Transcription
I set up Cloud Code to run 100% free on my computer using free AI models. All the $200 subscriptions and API fees make it hard for people to get started. So, I built a solution just for you.
In this video, I'll show you a live demo and how it works, some customizations, and I'll walk through a full installation. By the end of this video, you'll be using AI to write software for free on your own computer. Let's get into it.
All right. So, now you can see I'm inside of Cloud Code here, and I'm running a free local model through Olama. This free software allows you to download free models and run them locally. And then my software here, which I'm going to explain more about in a minute, allows you to connect all these different things together in a very simple way, which makes it super easy for you to get started.
I'm going to show you some other cool tools in just a second. But let's go ahead and just start with a simple prompt. Please create me a simple node script that will print hello world 10 times. If you don't know, hello world is a typical first program that anyone writes. I'll go ahead and kick that off.
Now, one of the other benefits of using the popot is that it gives you a complete development environment. So, if I want to get an editor, which will show me all of the different files that I'm editing, I can just simply click on that and then I can toggle back and forth between the editor and also cloud code. If I want to spin up other cloud code instances, I can do that as well. And then I can also open up a shell. If you're new to programming, this might be something that's a little intimidating, but it's just a place where you can type and interface with all of your code. This is very common in the developers landscape. I definitely encourage you not to be intimidated by the blackbox. It's very easy to actually use. But again, one of the nice things about the Popbot is that all of this is contained into a single system for you to take advantage of.
So now I came back over to Cloud Code and it wrote out all of the code for us. You can see here it wrote the little program with the hello world text and some code here that will make it loop 10 times. And then from here, if you're a beginner, what I always recommend is to have cloud code teach you about these different things. So you could always ask it, can you explain how this code actually works? And you can just actually teach yourself how to learn through cloud code.
Now, to take a step back, one thing I do want to show you here is that I'm going to go ahead and open up the settings here. Now, there's quite a few different things in here, so try not to get overwhelmed. So what I'm going to show you here is how easy it is to set up Ola. See, I already have it configured here, and I'm going to go over a full install in just a minute, but I just added this extra provider, O Lama, connected it up to my local computer, and I'll explain more about this. And then you just add the models that you downloaded in your free installation here. And then from here, you would come up to coding agents. And the system is currently set up to run cloud code. And then you can see here that I have cloud code enabled. And here's the really cool part is that for the backend API that cloud code itself is using, we have it set to use OAMA, which again is what we configured right here. And then we have the different models which again was configured right here. So once you download O Lama, you can experiment with all of the different models they have. And depending on your computer and all sorts of different things, you can pick the model that's best for you. And if you have questions on any of that kind of stuff, I always encourage people to just open up chat GPT and you can just ask, hey, I'm using Olama on my local computer. I'd like to do coding. This is what kind of computer I have. What type of model should I use?
Now, one thing I did want to talk about with regard to using free models is that they are free and so they are good for getting started, but there are some drawbacks. They tend to be a little bit slower than a model that you might use with Claude or OpenAI. And depending on what type of computer you have, that's going to affect how well they work. But one thing about the Popot that I want to show you here and also explain is that not only can you use the free models, but it also gives you access to all of the other model providers. And a lot of these are a lot cheaper than Claude. So there are three other models that I want to introduce you to that will work really well. They are very powerful models. And when you compare the cost to what it would cost if you were to use Claude Opus directly for 10,000 messages, it's $5 compared to $225, which is $43 cheaper. So you can always get started on free and then you can just level up to the next level. So you can use something like DeepSeek or Miniax for a really low price and get much better performance but still keep the cost super low so that you can get into this without having to spend a lot of money. Or if you just want to stay on the free you can do that and play around.
So back here in the pop just to show you how easy this would be. If I come back to the LLMs where we configured Alama the local model you've got all these other models here that you can use. I've already got Deep Seek configured, but just to show you how easy this would be, you would just go to your API keys, create a new key, copy that, bring it back to the Popot, we would update it, save, and then we would come back here to the coding agents. And then for Cloud Code specifically, we would just come to the back end here, and we would just change this to DeepSeek. And then you would pick the DeepSseek model. And then we could come back here and open up a new chat. And you'll notice that this feels a lot like OpenAI or Claude or any other AI platform that you might already be using. And it's always starts off with this chat here. So, it's a good time for me to just explain something interesting about the pop because it's very flexible and very powerful. I was showing you the cloud code interface, the interactive interface where you have cloud code and you have an editor. You can open up files. You can edit these files manually if you want. And of course, you got the dark mode, light mode of course, and you've got the shell. But you can actually code directly here in the chat. So if you're a little intimidated with the terminal or anything here, you can get started just by chatting and coding directly through a chat interface like you might be used to in other platforms.
So I'm not going to go over in great detail here what the difference between agent and code mode is. This is a little bit more like openclaw if you've ever seen that where you're actually working with your personal assistant. This is if we're developing a new application. So, I'm going to get started here and I'm just going to say I have a new app I want to develop. I'll tell you what I want. Okay, so this is going to feel pretty normal to you if you've been using OpenAI or Claude. You've got all your chats along the left hand side. We just opened up a new one. And then you've got the chat name here. And of course, you can star it, rename it, delete it, all that kind of stuff. But here's what's cool. I'll show you how to get back here again using DeepSeek. But just check this out. help me add a NodeJS script that prints hello world 10 times. So when you do it through the chat here, it's actually going to load up the coding agent. You can see here it's using cloud code with DeepSeek and it's going to run the agent and it's going to do all of the coding here in a chat style interface which might be a lot less intimidating for beginners. And you can do a lot here. And what's also cool, too, I'll have to simulate this just in the developer tools here. But you'll notice it also works in mobile mode. So, if you were out on the go and you wanted to start a project, you could do it on your phone and you're going to see it in a nice formatted style like that. So, you can see here the agent ran and it did all the coding and it's outputting that stuff here and it created the file directly inside of the workspace. And so now if we wanted to get back to the interactive mode, then all we would need to do is come down here and click interactive. And so this is going to load cloud code back up. And notice what's really cool about this is that it kept the session. Everything that we just did in that chat style interface is now here. So we can pick up right where we left off. Of course, like I said before, you can click the code editor and open up those files. You can open up the shell if you want to get to the terminal when you start to get a little bit more advanced.
And so that shows you how you can use Cloud Code, not only with free models, but with models that are slightly more expensive. So $5 could last you an entire month. You could build out a whole application through Cloud Code inside the Popot, whether it's the interactive mode or the chat mode. You can just jump right back to the chat mode if you want. But not only that, there are some other controls down here that just make development real easy. If I click on that, it shows me all of the files that have been modified. And what I mean by that is that we added this file here. We told it to create us a hello world script, which it did. So this file exists in the system now, but from this mode, it might be kind of hard to see what files are new and which ones are not. And so right here, you can see it says plus 6. And what that means is that it added six lines of code. And you can see 1 2 3 4 5 6. And then if we click on this, we can actually see the file that it created. Hello world.js. And now this might be a little bit more advanced if you're just getting into this. But if I were to click on interactive mode and come back over here, and then we were to come over to the editor, we're going to see that hello world script right here. and we can open it up. And now we can actually type right here and actually interact with the file. And then if we were in the shell, just to show you everything from different perspectives, if I were to type ls, which is just means list the files. And L just means list them in a long format like this. If I were to just do ls, it would list them like that. I always like to do -l. We can see the hello world here as well.
So what the popebot is really doing here is that it's building this entire interface for you to make it super simple, super flexible. You have the interactive mode, you have the chat mode, and then you have the flexibility of using any LLM, any provider depending on whether you want it to be free. It also supports open router. It supports the really cheap models that are actually still very good. And then, of course, it also allows you to use the anthropic keys, your OOTH keys. And here's something that I just wanted to show you real quick as well is that not only does it support cloud code, but it supports a bunch of other coding CLIs as well. So this might be new to you if you've only heard of cloud code. But there are other coding agents. There's the PI coding agent. This was made popular by OpenClaw. So if you want to use that coding agent, you can do that as well. And of course, you can configure it with any LLM you want. There's also the Gemini CLI, which is from Google. And you've got codeex, which is from OpenAI. And there's also another one called Open Code. So down here, you're going to see all of the other configurations for the other CLIs. So, we've got the PI coding agent right now enabled and it's using Enthropic, but we can move this down to use any specific LLM we want. And then we have Gemini here as well. So, now Gemini doesn't actually support you being able to use other models. So, if you want to use Gemini, you do have to use their built-in models. Same with Codeex, but if you want to use Open Code, that will also allow you to use other models as well.
So, let's try this out real quick. Let's go to Gemini. I'm going to open up another chat. Move it to code mode. Just going to say just get the chat started. And then if I go to interactive mode, we're going to see the Gemini CLI pop right up. And now we can get started with Gemini. Or if we came back and we wanted to try OpenAI's codeex, just open up a new chat. I'll tell you what I want to code. Oh, and by the way, you can also use voice here. Hello. Hello. Hello. And we'll get that started. And if I jump into interactive mode, it'll open up the OpenAI codeex. And we can just get started. help me build my app. And again, you can open up multiple codec sessions and work together with the tabs or you can open up the shells or you can open up the editor. So, you've got the editor here. And just as a reminder, if I were to come back to the chat mode, help me build a hello world node.js script. This is going to again use the codeex CLI with whatever LLM I set it up with. And there I went ahead and built it out in chat mode.
Now, if you want more content like this that helps empower you to use all of these new AI tools for free or at a very low cost, make sure to like and subscribe. I got a lot more coming.
All right, so now let me show you how to get the Pope installed for yourself. A link will be in the description below. You're just going to scroll down to where it says get started. Now, there are a few prerequisites. These programs have to be installed beforehand. Now, this should work on Mac, Windows, or Linux. So, if you're trying to walk through and get any of these installed and you're having trouble, just go to Chad GPT. Say you're on Windows, say you're trying to install these different applications, I already have links here directly to the sites as well. But if you're having issues, this will be your best friend. Just tell it what issues you're running into and it'll help you resolve it and then once you have those installed, then you'll be able to continue through. These are all very standard, so you shouldn't have too much of an issue.
So, now to get started, you will need to open up your terminal. If you're on a Mac, you just go to your Launchpad other and you're going to have a terminal right here. If you're on PC, I believe they call it PowerShell. I'm not sure, but either way, if you're on PC, just say, "How do I get to my terminal so that I can install a program?" And if you're having issues along the way and you want support, you can always jump into the AI Architects community. It's really just two steps that you have to go through and there's a copy right here. I'm going to copy this and I'll explain a few things along the way. Now, all we need to do is paste in the text that we copied from the instructions. It's going to create a new directory and also install the Popot. Now, one thing I'm going to do here that you should not do is I'm going to change this to beta because the software that you're going to install isn't live yet. So, I need to install the beta, but you're going to keep it just as it was. I'm going to go ahead and install that. Now, while it's installing, I do want to mention if the instructions that you see in this video ever differ from what you're experiencing, just come here and check out the instructions on the GitHub page themselves because from time to time it changes a little bit. I'm always improving it and so it's good to know that.
Now everything installed, it automatically moved me into the new directory, my agent. It also gives you the next command to run, which is the same one that we have here in step two. So you just type npm run setup. It's going to check these prerequisites that we talked about earlier. If you don't have those, it's going to tell you to install it. And then it's going to ask you to name your project. And it's going to use the same name as the directory that it created. So I'm going to go ahead and just hit enter. Yes. Now it's going to walk you through a few different things that you need to go through. First thing it's going to do is it's going to open up a GitHub repo creation page. Now, if you're new to software, this might be a new concept to you, but GitHub is just a place where you store and modify code. It was originally created when there was a bunch of developers working together on the same codebase, and they needed a way to organize that. I have training on this inside of the AI architects. I'll also have more videos on this as well, but go ahead and just create the repository. And then here you're going to copy this URL back to the terminal. It's waiting for you right here. Just like that. You're just going to copy it directly.
Now it's going to open up another page to create a key. So we're going to give this Popebot that we're installing permission to access the GitHub project we just created. And right here, it lists all of the different permissions that you'll need to create. So again, you just hit yes. It's going to open that up. So right here from this screen, we can go ahead and type my agent token. Scroll down. I recommend going to only select repositories and just selecting the project you just created, my agent. And now we're going to go ahead and add the permissions. You can always just flip back and forth. You're going to need read and write. So we're going to need actions, administration, contents. You can also just type. We're going to do pull requests. We need workflows. We need secrets. That should be all of them. Now you just need to remember to come here and do read write, read, write, read, write, read, write, and read write. And now once you have these, you can just go ahead and generate that token. Generate token. You can copy it right here. And then you just drop it right back into your installation process. It'll double check to make sure you did it right. And then you can move on to the next phase.
Now, in this demonstration, it's asking you, what URL do you want to use for your new bot? And one of the easiest ways to get this up and running with an SSL certificate. That's that little secure lock that you get with the HTTPS is to use this application called Enrock. If you come here to their website, you can go ahead and sign up. It's free. Now, this might not be the way you want to do it long term. There's lots of other ways that you can set this up, but to get up and running for testing, this is the best way. You can just click this and get started for free. So, get this all set up, install it, create your account, and then once you do, you can come back to your terminal, and you'll type enro http80. That's it. You're just going to type that one thing, and it's going to load up. And right here, it's going to give you this URL. And again, you can walk through their documentation. It's going to tell you if you run into any issues, just come to chat GPT and say, "Hey, this is running into an issue with Enrot. Can you search and help me out?" So, I'm going to take this URL, going to copy it, come back to the setup, gonna drop it right here. It's going to do the final pieces of the install. It's actually going to start everything back up again. And then it's actually going to ask you, hey, do you want me to go ahead and open it up for you? If you open it pretty quickly, it's possible that you might get a 404 error, but it's just because things are still loading in the background. So, if you get that, just refresh it a couple times and it should pop up.
The first thing it's going to do is it's going to ask you to create an admin account. Just go ahead and put in your email and put in a secure password. and you'll create that account. Now, the next screen you'll see is like if you want to get urgent updates and features as I release them, go ahead and put in your email here and sign up. And then it'll take you to the sign-in page where you use the email and the password you just created. You sign in and the site is up and ready.
All right. Now, from here though, we can't send a message yet because we haven't set up any of the backend LLMs. We're going to come down to the settings down here. We're going to hit admin. And then here is the page where we can add in all of the different keys. So I'll show you how to set up the provider ones as well as well as the custom LLMs with Lama. Enthropic is super simple. You come here. You're just going to go to your API console. You'll get a key. You'll add that. You'll copy that key. You can set it right here. Now, if you are already a power user and you want to use your OOTH token, that's going to allow you to tap into your Pro or Max subscription. and you just click add token. And then from the command line, you're just going to type claude setup token. And this is going to walk you through the process of getting your own token. You just go ahead and authorize. It's going to give you that token. You're just going to copy and paste that back to the popot. And again, if you want to use OpenAI and you want to use the OpenAI codeex CLI instead of cloud code, you can do the same things here. You just go into OpenAI, grab their API key. If you want to use DeepSseek with cloud code, again, we can set that here. Going to add that right here. Save it.
And now, let's come down here and set up the custom LLMs that are running on our computer. So, before you can do this, you do need to download and install OAM. It's very easy to do. Go to olama.com and you can just go to download OAM. You can pick the instructions for whatever platform you're on. Now, you will need to experiment with which model you want to use. And again, I recommend just chatting with chat GPT. Now, these were the models that I was playing around with locally here. And what I did find is that not all of their models are listed here. And for myself, I actually found these specific models when I was chatting with ChatGpt. Actually, I was using Claude. It gave me the instructions on how to do it from the command line. So, back here at the command line, if you type O Lama, you can actually interface with your own program here. And if you type list, if you just installed it, there's not going to be any models here, but you can see the ones that I was using. And so if you find that whatever you want to use isn't actually listed here, then you can just type pull and the one you want. So the way I downloaded all of these was I just typed pull and then the name of the model. If I do that, I'm sure it's going to say it's already there. See, just like that.
So from here, you're just going to click add OpenAI compatible API because there are others other than Lama. And then you're going to put in the base URL. Now, specifically for Olama, we have these instructions down here. So, if you just grab this, copy it, put it right here. Now, for the API key, you can just type not dash needed. Anything will really work there. And then here's where you can add all the different models. Okay. So, I'm going to come back here. And we had them listed here. So, I'm just going to grab this one. And then we'll add another one. And we'll add another one. And then we'll add one more. And we'll go ahead and save it here. And now we have this.
And so now from here what we need to do is we need to come to chat. So there's really two interfaces here. We have the chat, right? We have a chat. We need to assign a model to the chat. And then we also have a model that we need to assign to the actual coding agent. And you can use the same ones, but they are configured differently. So if I come to the chat here, I have it set up to anthropic. But here you can see you can set it up to anthropic, deepseeek or a llama. And that is because we configured those. So if I came here and added open AI and copied that and we saved that and then we came back to the chat, we'll have open AI here as well. So whatever you want to use here, if you wanted to go to Olama, notice it's going to show you the four models that you installed. So you could select mistrol or whatever you wanted. And then now we can come over to coding agents. And now from here it's going to be blank because none of the coding agents are configured yet. Cloud code is turned on but it's missing the credentials cuz remember I didn't actually add the OOTH token and that that's what's set right here. So now if I click on API key notice it reconfigured itself. And then in here as well I can pick any of the model. So if I want to pick Olama for cloud code then I can pick whichever one that I want. And now if I come back to the chat and open up a new chat window, I want to code. I'll tell you what I want to do. That's just a good way to get it started. And now here is where you can use the chat mode and you can start to code from here. Or if I were to open up the interactive mode. If you just install this, do give it this a second. It will take a second to boot up the first time. And here you can see we've got cloud code booted up and it's using the Olama model. And then again, if I wanted to come here and turn on the OpenAI codeex CLI, I just turn that on. We already have our key in there. It does have the OOTH support as well. And now, if we come up here, we're going to see that codeex is here. And if I were to open up another chat, I'll turn on the code mode, pick a repository. And now, just for a moment, let me explain the difference here. So, when you're in agent mode, you're actually modifying the Pope itself. So, I originally built the Pope as a openclaw replacement. And so when you come to this page, you kind of have a choice. And they're very similar in a way. One choice which is agent is really kind of like more modifying your own employee to help you work. And the other one is to code. It could be any different project that you want to code. So notice how when I hit code mode, it's allowing me to select between my different repositories. Now I have a lot more repositories than are listed here. And the only reason why you only see one is cuz if you remember when we set up our key, remember I only set up one repository. If I were to come back in here, if I were to edit this and say all repositories or select additional ones, I'll just pick all again and I'll update it. I'll go ahead and refresh and then we'll go to code. Notice we've got all of my repositories. So if I'm in code mode, it means I want to write code for any project that I'm creating. It could be my startup or whatever. If I'm in agent mode, notice that there's no selector. It means that I'm actually modifying the employee. So, you can see here I've got a bunch of different projects. I had a I just did a video on a on building software just like this. And so, you can see this AI chatbot program I had. We can just go forward. What I tend to do is once this loads up, if this is going to be like an ongoing session, I'll come up here. I'll rename it to, you know, updating my startup and then I'll star it so that it's starred. And so if I want to start coding here, I can. And then of course I can turn on interactive mode. And just remember we changed this to the codeex CLI. So if I come back here, it opened up the codeex CLI. And if anytime you're running into any issues and you need support, just come over here to the lefthand sidebar, click support, and we'll be there.
Now, if you want to learn how to build software like this or you need support for the popup, make sure to jump into the AI architects. I've got a complete classroom and I'm also building out a completely new course that will take you from an absolute beginner with no experience to an expert in 3 months. I'm dropping in a new module every single day. I've also got courses on how to make money, vibe coding, AI automation, and you can get on a call with me personally every Tuesday or Wednesday, and we have different times for different time zones. I'd love to see you inside the community. Either way, I hope you enjoyed this video and I'll see you on the next.