PewDiePie's New FREE Odysseus AI Agent!

Today, we're going to be looking at Odysseus, which is PewDiePie's new AI agent. And we're going to be testing out. It's a free open source project. You can see an example of it right here.

We already got it plugged in and it's working offline locally. Additionally, we tested out with Gemma 4, which is a local model. So that means you can actually use Odysseus as an AI agent locally, and we'll be testing now and seeing how it performs. Now, if you're wondering, okay, what is Odysseus?

So basically imagine like ChatGBT or Claw, but it lives on your computer with your data and it's free and open source. That's Odysseus, so it's a self-hosted AI workspace you can run yourself. And it's got one tidy app with a chat, an agent that does tasks, researches, that you can connect your emails, your calendar, and I'm going to show you exactly how to use it, plus how you can get access. Now, if you're wondering, how do you set this up?

You can see the GitHub details right here. So this is a free open source project. And you can set it up with Docker, or you can go natively with macOS. Just follow the instructions right here.

Or you can just give the GitHub to Claude Code or Hermes and it'll actually set it up for you. But basically it's your own AI workspace. So most tools really, they live on someone else's computer or they live in the cloud, et cetera, if you're using something like Manus, whereas this is like the same experience as ChatGBT, except that it's running locally and it's your own setup. So you can see how it works right here.

You can, there's a chat over here. Then you've got the email section. You've got the tools. So you've got the brain, the calendar.

There's a cookbook here. It's pretty interesting as well, because you can select which model you want to download. So these are all different local models that you can actually use. You've got the deep research section right here.

I will say that the UI is a little bit messy. Like you see, we've got a pop-up over here, one over here. And also it feels very retro, but again, this is a free and open source project. So, you know, it's one of those.

And, and this actually comes from PewDiePie, who's one of the, I don't know if I'll know, I don't even know if I've pronounced his name correctly, but he's a pretty famous YouTuber. So he's created this and then launched it open source, and then you can switch between models. Now, one thing I will say here is like for most people, they may struggle to run these models or to get the most out of them. So that's one thing that I've experienced with local models.

Obviously Gemma 412B came out today. So that might be worth testing out with it, but you can switch between all these different settings right here. I think for someone who's new to local SEO, that might be a little bit confusing, but you get the point. And then here, you've got this option to manage your memories, your skills, your settings, and everything else.

You can also run deep research here. So what that means essentially is you can have like multi-step web research plugged in. We've got the gallery where you can add like photos and it takes your documents as well inside the library. So you've got like an archive, research documents and chats, and then you've got schedule tasks over here.

So you can actually schedule in tasks. Like Hermes agent, with the activity, you can add new tasks. You can actually decide how it's triggered as well. So you can add new tasks here and say, okay, prompt on a schedule or prompt on an event or research on a schedule.

Right? So for example, if you're doing research every day, then you can click on this and then type in the research tasks that you have, it could be like, for example, check the latest daily AI automation news every day for me at 6am or 9am. And then store it as a notification or email me once it's done. So you get many different customizations to this.

I'm actually surprised by how many customizations you get inside here. I thought it'd be way more simple than this, but you can see here. And then you can switch between different models as well. So you can switch on open router between all these different models too.

You've got a notes section here, tasks, but again, like the UI is a bit messy. So you see how there's loads of different pop-ups here. And then finally you can switch between the theme here. So you can see all these different colors you can have, which is pretty cool as well.

But yeah, it's an interesting project for sure. Let's just check it's working. It seems like you've got two options as well. So you can switch between chat and agent.

Like you can see here, it's pretty fast to respond. We're running it through open routers, Jemma 4.26b. So it's quick, pretty quick to reply. And this way you've got your own AI and it works in three parts.

So you've got your machine, which is running the app locally. You've got your data. If you're running it with a local model, then it's got your local documents tied in, but also anything that you do inside the app will be stored locally as well. And then you can plug in the brain, right?

So you can plug in like a cloud-based brain, like we plugged in open router, or you could use a local model as well. So you can chat with it. You can also use it as an AI agent. So if you go over here to a new chat, click the agent, then we can get it to do different tasks.

So let's try that right now. And then you can see it's beginning to get to work. It does actually say trigger research failed, which is strange, but then it's come back to us. And that does seem to use the latest news right here, which is pretty useful.

So it is useful in many ways. I'm not sure about the UI, but does it work? And can you use it? Absolutely.

Then you've got the cookbook. So basically what it does here is it scans your computer and it tells you which AI models locally will actually run in it and downloads them if you click. And then you can just click them and run local models. So it's pretty good.

If you're really big into local models, I'd say it's a good option. It's got the deep research option. And then you've also got this compare option where you can compare two models side-by-side, which is interesting too. So we could compare these two models and see how they perform for a different task and just hit start, and then it will go from there.

You also got the document section where it's a writing editor and you can write in the AI assists. You've got memories and skills, so you can manage everything in one place. And then you've got the email section and everything else. So how would you set it up?

So you just download it, start it up. It'll run on this address locally, then you log in. So you just create an admin account and log in, add a brain. So an API to it, and then you can just use it and start chatting.

So it just works in six simple steps like you can see. Some people say like self-hosted AI agents like this can't be that good. Actually you can see that it actually did work and it was pretty easy to get set up. Other people say I need a fancy setup to run it.

So the app itself is pretty lightweight. The only thing that's going to be heavy on your setup is the local model, but you can skip that by using a cloud-based model instead, like Jemma 4, which is free. And then it's open source as well. So it's free to get.

So who's it for? I would say it's a great choice if you love the idea of just having one tidy app in one place. I think for a lot of people, they're going to be already focused on using Claude or Hermes. I can't see a big differentiator between them, but I like the idea of an open source project.

Obviously it will get a lot of attention because it's PewDiePie's project as well. But if you want like multiple different agents in multiple different places, then I would recommend just having a mission control instead, because it's going to be easier to organize everything. And also you can customize the UI and everything else and keep it clean how you want it. Whereas this sort of thing, you can't customize quite as much and it's set in the way that it runs.

So just to recap here, Odysseus is a free open source AI workspace you run. Like Chatube, it's here at Claude, but it's private and yours. One app gives you a chat, an agent that does tasks, deep research, document writing, memory, email, triage, notes, tasks, and a calendar. You pick the AI brain, which could be cloud or local, and your data stays with you for running locally.

Now, if you want a full guide, setup, roadmap, SOP on how to use Odysseus and everything else, you can get that inside the AI Profitable Boardroom. Link in the comments description or go to the AI Profitable Boardroom.com. Inside the community, you can ask questions, get help and support. There's always people online, which is great.

This was actually based on a question we got from Dale, who's a member inside the AI Profitable Boardroom, so it's a great question. That's what inspired this video. So thanks for posting. And if you want me to create video tutorials for you, like you can see today, then feel free to post inside the community and I'll be happy to do that.

For you inside the classroom, you can get access to all of my new trainings. If you want the agent operating system where you can manage multiple different AI agents in one place, you can get that inside here and you also get the zip file and everything else. Plus we add new daily trainings. Like you can see, you can jump on daily coaching calls.

So we have four weekly coaching calls where you can ask questions, jump on a call, share your screen, et cetera, meet other people. Inside the map, you can also meet people locally who are using AI agents like you. And this is all inside the AI Profitable Boardroom. Link in the comments description or go to the AIprofitableboardroom.com to get access.

Thanks for watching.