21 Powerful Use Cases for Claude 3.5 Sonnet (FREE)

All right, so as you probably heard, Claude 3.5, sonnet is freely available with limited usage and it beats ChatGPT 4.0 across many categories. The one that is absolutely undebatable though is code generation. And not just code generation. Their Artifacts feature allows you to see that code in action as it generates it, making this an unprecedented development in the AI space. What does this mean? Practically? You can create websites, you can create games, you can create visualizations, and so much more. There's a ton of use cases and I actually really took my time with this video and researched this across the past week since Sonnet came out and now compiled all of them for you in one video. The beautiful thing here is though, as opposed to the GPT4.0 use case video that I did, a lot of these use cases you can try, you can replicate, you can use for yourself today. Because Sonnet 3.5 and its capabilities are not some sort of announcement or promise of things that are yet to come. It's available today and we'll be looking at all the various ways you could use it today. If you want more details on the model itself, I created a separate video for that. This video is going to be all about how you can put it to work, which is going to be a combination of examples of how other people put it to work, and prompts that I crafted myself. Some simpler, some a little more intricate. All of the prompts that you'll see here I'm providing to you for free in the description of the video. If they're too long in the description, I'll probably just have to link this notion page as you can freely access just like a website. Okay, so let's get into it. There's a lot to cover and I could have actually quite easily made a video titled the 50 Most Mind Blowing Use Cases. But there's the thing is, not all of them are created equal. Some are more impressive than others and some are more practical than others. So I use my personal judgment and experience to pick the use cases that I figured you might find really interesting here. Because we're going to be visualizing data as an animation. This is something the code interpreter, aka the data analysis tool in GPT4.0 cannot do. So namely, the prompt I'm using here is act as a data visualization expert, create a real time dashboard to effectively represent the spread of a disease over time across various countries. And this is one of the first capabilities I want to highlight. It's really good at creating these dashboards type websites. We'll be Looking at more examples shortly here. But look, from this simple little prompt, it writes the answer and then here you have the artifact that essentially is all of this code. But you don't even see the code in the first view, it just writes that in the background and it gives you this interactive preview. And you can look at the little app. Look how great this is. I didn't even give it any data. It just came up with some mock data for me and I can switch between the different countries. And it has several features like it simulates real time updates. So you could hook this up to API with code that provides it with real time data and it will be updating as it goes. Quite simple, right? But if we go further in this conversation, I'll present you with by far my favorite follow up prompt here. It's so simple, yet so effective. By simply saying make it better, it will interpret what you mean by better. It will look at the context so far at what it created, and come up with a list of iterations that would make it better. In Claude's eyes, that is. But it turns out it's really good at figuring out what would make it better. So by simply saying make it better, it came up with this dashboard with way more information. At the top you have the base data, it added a bar chart option and there's a global map view. And here we get to the first important caveat. This will write code for you that might not be executable within the artifacts feature, but if you bring this over to something like Visual Studio code and run it, you can have additional packages that are not available in here. Just as a recap for non technical people, which by the way, is what I assume for this video. You don't need to know code to follow or recreate these use cases that I show you. A package is a set of functions and instructions that do a specific thing with code. Usually they're more advanced functions that by using a package you don't have to bother with. So in this case it's a package that represents maps of the world really well. But this package is not available in here yet. Claude will still create this for you because it assumes it'll probably take the code and host it elsewhere. This is just a preview, right? So this is something to be aware of because sometimes the more advanced prompts that I'll show you will use packages that are not accessible in here. But if all you want to do is stay in here and use the artifacts feature, you can absolutely do that. We'll talk about the nuances of that once that comes up. But as you can see by simply saying make it better, it really improved this dashboard. It added this global map view that is not accessible in here because it's a package that's not in here as I outlined. And that's the very first use case. But it doesn't stop there. I actually went in and used some of the prompt engineering skills that I've been teaching on this channel and I crafted a prompt that helps improve some of these use cases. And the result of it is this. It's this larger prompt that goes into all the details of the data visualization that I'm looking for. This is something similar to Sam, the prompt creator that I showed you on the channel about a week ago. But this is for the use case of creating projects using artifacts in Sonnet. So what I'll do in this video is I'll give you all the use case prompts. But this improvement prompt, we're refining that in the community. We're actually hosting a weekly challenge right now where all the community members get to use that prompt to flesh out their own use cases. So in the spirit of the AI Advantage channel, I'll say what I always say is that these videos are always there to give you fish, but then the community exists to teach you how to fish so you can catch them yourself. And in this case, the fish that I'm giving you is this fleshed out prompt. So if you copy paste it into here and simply run it, you will see that it does a few things. First of all, you can you have manual control over the various features in here. Okay, so instead of having a quick one liner and it assuming a lot of things, you tell it exactly what you want. And if you're building a visualization like this that represents the spread of disease over time across various countries, then it might be really good to have manual control over these things because here you can go into the prompt and you can actually edit. You can actually see what decisions have been made rather than telling Claude, bro, figure it out yourself. And here is the result. And look, I kept it like this on purpose because here we ran into our first problem. There is an error message, right? This generated code includes unsupported libraries datefns and calendar as calendar icon from Lucid React. You don't need to know these packages. You don't even need to know what this means. You just have to know that this is an error message. And what you can do in Claude and in GPT 4.0 every time is just copy the error message and paste all of this back into it. You could also copy just the error message in the middle. But this works like a charm. Just paste it in there and and it apologizes to you and it will rebuild the app without those packages. Now you want to do this if you want to stay inside of the artifacts feature, if you want to bring this outside, if you want to host this yourself on the Internet, you don't mind. These packages will probably do something even better than you can do it here in these artifacts. But if you want to stay in this web interface, just take the error messages that it gives you, throw them back in, and there you go. Here's the improved app. You can see there's various improvements, different views. You have an option to share the page, you have an option to export the data. And all we did is lead with the more advanced prompt. And then if you want to stay inside of the artifacts feature, we just paste the error message and it readjusts it so that it works in here. Just be wary, that will usually result in worse code, objectively. So if you want to host this elsewhere, take the first version. And also the obvious question, Igor, how do I host this elsewhere? Well, this is not a technical tutorial on how to host websites or web apps on the Internet, but what I did do and in my prompt I included the blog that tells it to give you a step by step guide on how to host it. So if you look at the bottom of its output, it doesn't just give you the artifact, AKA the website that we works, but also here at the bottom it tells you to fully implement this dashboard in a production environment. You would need to add a proper date range picker, implement real API calls, add proper error handling, implement the sharing and data export functionalities. It tells you all the steps that you could then communicate to a developer or work with Claude to guide you through it. And also what you'll see in the first result, it gives you a step by step guide on how to host this yourself. So this app goes offline as soon as it closes the tab and you can share it with the world for others to use it. But if you follow this step by step guide that comes from the initial advanced prompt, I'm confident that if you take some time and use Claude's assistance in it, you will be able to get this up and running. It's not easy, but with the assistance of something like Claude, it's not that hard either. And then beyond that, I have a more advanced example of a visualization here on Twitter by Louis Batala. And it took him five minutes and a couple of iterations. That's what it takes sometimes, right? If it doesn't work, you got to follow up, you got to tell it what you don't like, what it should change. And he just visualizes a pendulum with a bunch of random numbers in between, and then it generates a bunch of other pendulums. And look, you get this visualization. I'm not exactly sure how you practically use this, but you can do things like this. And while the options might not be infinite, there's so much you can do which just wasn't possible a month ago. If you want the final prompt on the tweet, links are in the description, as per usual. All right, next up we'll be using a new chat, and this is a very simple prompt yet again. Create a narrative driven website from a PDF that weaves together unexpected visual elements and storytelling techniques to reinforce the information in the PDF. This one comes from my personal prompt library. I have various interesting and esoteric prompts that I like to play around with here and there. This was a variation of one of it and I used its ability to convert PDFs into websites and combine that with these really powerful LLM keywords like storytelling techniques and unexpected visual elements. And what do I need to do here? Of course I need to provide it a PDF, otherwise it will be a little lost. This is a PDF I shared on the channel before. It's one of now 60 or 70 AI Advantage community guides. And it's this midjourney technique where you can generate screens with green screens in them, which makes it really easy to create custom marketing materials for your very own brand. Because it's really simple to replace the green screens in, either with a chroma key effect in a video editor, or by selecting that one color inside of Photoshop, Canva or similar. Anyway, the point is that this is a step by step guide and we're going to try to visually represent this and just run this prompt and see what we get. Now this one is so interesting because the concept here is not the idea of turning a midjourney tutorial PDF into a website. It's the idea of transforming any PDF into a different medium. LLMs are essentially transformers that generate various outputs, right? To generate text, image, even video. Now, with Gen3 and Dream Machine or with Cloth Sonnet, they create web apps, websites and games or interactive visualizations. Now look, this is a very simple prompt and I just gave it the PDF without any additional Context and it creates this website for me. And as you can see, it is narrative driven. Chapter one, the discovery. And it turns my little tutorial into a story driven website. Isn't that interesting? You could enter the mid journey prompt here, generate the image. Obviously you would need to hook this up to the API to actually work with Midjourney and Midjourney doesn't have an API, so you would need to do stable diffusion or something else. But you get the point here. There's placeholder images in the background. Look at these scrolling elements. This is not your typical website. Powerful little prompt right there that you can play around with yourself. Here's the formula of how to do it. And then here at the bottom we should have the different use cases. We always make a point out of including use cases at the bottom of the community guides. And there you have a few of them. There's an epilogue with a little motivational message. Very interesting, isn't it? And if you take this a step further and you flesh out this prompt into this yet again, the prompt is available right here. It fleshes out all the details of what we want on the website. And in this case it's interesting because this is actually less creative and less unexpected. The more guidelines you give it in the prompt, the less room it will have for interpretation. So in this case, maybe going with the more open ended prompt of narrative driven website that weaves together unexpected visuals and storytelling and then figure out all the details of that. Claude might be a better option than going with this advanced one. Nevertheless, I just wanted to give you the options. This looks certainly more formal, but does the trick. This menu on top jumps to the various sections of the website and it's just a very powerful way of turning any PDF into an interactive website. And there you go, you created the website from a PDF in seconds. And here's another interesting capability. You have this menu at the top right here. And you just got to realize that you can always add to these web pages, add to these projects. Just because I show you how to do things in one prompt doesn't mean that you can't have a conversation with it. I'll just go in here and paste this prompt here. Add a tab with an interactive quiz. I feature this many times on the channel, creating little quizzes, questionnaires, flashcards. All alums are really good at that. And we can use that capability and our awareness of that capability in a little prompt like this by adding a tab of a quiz that consists of five multiple choice questions. I'll just run this and it will update this project here for me. All I have to do is wait for a few seconds. The funny thing is with GPT4, I usually had to stop and wait for a minute until it's finished and then we cut the recording. But here I'm just talking over this happening a few moments later. Okay, fair enough. We had to fast forward a little bit just because generating the code for all the multiple choices, choice questions took a while. But look at that. We have a little tab with a multiple choice quiz. You could submit that quiz, you could ask Claude how to implement this practically. And there's some console messages that I could now take and paste back. And this is the way you want to work with it. You want to add your ideas and if something doesn't work and you get error messages, you just paste them back in and then it will improve on it by itself. And here I want to highlight another use case here from Sabo Schubman on X and what he actually did. He gave you the Tesla earnings report for the first quarter of 2024. And because that is very well formatted, he gave it a simple prompt like this. Come up with a detailed analysis and summary with stock recommendations by carefully reading the attached earnings report. Put together a nice interactive and highly detailed but visual appealing dashboard. And here you have the dashboard keyword. Again, very useful if you just want to create overviews of something. And here it is. He transformed a PDF into a website. This is just another use case after the step by step guide that I showed you. You could do this with any PDF, Just try it out. I mean, it's free to use. Isn't that crazy? Sure, there's a limitation of only a few messages, but you could just try all of this your today. All right, next up, I'll be using another one of my prompts. Create an interactive dashboard with editable sliders and text fields to visualize population growth in the US over time. And mind you, we're not giving it any data here so far, right? We're just telling it to create in another dashboard. But what I did here is I took a screenshot from this website, Worldometers.info you can take your data from wherever you want, but because it can recognize images, I want to show you this capability of you feeding it images and using the various dashboard website game generation capabilities that we have here to create something that might be useful to you. So all I did is a screenshot, a part of this data that is relevant to what we're doing. Here, here's the screenshot and I'll just drag and drop it onto Claude now. Now I uploaded the content and as soon as it's uploaded I can run this and it will generate this dashboard based on the data that I gave it. Right? You could also run it without the data. Then it will just invent some random stuff or tell you to connect it to an API later on. So this time we're not doing PDF prompt, but we're doing image plus prompt and we're creating an interactive dashboard from that. And again, this should happen so fast that maybe we don't even need to fast forward or cut too much. It's already working on it, it's already generating code. It's going to open up this little IDE over here. And as you can see here, I love this. You can see how it populates every single year. You can always double check that. But the vision capabilities of Claude, as I talked in my summary video, they're also best in class when it comes to the benchmarks. And practically that's the case too. So now it packages all of that, it finishes up the prompt and we should be presented with a visualization any second here. Let's have a look. There's a little bit of loading. Oh, and I'm out of messages. Okay, I'll have to wait another 20 minutes. That is okay, but look at this wonderful dashboard that it created in the very first try here I can actually adjust these sliders. Like, what is the range? Do I want to go from 1955 to 2024? Now here I would maybe like a second button here where I can readjust this. I could just follow up prompters if I had more messages this very second. But I think you see the point here. By the way, I'm on a paid plan and I've been testing like crazy over the past days and especially today as I finalize all the research for this video. But no worries, in 20 minutes we'll be back. And as you can see, this works as expected. And we have an interactive graph that we easily generated from a screenshot and a simple prompt like this. Pretty amazing. And I want to highlight one more thing. In the original chat, it cut off some numbers here for me. I didn't like that, so I just told it the numbers are cut off if I go into billions. Make it more visual, appealing, unique and better. In this example, I didn't even give it the data, so it just embedded some. But as you can see by following up with your feedback, with your human Feedback as you would be giving it to your assistant or another human that works for you, or another human that works with you. It makes these adjustments on the go, but now there's a graph missing. So I say include a graph and then it creates a graph. But what is this graph? This is more graph than website. The heck? This is some mutant graph. I don't like it. I simply told it, this graph is crazy. Fix it, and then it interprets crazy by itself. Okay, this prompt might be a bit ambiguous going against what I teach on this channel, but hey, you gotta have fun sometimes too. And then eventually, after telling it that it should fit on the screen, being more specific than just telling it, it's crazy that didn't work. It does this for me, which I think is a fantastic end result, considering I give it, what, like three sentences in total? Incredible. Okay, here's the same use case, but with the flashed out prompt that we show you how to do in this week's Community Challenge. So by copy pasting this prompt into here, you get this artifact generated for you. And as you can see from the prompt straight up, I didn't even edit the prompt right. I just ran it from my little workflow there. I get something that is immediately usable. There's a lot of details like export data, and it's just the more advanced version with more customizable building blocks in here. But this is pretty great. I get multiple ways to visualize this export button and so much more, and it all works in one shot. So this is what I say in some of my other videos. If you're there to give it feedback and if you know what you expect, it's fine to have short and simple prompts. But if you want reliable results or you're going to automate it away or you're going to delegate the prompt to somebody else, you want one prompt that contains it all because that eliminates room for error and provides you with consistency. The results from this prompt will be way more consistent than from a prompt like this, just because we're nailing down all the various variables and telling it exactly what we want, rather than letting it figure out everything itself. All right, for the next use case, we'll be creating a web app. And concretely, we'll be creating an AI powered web app. Now, this is something that won't fully work in here just because you can't link to API endpoints from within the artifacts feature. This you would need to copy over to some IDE or host it yourself on something like Vercell. But by simply telling it Create a summarizer tool that links to the Claude API. Include a field to paste text and a field to add the prompt, and you get this. Here you paste the text you want to summarize. Here you have the option for a custom prompt and all you do is hit summarize and it will give you the summary. Good. But then I'll follow up with the good old make it better. What do we get? Well, we get a better version. It figures out what better means in this context itself. Concretely here it adds an advanced mode where you can actually adjust the max tokens and the temperature. And beyond that it gives you the three different cloud options to choose from. So if you want to be using Sonnet, you can pick that. It's going to be faster, it's going to be cheaper. Again, this won't work by itself. As it outlines here in the prompt you will need to go into the code and you'll need to replace your API key here with your actual API key. If we head on over to the code, this shouldn't be hard to identify. You can always just command of API underscore and here it is. This is what you would need to replace with your very own API key. I showed you that about 20 times before on this channel. Just look up one of the Chatbot building tutorials to retrieve your OpenAI API key. Or just ask Claude where you can find it. Now beyond that, I continued in the conversation by telling it to make it more aesthetic. And look at these simple little prompts. Make it more aesthetic, make it better. Or in some cases make it more unique also works really well. And it does that. It makes it more aesthetic. Look at that. It adds this universal symbol for AI that the Internet seems to have agreed on at this point. It's kind of funny. And then here we have the advanced features. Now, it's not perfect, there's a bit of padding missing on this, but again, we could follow up on this and I just want to highlight this is not supposed to build 100% perfect apps. It's just the best LLM we ever had for building applications like this. And it allows non technical users to actually build things, which before wasn't really the case. You need to download ide, install Python, make sure you have all the packages that are needed. I covered this on the channel before. Matter of fact, there's one video that teaches you how to build a basic Chatbot. It's the first one in the Chatbot building playlist on the AI Advantage channel. And that teaches you how to set up local environments for you to copy this code into where it will work. It's about building a ChatGPT chatbot with Python. But most of the setup steps and retrieving the API keys, et cetera, are all the same. So check out that video if you want a detailed step by step tutorial on how to use this code yourself. Anyway, it's time to move on to the next use case, which in this case is something that me and the team found on Reddit. And this one is incredible, let me tell you. Now look, before you even look at this, I just have to clarify. This is not something you will get from one prompt. This must have took a lot of iteration. And as you can see, this is not hosted inside of the artifacts features. So you have to bring the code out, host it somewhere, and then you can get something like this. As you can see, this is hosted on the Internet. These more advanced prompts that I include in tutorial always give you a step by step guide on how to host. Plus, there's a plethora of tutorials across the Internet. It's not too hard. Almost everybody can figure that out in an afternoon. But as you can see, this is a web hosted app and it's this travel planner. And I think it's absolutely incredible. It includes images which were obviously added, it includes a link to Google Maps, and all of this was set up by Claude. Is there additional steps after Claude does the generation? Yes, absolutely. But can Claude Sonnet guide you through every step of the way also? Yes. You can provide screenshots of your progress as you work for something and tell it what kind of problem you have. And it will act as a tutor for you in implementing it too. But look at that. You can search for various locations across Bali with custom images. It gives you the different points on the map. Incredible. Use case right there. And here's another one from Sabo Schulman where he actually took a Screenshot of the ChatGPT interface and he let Claude 3.5 Sonnet recreate Chat GPT. Kind of rude, isn't it? But it did exactly that. It wrote the entire code for a chat GPT4O clone with the prompt presets. With all of that, you could hook that up to the GPT4O API and have yourself built ChatGPT4O clone just like that. Again, the links to all these posts are below. You can check out more details there if you're interested in that. But this brings me to my last Category, which is a bit of a tricky one because it is game creation. And I do have some very basic prompts here. But look, I just have to say, I'm not a game developer. I might be an avid gamer, always have been, but I've never created games. But with Claude, you can get your feet wet and you can build your own interactive little games yourself. And they actually work in artifacts, features, so you don't have to bring them out, you don't have to host them anywhere, they just work in there. Now, is there limitations? Of course, you won't be able to build GTA 6 in here, let me tell you that. But you can create simple little things like Pac Man, Backgammon, Tetris, or the good old Snake. So we'll do that in a second here, and I'll even show you how to create custom game assets yourself that you can then use in your very own game. But before that, I want to actually dive into the deep end of the pool and shout out the friend of the channel, Chris, from All About AI and this video that he posted on YouTube. I think this is an absolutely incredible one. If you care more about game creation and you want to see what you can do, if you're ready to bring the code out and to do a little bit of code editing, which does require some coding knowledge, then check out this video. You can also follow his project step by step and rebuild this game that he did. But here, in under 30 minutes, he actually went for a step by step tutorial through all the asset generations. He generated these little, little flappy bird assets in 8 bit that. I'll show you something similar in a second here. Himself, or maybe not himself, maybe rather in tandem with Sonnet 3.5, but that's the new doing it by yourself, I guess. And then he just goes through all the step of adding various sprites, adding AI opponents. Yes, that's something that Sonnet can do really well. Adding music and sound and background and finishing up the game. And then he has a fully functional game that he can actually play, and he uses Claude's new project feature to build it. As he generates more assets and more code, he adds it to a project and then he keeps talking to it. And Sonnet is aware of all the different code and all the different things that have been added to the game. So if you're interested in building something more complex and you try out the various use cases in this video, then Chris's video here might be a fantastic next step. And let me just note, I forgot My table mount for this mic, so I keep handing it off from one hand to the other because it's really heavy. Anyway, let's get back to the video here. So here you have a full project. This is kind of as far as you can take it with a tool like this. When it comes to game development today, I mean, I'm sure you could do a little more, but this really shows you what is possible. Very well. But again, that does require some coding knowledge. If you want to kind of just play around and have a good time. Then let me show you some other uses. Use cases before we build something ourselves, because you could not just ideate, but you could also build with Claude 3.5 Sonnet when it comes to games. In this Reddit example, Sir Wood Dafu here made a simple 3D first person shooter touchscreen game right in the chat interface. Okay, so all of this is created within the artifacts feature in the game. You shoot happy emojis at sad monsters to make them happy. Kind of a neat concept, right? By the way, the ridiculous idea for this game is Claude's. So you could first generate the idea for the game and then create it. All you need to be aware of is the very various things that you need, which Chris's video here perfectly outlines. You need opponents, you need sprites, unique music sounds, maybe a menu to navigate it, etc. And as we're going through and exploring some of these ideas, let me show you another one here, because this one is all about adding AI players, and this is something that got a lot of people excited because you can create single player games and add AI players to it really simply with Claude 3.5 sonnet. This was something that was a little tricky with GPT4.0, and let me tell you from personal experience, it just did not work reliably. If you want all the prompts, they're also here in this Reddit thread. Look, he posted every single prompt and as you can see, this is a 15 prompt sequence. So don't expect to build some of these more impressive use cases in one or two prompts. Gonna have to iterate a little bit. But I hope that this video is enough inspiration and points it towards enough resources that you're gonna be able to figure it out yourself or you join the AI Advantage community and you can do it together with others. Shameless Plug. Okay, here's another use case that you can build. As I mentioned, these basic games are easy to buildable. So here it combines a basic game and an AI enemy to build a checkers game where you're playing against the AI. And as you can see, this is running in the browser on your your phone. Okay. So you don't need to host it externally. You can build these things while you sit at dinner with your family and show them, hey, mom, I built a checkers game where you can play versus AI myself while you ate your appetizer. Actually think about it. Don't do that. Most people are going to consider that weird. Not me, though. I think that's kind of a badass move. Or here we have one more example from Ethan Mollick before I show you the last two prompts I have here. And he actually turned the academic research paper that he wrote himself. You could do this any academic research paper, interactive game. And this is a 3D shooter that references concepts from the paper to be used in the game. Again, I can't say everything is possible in here, but a lot is possible. And my hope is that for examples like this, that becomes clear. So there we go. Finally my Claude usage actually reset and we're gonna be able to do the thing that I've been waiting for all day, and that is generate an image of a cat with a hat in the eight bit style. Okay. If you're not familiar with eight bits it in a second you'll see exactly what that is. And you could see it writing the code to generate a little cat with a red hat. Is this a cat in a hat? Isn't that beautiful? And then what I want to do here is I want to follow up with a prompt and I just want to show you this combo. This is very basic, right? This is something you could figure out yourself. But it's this ability to combine that you really need to become aware of, because as I showed you, some of the best things in here are built with 15 prompts or more. So if I now say create a snake game with the cat with a hat as the player character, you can create a snake game, but you can use the little 8 bit style cat in SVG format that we created before in the same conversation as your player character. So let's see how it handles that. It should be able to pick up the svg, write the code for Snake. That should be pretty straightforward. And in just two prompts, we should have a playable game. If that's not the case, I'll follow up further. But let's see. Use the arrow keys to move the cat. This is really great. Look. And my little cat is getting longer and longer, but like, I'm missing one More thing, right? Is it just me or do we want the cat to actually hunt? So let me see if I can do this in one prompt. Create an image of a hat in 8 bit style and use it as the food. And then why am I saying food? Because I looked at what it wrote and I saw that it references to the thing that we need to eat up as food. This is a very old trick, but it's one of the most reliable ways to reduce ambiguity. Just reference the words that it uses itself. Can't go wrong with that. And there you go. The very first time on this channel, we created a little snake game where a cat with a hat is hunting more hats and extending its empire, extending its body by finding more heads. It might look basic from the outside, but considering the fact that we just made this while I was talking in under a minute, I think that's seriously impressive. And this is the worst these tools are ever going to be. I can't wait for the large model Claude 3.5 Opus or OpenAI's response to this move. I hope this video brought you some inspiration and some knowledge. And if you want to get more into this, if this sparked your interest, I would again recommend this little playlist that is still relevant on how to build a custom chatbot with code without any coding knowledge. It's the most popular long form tutorial on this channel for a reason. And using something like VS Code and installing various coding languages yourself on your own machine is the logical next step to bringing games like this to life yourself. All right, that's all I got for today. See you soon.