Okay, why not we can create a agent? So that agent will be connected with some tool. Okay, tool means we can use different different tool here. Uh let's say we can use any kinds of search tool, we can use any kinds of storage tool, we can use any kinds of calendar tool, Google Drive tool.

[00:22:33]Whatever we can use, but there should be some kinds of tool. So with the help of that particular tool, my AI agents will try to fetch the information and it will give to the user. Okay, so what they introduce? That time they introduce uh agent system. Let's say this is your agent.

[00:22:53]Okay, agent.

[00:22:54]Uh internally it is using a large language model only. Okay, whenever user is giving any kinds of input, it is connected with some kinds of tool. Okay?

[00:23:03]So, let's say if I'm asking for uh uh any latest information, that time it is connected with a search tool. Okay?

[00:23:09]Internet search tool. So, mostly this will search on the Google. And Google is continuously updating. Okay? With latest information. So, if you're asking any real-time question, first of all, what it will do, it will um use this search tool. It will search over the internet.

[00:23:23]It will get the information. Okay?

[00:23:25]Latest information. And your agent LLM is trying to refining that. And it is giving you the response again. Okay? And why I'm calling this particular system as a agent? Because your application is smart enough to understand when it needs to call this call this tool, whe- when it doesn't need to call this tool. Okay?

[00:23:45]This kinds of uh reasoning capacity your application will be having. Okay? That's why we call it as a agentic agentic system. Okay? Here, we are not deciding when to call this particular tool.

[00:23:57]You just give the prompt. Okay? To the application application application will decide whether I has I have to call this tool to get uh give the response or I have this information with me. So, I can give you the response. Okay? So, this kinds of capacity it was having. So, this was the first agent they have introduced with some real-time tool.

[00:24:19]So, if I uh take you to the ChatGPT, so let me give you the example. So, I'll open the ChatGPT.

[00:24:27]And uh here, let's say I'm asking a question. Let's say I'm asking, "Tell me about Okay?

[00:24:39]Python."

[00:24:41]Now, see. What will happen?

[00:24:44]Uh this is directly giving you the answer. Okay? It is not referring any kinds of tool search tool. It is not searching on the internet. Okay? Instead of that, what it is doing, it is giving you the direct answer. Okay? Because this information is already available in the knowledge bit itself. Okay? Knowledge knowledge uh LLM knowledge itself. Okay? Because it is already uh having uh before the knowledge cut off.

[00:25:11]You get it? But whenever I'm searching for any other question, let's say I'm telling tell me the latest news latest news of India election.

[00:25:36]Now, if I search that Now, see it is searching for web. Okay?

[00:25:40]It is searching for web. It is using our internal search tool and with the help of that it is searching over the internet and it is referring some trusted uh let's say sources like Al Jazeera ABC News. Okay? That's how it is searching on different different website. Okay? Now, if I open this website, you can see that this is website. This is another website. Okay?

[00:26:04]And this website has already this kinds of latest news. It is bringing that particular informations. It is passing it to the LLM. LLM is trying to refining. LLM is trying to summarizing all of this let's say uh all of this content of this kinds of sources and this is refining and giving you the answer. Okay? Refined version of answer.

[00:26:28]Okay? So, this is called actually uh agent system. Okay? It is utilizing some kinds of uh tools in the back end and this application is automatically deciding when it needs to call that tool when it doesn't need to call that tool.

[00:26:45]Okay?

[00:26:46]Not only that, this is a simple example I have shown. If you have already used uh like anti-gravity, Let's say if I open up my anti-gravity.

[00:26:57]So, this is my anti-gravity. So, here I can uh give the prompt to the agent. So, let's say here I'm telling um create a car racing game using Python.

[00:27:20]Okay, Python.

[00:27:22]Now, here you can um select different different model because internally I told you agent uses a large language model. Okay, because this is the brain.

[00:27:31]Okay, it is having the reasoning power.

[00:27:33]And it decides actually when to use the tool, when uh it doesn't need to use the tool. Okay? So, here you can select different different model. So, anti-gravity supports these are the model. You can select any of them.

[00:27:45]Now, if you give this prompt, you will see that automatically, first of all, it will try to make the plan. Okay, what to do.

[00:27:52]Now, see it is telling generating. Let's wait. Now, see it is thinking. Okay? It is thinking. Now, it is trying to making the entire plan for you.

[00:28:02]Okay, how it is going to uh create that particular racing game with the help of Python. What are the resources it need? What are the tools it needs? It will try to make the entire plan. See, this is the plan.

[00:28:14]You can see this is the plan. Okay, proposed plan.

[00:28:17]Now, what it will do in the plan? First of all, it will do the initialization.

[00:28:21]Set up the Pygame display front and clock. Then player car, obstacle, uh collision detection, score system, game over screen. Okay? Then what are the requirement it needs? Verification plan.

[00:28:36]So, this is the agent plan, guys. That's how one agent works. First of all, it has to make a plan. And based on the plan, it will start working on that.

[00:28:46]Okay. Now, I told you in the definition itself uh it will seek for help when it necessary. That means little bit of human interaction is also needed. Now, this plan is proposed to me. Now, I can review the plan. Okay, I can make some changes. Okay, so let's see if you want to change anything. Let's say you don't need this particular step. You can change anything. Okay, you can change anything. You can edit anything.

[00:29:12]Okay. Then, you can review it. You can like tell, "Okay, this plan is completely fine for me. You can continue." Now, let's see if I do uh uh the plan is fine.

[00:29:27]Go ahead.

[00:29:30]Okay?

[00:29:32]Now, if I give the prompt I think prompt uh you can't see because this is uh just uh beside my image.

[00:29:41]But, I think you can see, okay? Uh the prompt I have given.

[00:29:45]Now, see. Now, it has started working on the plan. Okay, one by one, it will work on all of the plan, okay? And it will try to implement the entire game for you.

[00:29:56]Okay? So, this is called AI agents nowadays. So, AI agent is like uh this is not uh I mean um I mean uh restricted to the tools only.

[00:30:09]Okay, now it can automate the workflow.

[00:30:12]This is called automation, right? Here, I'm not writing the code. See, my agent is writing all of the code. And it is asking for the approved. Okay, if I show you if I let's say show you, so here you can see it is telling, "Do you want to run this pip install command?" So, it is asking for human interaction. Now, if I give the human interaction, if I give if I tell yes just do it.

[00:30:35]If I tell, "Okay, I accept the code."

[00:30:37]Now, the rest of the task it will automatically do that for me.

[00:30:41]Okay? So, this is called AI agents. Now I think you have understood this particular definition. Now let me show you the definition once more time.

[00:30:50]So this is the definition guys, okay? So here you can see agentic AI is a type of AI that can take up a task or goal from a user. So the here the task and goal I have given just create a car racing game with the help of Python. So then what it will do it will work towards completing this particular task is on with minimal human guidance. First of all it will try to plan, okay? Then it will take action, adapt the changes, okay? Let's say whenever it requires any kinds of changes it will automatically do that and seek help when it necessary. That means it will ask for my help, okay? If I want to change anything, so it will ask for that particular help for me. It will ask for ask for my feedback, okay? If I give the feedback it will start working on that, okay?

[00:31:34]So I think guys you have understood the entire uh evaluation of this agentic AI, how this agentic AI came, okay? Right now in the market. Now in the next video guys what I'm going to do, I'm going to discuss this agentic AI in detail. Okay, the application working mechanism, okay?

[00:31:57]I'm going to show you one example like how one agentic AI application works whenever we give any kinds of command, okay? We give any kinds of prompt. I think you have seen all the anti-gravity we given a prompt and it creates the plan. After creating the plan what it will do, okay? Each and everything I'm going to give you. I'm going to tell you the characteristic of this agentic AI, what are the characteristic it follows, what are the component it is having, okay? So with a good example we'll try to understand the entire concept in the next video. So yeah, this was this was only the understanding like about this agentic AI evaluation like how this agentic AI came in the market and whatever traditional application we used to create in the GNI, uh nowadays people are uh actually um people are moving to the agentic protocol. People are moving moving to the workflow automation. Instead of creating the simple uh actually text generation based application. Because right now uh everything can be automated. All the workflow can be automated, okay? Instead of working on manually, we can create uh agents. And they that agents will try to complete that particular task for me. That's how you can also scale up your business. You can uh you can actually uh uh create some agents for your business.

[00:33:14]So, it will run automatically. Let's say customer support agents you can create, okay? Automatically uh email sender agents you can create.

[00:33:21]So, that's how you can minimize the uh employee in your company. And you can uh save your budgets, okay? But let's say if you don't have these kinds of agentic AI system, that time uh what you have to do, you have to hire someone to do that particular that task, okay? That's why companies are uh adapting this AI agents in their uh application development, in their uh workflow automations, okay? They're replacing some low-level employee uh which uh they feel like, "Okay, I don't need these kinds of employee and I can do these kinds of work automated way, okay?"

[00:33:54]People are uh thinking in that way.

[00:33:57]Okay, you have to also be smarter. Now, people ask like uh whether we'll have the job or not, okay?

[00:34:03]Definitely we'll have the job, but you have to learn these kinds of technology.

[00:34:08]If you know these kinds of technology, then tell me, who will replace you?

[00:34:12]But if you don't know this technology, let's say still you do the Excel uh let's say uh data collection automa- uh data collection, let's say strategy. Now tell me, I can easily create a agents and I can do the Excel data collection.

[00:34:28]Okay? I can um easily handle the customer automation. I don't need someone to handle my customer. Let's say whatever customer are coming to my website, okay? I don't need to like hire someone to sit and reply for that. So, what I will do, I'll just create a agents. I'll give all of the informations about my website, all of my services, my agents will take care everything, okay? This is the things nowadays people are moving, okay? So, yeah, trust me, guys. This particular skill is having high demand in the market. So, if you can master this one, definitely you can you can get lots of opportunity, okay?

[00:35:04]And I will try to complete this agentic AI in a such a way so that after completing it, you can create any kinds of agentic AI applications. So, yes, guys, this is all about from this video.

[00:35:16]I hope you liked it. And if you have liked it, guys, so please try to subscribe to my channel and hit the like and just let me know your opinion, okay? Nowadays how much this agentic AI is required and whenever you are looking for any kinds of jobs, so I think you are seeing this particular agentic AI concept in their job role, okay? And let me know your difficulties for sure, what can I do so that your difficulty can reduce and what kinds of let's say topic framework you need, okay? Just do let me know. Definitely I'll try to cover. And let me know in the comment section because comment section I usually check and I'll try to reply you and definitely I'll try to take your feedback, okay?

[00:36:01]So, yes, guys, this is all about from this video. Thanks for watching. In the next video, guys, we'll try to understand this agentic AI in more detail. Thanks for watching.