Generative AI vs AI Agent vs Agentic AI — Explained with Claude Code & Codex

What if I told you that I created an e-commerce website? in which we have front-end, back-end, database, API, admin panel We tested it, fixed the bugs, and posted it on Hostinger BPS. without me touching a single line of code. Yes, guys, not a template, not a website builder actual production code written, tested, and deployed by AI. Yes, guys, with the help of Artificial Intelligence AI. Today, in this single journey, I will tell you about four terms. which people confuse or are not familiar with. That's your LLM, Generative AI, AI Agent, and Agentic AI. So, I will explain these four terms through a single mission. What are the differences between them? By the end of this video, I guarantee you that you will never be confused about these four terms. akashic records, the mind. First of all, before anything else, we will first establish our foundation. And that is LLM, Large Language Model. Before understanding this, I would like to share with you A mini version example of this. And that is, whenever you type something on your mobile Let's say you are sending a message to someone on WhatsApp. Then you have noticed the suggestion for the next word at the bottom. Now, the suggestion for the next word is entirely based on your previous chat history and also the linguistic patterns that have been taught to them And it's very limited. Now, an LLM is that same concept taken to an insane scale. Here, the LLM is trained on an extremely large collection of text, code, and other data. Now, exactly their training sources depend on their models. and based on all those models, whenever you ask an LLM anything It will predict the next most suitable token. Here, I said "token" and not "word." Why? Because this token could be a full word, a complete word or a part of a word, a character, or even a punctuation mark. So, basically, it generates a token for you. then the next token, then the next token In this way, you get a complete sentence. You receive a complete answer. and if you have requested a code Then that code also generates it for you. That's it. Added score. This is your LLM. And all your AI models use LLM as a brain. I hope you know now. It's a very powerful next token predictor. You can call it that. Part 2. The mission. Now, guys, let's move on to the mission. So, what is the mission? Let's see that first. Build a complete e-commerce website. Front-end, back-end, database, REST API, admin panel and host it live on Hostinger BPS without me writing or touching a single line of code. To achieve this, I am going to use Cloud Code. If you have an OpenAI Codex, you can use that as well. It's the same concept but different tools. And you will watch this video very carefully. because as we move forward on our mission our concepts such as generative AI, AI agents, and agentic AI will also become clear. Yes, guys, just one mission. Today I will clarify your doubts regarding these three terms. Part 3. Generative AI Deconversation. Now, guys, step 1. Planning. When you go to build any project, Your first step should be planning. Yes, guys, you have to plan first. And see what I am going to write for that plan in our Cloud Code. I want to build an e-commerce site. Here are my requirements. I'm considering this tech stack for the back-end. I have a VPS hosting. Help me plan this out. And that's it. This prompt, although I gave a very simple prompt, We could have written it better. But for example, when I write this prompt to my Cloud Code, It will ask me in response Howe payment gateway do you prefer? How many product categories do you want? Do you want multi-vendor support for this? So what's going to happen this way, guys? It will ask me questions. I will give my answer. Sometimes it will ask me questions. It will answer me. In this way, we will establish a solid plan. But there is one thing you must keep in mind. And that is, during this conversation, I asked something in the form of text and in the form of voice, It answered me in the form of text. That is, text in, text out. So what it is doing is It is generating the content I asked for. So this is your generative AI. Yes, here it is not doing any extra work. Like creating a file, Opening a terminal and issuing a run command. It's doing nothing. Whatever you are asking it, It is simply providing the answer in the form of text. And that is generative AI. You can also understand it in this way. A brain of an LLM generating content or output based on your input. Now, let me tell you an example. So Chad GPT, your Claude, Gemini, when they give you simple text Whatever you ask them, that's it. This is your generative AI. And remember that in this variant, They do not have access to external tools. or real-world actions. Just remember this much. Where they are simply generating text, image, video, or audio for you. Whatever you asked them, that's what it is. This is your generative AI. Part 4, AI isn't. Alright, everyone, we've completed our planning. What is the next step? Tell me. If you have completed a good project in college or in a company, then the next step is that now we have to create diagrams. There can be flow diagrams, ER diagrams sequence diagrams architectural diagrams So we have to create those. so that all our entities, tables of our database or whatever it is We find out about it. Alright. So we have made a solid plan. Now see what prompt I will write. I will say, This plan is great. but I want a proper flow diagram and an ER diagram for the database Before we write any code. Now, this is it. The moment you write this prompt, Here is the problem. And the problem is, Cloud Port will create a diagram for you. but in the form of text which you will have to copy later and paste it in the plugin to obtain the actual rendered image. Now, a pure generative AI Cannot create a proper diagram and neither can it create it and save it in the file. Here we will need an external tool with the help of which Cloud Port can create a diagram for us. And there will be many such tools, but the most popular one is Mermaid MCP So finally, in Cloud Port We will connect our Mermaid plugin. or MCP. And that's it. Now, based on the prompt we gave, It will create a flowchart. sequence diagram ER diagram for you. But here is one thing that you must keep in mind. If you are a generative AI, like ChatGPT or Cloud, if you are connecting MCP tools, and now that AI is giving you an output with the help of that tool, It doesn't make it an agent. Yes, guys. We will call it a simple tool. You have just granted access to a tool. Used the tool and that's it. Now, when is it actually called an AI agent? Actually, this AI agent is called when Now you have multiple MCPs. and you gave the same prompt that I want a diagram ER diagram, flow diagram based on the project that we have planned. Then here, your AI Your Cloud Port or ChatGPT generative AI It will now think. It will decide. Based on that, It will act. Out of so many MCP, Which MCP should I choose? so that I can make this diagram? So there it will select Mermaid MCP. So here it is, deciding, It is acting. Based on that It will observe. Once it has created the diagram, whatever we were planning Based on that, It will observe. It will check whether it is correct or not. In case it is not correct, Then again, it will recreate it. and then observe. And now if it sees, Yes, now it is absolutely correct. It will render it. and save it in our file. Yes, guys. Here, any AI, You can finally call it an agent. when it learns to use loops. As you can see here, when it learns to decide What will be the next step? then reason then act, then observe, then repeat and it will keep working until it reaches the goal that we gave it. And that is your AI agent. LLM brain plus tools plus a feedback loop. Yes, guys. So all this together makes you an AI agent. I hope you know now. Generative AI was producing normal text for you and nothing extra. The AI agent is carrying out the task. that you are giving. It's like a skilled worker. Who is doing the work for you? Part 5 Agentic AI the owner Now guys, In this mission, We have almost reached to the final stage. And you had to find out about generative AI and you also got to know about AI agent One thing, What is an agentic AI? Now here, at the end of our mission, We have come, So this will also be clear to you. Now I am going. Okay, I am handling Above everything, I give it all to you. You decide. Database first or backend or frontend. Build all of it. When you are done, write test and run them. If the test fails, Review your own code. Fix it. and test again. And when everything has passed, I securely connect Hostinger MCP using a scope and revocable API token Define what it is allowed to access and keep approvals required for the critical production actions Deploy it. Make it live. Yes, guys, Now, after giving this prompt, We just have to relax. and sit. And just keep watching. We don't have to do anything extra Now if you see, After this prompt, What will our AI do? Hey guys, First of all, It will think on its own. That's what can be The first step. Most probably, it will create your database schema Then it will build your API Then it will go to the frontend. Then it will be completed. backend like this. After that, guys, It will be tested. It will write. unit test and then test. And let's say, three errors Came. Like your API Got mismatched or inventory Did not update properly or Admin URL Got exposed. So first of all, What will it do? It will read the error. Now it will try to fix it. How can we Fix it? And once it Fixes the code. then again it will test. And this time, if everything becomes successful means all the tests get past then it will go to Hostinger BPS to deploy it. And then, We have already to our Hostinger BPS. Again, with the help of connector, That is your MCP. It will easily our website on Hostinger BPS Yes, guys, Our mission is complete. But there is An important thing which you have to pay attention to And that is in agentic AI everything AI is giving us A to Z from start to finish But it doesn't. mean that You just sit. and relax and you don't need to see It's not like that. Again, This is a very crucial step And that is that there will some parts some moments where you have manually Give approval to yourself For example, when it comes to changing the DNA, or migration or replacing live production or deleting something. So these are Such important steps Where, guys, You can't blindly empower In auto mode. You have to Review it manually. Okay? And don't worry, in agentic AI manually sometimes in such critical conditions giving approval also comes in agentic AI Agentic AI does not mean that It's doing Everything on its own. Obviously, It's doing. But there are some steps where it is important for you to be there It's important to give approval Yes, guys, This is yours agentic AI where if you notice the same AI now we're looking at Cloud Code Cloud Code normally if it's just generating a text for you now you'll say, this Cloud, simple chat this is my generative AI even ChatGPT. Then in the same Cloud if you give external tools access to it, and draw an image like It thought, which one Should I If it's wrong, how should I Fix it? This is yours agentic AI skilled worker One task, Done. Now, the same AI is also known as agentic AI because of its behavior. And what Is that behavior? That behavior Is that it? the same AI has the power to plan the multi-step tasks by itself. For example, It can plan. 10-12 steps by itself. It can also Do it. After making it, It can also decide. that first I'll do this. After that I'll do this. And again, if there are Some mistakes, It can correct. And this behavior makes an AI agent into a agentic AI Yes guys, As I told you, in one mission and in the same mission I'll clear it. All your doubts. I hope your doubt has been Because your Claude It's simple. Your Claude is also known as generative AI Your Claude is also working as an AI agent and also showing you the behavior of an agentic AI I hope now you Any doubt. I hope now finally LLM, generative AI AI agent and then agentic AI the gap between differences is now Clear. So, That's it. Guys. That's all for this video Please comment and tell me whether your doubt was cleared or not. And please Like this video. Our likes target is 1,000 likes. Share it with your Because in today's time these three terms are Very important and everyone needs to the differences between them I hope You finally Got to know. If you liked Please subscribe to the channel See you in the next video Until then, Thank you so much Be careful.