Omarchy + GPT-6 Astra is SCARY GOOD!

Julian Goldie (0:00)

So today, we are going to be looking at Omachi, which is a brand new, free, agentic OS from Linus, well, it is a Linux OS.

So essentially what this means is like, it's an operating system for your whole computer that is designed specifically for using AI agents. And I just kind of want to test it today. I've seen a lot of hype around it. I want to see how it performs, how it goes with using this. And the cool thing about this is if you're on Mac, you don't need to basically like delete everything you've already got, which is kind of what I was already worried about. So if you're watching this and you're on a Mac, what you can actually do is you can get Astra 6 So I got GPT Astra 6 just to set it up for me. And before I installed it, I just made sure like, you know, can I use this at the same time as having Mac OS. So you can use something called Triomachi. And with Triomachi, you can basically run it without any setup. So here's what the free GitHub repo looks like. And then if we want to test this out, you can see I've got it open down here. Now, if we take a look, this is an example of what it looks like. And then we just need to start the setup. So if we press return like so, we can go through the setup guide here. So we're going to go with English UK, and then we can give it our username and see how it goes from here. So I'm just going to sign off for this now. Actually, this is a bit, so it's asking for a hostname as well. Again, like this is the first time I'm setting it up and configuring it. So I'm just going to get Astra to help me here. Can you help set this up? And then we're going to go from there. So basically, Astra installed it for me, and then configured it for me as well. And then basically, it's running its own windowed mode.

It kind of feels and looks like terminal, but this is basically like a whole different operating system, which is set up and ready to go. If anyone's watching this, by the way, and you have any questions about that sort of thing, feel free to ask.

Man says, would this run in an M3 Pro Max? Absolutely, I think. Yeah, I don't think it's super heavyweight. So I think you just need about 8 gigabytes of space. So for example, I'm running this on a Mac Studio. You can see here, I've got 36 gigabyte memory. It's an Apple M4 Max, and seems to be working like totally fine right now. But we'll see how it performs with the rest of the setup in a second as well. So let's see what we got. So we're going to just type this in. Now we can get this set up here. Looks all good to me. And then it's just going to start setting up the machine as you can see right here. When I'm using it, it's pretty fast and responsive. So if you ever use like a VPS, sometimes VPS can be super slow when it's in a separate window like this, but this seems to be super fast, which is good. And then it's asking me to update, so we're just going to update it right here. And then you can see it's updating now. All right, that's finished. Nice. And this is what it looks like once it's set up. Again, this is something that I'm learning as we go along here. So it's kind of fun to test this out and see how it goes. So let's see, I actually asked GPT-6 for some ideas on like what we could do and how we could use it. So it says GPT-ASTER actually says map control connection times out inside the Linux window. But let's see what it recommends for this. Let's see what we can do with this. Also, if you ask Astra, the good thing about this is like it's got information about you, it's got its context and memory about you. It knows what you work on day to day, and so it can give you like specific use cases. Got another question from Truekish, which is, hey, I have a question. Is using OpenCode Go plus Hermes good, or is they a better alternative? So OpenCode is pretty good. With Hermes, you can combine them together.

Yeah, I think that's a pretty good combination. Especially for example, you can get a lot of free APIs on OpenCode that you can't get elsewhere. So it's pretty good for that. And you can use OpenCode as the brain, and then you can use, for example, Hermes as the actual agent harness. It depends what you're doing as well. If you're doing like agentic stuff with Hermes, then that's a great harness. But if you're just coding, then you might as well just use OpenCode directly, if that makes sense. Now, one of the benefits of this with your AI agents is that when you're using something like a Marge array, basically you can get your AI agents to control this operating system in its own window. So what this essentially means is like, you can customize it how you want it to be done. So if you have, for example, certain apps you want to build, if you want to customize the actual computer system itself, then you can do that with your AI agent inside a Marge, which we'll come on to in a second. So what we're going to do now is allow access with our AI agents. You can see here that Astra is actually configuring a Marge for me. So what this means is basically we can give Astra 6 or whichever AI agent you prefer for me. Right now, my favorite is Codex inside Chachibiti with Astra 6, and that just allows Astra 6 to be able to control and configure it. So for example, we were talking about VibroCoding cool stuff into a Marge, and now that we've connected them together, we can basically give access to Chachibiti Astra 6 to be able to control a Marge.