NEW Local AI Agent Operating System is INSANE!

**Julian Goldie** (0:00)
Today, I'm going to show you how you can basically build and automate anything with local models. And this is something I've tested. I've tried a lot of different local models recently. I'm on a Mac Studio, but you can use whatever you want. There are some models that are great. There are some models that will slow everything down. I'm going to talk about the benefits of using a local model, how to run them for free, how to run them privately, how it works, etc. And one we've built as a system to basically automate anything. Now, as an example of this, this is our local AI engine, as you can see. So we can actually voice control this. We could just say something like, okay, build me a to-do list app. And then we can control it with our voice. We can click send here. And what it will do is start building. And we can actually preview that once it's done, right? Which is pretty powerful in itself. Now, also, if you want to see, okay, what have we built in the past with this stuff? So whilst that's thinking and building locally, I'll show you some stuff that we've actually built with this. So you can see here, for example, all this stuff that we've built, we can open this up. And this was just like for fun, but you can see how you can actually build things and then save it to your workspace inside this agent operating system that we've built, which is quite a lot of fun. And also the cool thing about this as well is that once we've built it, so you know, we just asked it to build a to-do list a second ago, you can see this actually ready to go and ready to preview here. Now, this is not like, you're not going to get like frontier level stuff unless you have an amazing set up. And I'll come on to that in a second. But you can see how this to-do list actually works. We can control our voice, we can open it full screen. It's going to be saved inside our workspace, so we can preview here as you can see. And that was literally just one test. So you've seen a proven case study of how this works in real time. You've seen how we can operate our voice. You've seen how we preview and also workspace. Now, the other cool thing about this, well, is like if you have a really good setup, let me show you an example. So this is Daniel inside the AI Proper Boardroom. And he actually built out a full avatar video, which is crazy, like a full AI powered video with no APIs, because he built it locally on his setup. Now this is using RTX 5090 with 32 gigabytes of VRAM. But that just gives you an example of the power of this stuff, if you have the right setup. So let me talk through exactly how it works. I mean, you can see another example here.
So this was an image generator from Daniel, and he's basically been building images locally using AI and local models, right? The same local setup that we just talked about a second ago. So this can be super powerful if you have the right system in place. And that's what we're going to talk about today. So this is the offline agent engine, which means you have a real AI model running 100% on your Mac. It's free.
It builds instantly as you saw a second ago with the to-do list app. And also, it's completely private. So normally, for example, if you're using an API, it's in the cloud, which means someone else's server. Also, usually you have to pay for an API, and it's also your data going to the cloud.
Whereas, for example, if you use a local model, you can say, okay, build me this. It loops back.
It's on your Mac or wherever you're using it. And then you get a quick answer as we saw a second ago. And nothing leaves your Mac. It's all local and private. Like if you don't have Wi-Fi, well, no problem. You can still use your local setup. Now, one thing that I've found is it's a little bit slow to set up a local system unless you have it running, which is what we actually set up with this system here. So we can just build stuff privately locally for free. We can do it offline as well. And we can save everything inside the workspace. And because the Asian OS is an offline system, you know, it's hosted locally. It doesn't matter what happens. We've still got it ready to go. So you have the system and you can just swap models in and out. Now, if you look at the old way versus the new way, right, a lot of people will have a subscription. They've got a monthly AR bill.