NEW Agent Operating System is INSANE!

Agentic operating systems. Today I'm going to be showing you some of the latest questions we've had about AgentOS. How to get the most out of it, how to build efficient systems that really help you control and manage all of your agents in one single place. So all these workflows that you're looking at right now, these are part of our agent operating system.

We basically have all of our agents connected to one memory or one click away. And then anything, anytime something new comes out, for example, like anti-gravity just dropped the agent team previews today. We've actually added that inside the system. And the beautiful thing about this is it improves every single day.

It becomes more powerful. It doesn't matter what models come out. You don't need to chase them anymore. Instead, you can use this AgentOS to get the most out of the system.

So we're going to get straight into this and start answering some of the latest questions we've had inside our AI profit boardroom community. Because I know if people have these questions, you probably do. And I just answer these questions with a video. The first question that we had here is from Trevor and Trevor was basically asking about a new model called Bonsai 27B by Prism ML.

This is not something I've actually personally tested out, but I've seen a lot of different variations of new local models coming out from Quen 3.6. I personally haven't tested it, but from what I've seen about Quen 3.6B is a really powerful model. It's great for local. I just think like if you want something that's actually really good, even on a Mac studio, you're going to be quite limited to the number of models you can test.

But if you have something like an RTX 5090 or, for example, you have a DGX Spark, by all means test it. I'll be really interested to see what you can build. Lee posted about Cloud Code and basically, you know, when you close your laptop, it can stop running if it's running locally. Another option you have as well.

So this is a good option. You can use something called Cellage. I've not tested it out, but this is one that Lee suggests. Another option is you can actually run Cloud in the cloud.

So you can run a remote session with Cloud. That is one option. And also Hermes have a new cloud option. So we actually build that into our system.

Recently, we have Hermes Cloud and it works really well. It's really smooth and easy to use. And your agents can run 24-7 in the background. There's also a background terminal process for Hermes agents.

So even if you shut down your computer, you can still use Hermes. We got a question from Robert on how to build a backlink. So this is more for SEO. But one thing that we actually have is an outreach agent built into Hermes that you can use for link building outreach.

If you're wondering what a backlink is, basically, the more backlinks you have from authoritative sites, the more likely you are to rank on Google. So to get backlinks, all you need to do is just reach out to multiple websites, ask them for a backlink, follow up and play the numbers game, which is, you know, you might have to reach out to 100 websites to get one backlink in. How do you automate that? You can use agents that are connected to an email inbox to do the outreach for you.

We have another question from Pat. So Pat was saying that when he updated his agent OS to the latest version, and we do a regular new version inside the AI profit boardroom, so we actually have new updates that come out all the time. And then it's really easy to update. If anything does get removed that you want to add back in, you can just go to the agent that built it and say to your agent, Hey, I noticed this got removed.

Please can you add it back in? Sometimes agents just override the files. And so if it has history and you have a conversation where your previous setup was, you can just ask the agent to add back that sidebar inside your agent OS. You can also set up a rule when you're building with your agents to say, make sure you don't make any changes to the sidebar without consulting me first and asking for approval.

I'm going to tell you, I was talking about how good the agent OS with Hermes and chat GPT 5.6 is. I would 100% agree with this. So basically what we can do inside Hermes is we can create a new profile. I've got GPT 5.6 here and we can also plug that into, for example, like Hermes Oracle and Hermes Astros.

But GPT 5.6 is by far the best model that you can connect to Hermes agent on the subscription. Obviously with Paypal 5, you can't connect via your subscription. You have to use the API, which can get expensive with Hermes agent. But totally agree, GPT 5.6 is really powerful when it's connected to Hermes.

The other thing I like about it is it can generate images too, and it's right up there with Paypal 5. Plus, it doesn't cost you anything extra because you're using your subscription. We've got another question here from Foss and Foss was basically saying, you know, they've been applying for IT and data roles, but pattern is exhausting. You know, a company wants a job, they apply, silence, and then they just don't hear anything.

And so Foss has this idea where they can build an agent that tracks everything they've applied to, flags reposted jobs, skips anything they've touched in the last 90 days, etc. And this is actually something that I saw Robbie D build recently. So they basically build out a very similar process for managing and applying jobs. As you can see right here, they took our agent OS and then fully customize it, which is the beautiful thing about this, and then create their own custom workflow based on their use case.

So it was a great idea by Robbie. And I think that your idea for that workflow could work really well. Definitely check out Robbie's workflow because that's a similar system that tracks who they've reached out to, what jobs have been posted, who's in the pipeline, etc. In a nice kind of AI powered Trello board that helps you automate everything.

Then you could tweak it based on that because once you test it, you'll probably find things you want to improve. Also loving the positivity inside this community. For example, Christine has been saying she's been watching the tutorials, working on their own business, will need AI. And she was saying how great the tutorials are.

So I really appreciate that. Thank you, Christine. Now, Jason was asking about app building. Can you run Codex locally with a Quern 3 AB model?

Would it be effective? So you can definitely run local models with Codex or Cloud Code, whatever you prefer. Quern 3 AB, I haven't tested out personally, but don't expect it to be frontier level. Typically for local models, they're great as sub-agents.

They're great for working on the little tasks that require a lot of hard work, but don't require a lot of thinking. And then you can get a frontier model, for example, like GPT 5.6 to actually come up with a plan and orchestrate those local agents. That's the way that I would approach it. And GPT 5.6 can actually orchestrate those agents directly from Codex.

Now, one thing we've actually got inside the agent OS is the local AI engine right here. And it's pretty cool. Basically, you can type in a prompt, ask it to build something like so. We were testing out the new version of Jemma 4 earlier today.

And then you can preview what's been created over here. And there's a full workspace. So if you want to get your agents to run with local models, this is a great way to build into your systems. And then we also have GPT 5.6, where we can switch between GLM 5.2 and GPT 5.6 inside the same system.

And that's really nice too. We've got another win from Kenny, as you can see right here. So they've actually set up their own version of the agent OS, customized it, renamed it. Basically, they've taken our setup and then made it completely custom to them.

And you can see the full setup right here. So it's fantastic to see the people who are non-coders, non-technical, setting up an agent OS using the memory vault, which is basically our memory galaxy system over here so that we can give context to all of our agents. And they've even wired in a free local model, which is absolutely amazing. So congrats on setting up.

Congrats on setting up Obsidian. And it's really inspiring to see you sharing this. So keep sharing your journey. Thank you so much, Kenneth.

And that's basically it. I've answered all of the new questions inside the community. Thanks so much for asking these questions. If you're watching and you're like, you know what, I want Julian to create a video tutorial for me like this, where he answers my questions.

You can just post inside the community here. And every single day like this, I'll create a video tutorial for you, answering all your questions, helping you as much as I can. It's a really supportive, positive community where there's always people online as well. So if you post a question, it won't just be me answering it, but also you get help and support the whole community.

It's just a great vibe. Lots of positive people, lots of people growing and learning along the way. As you see, we actually have over 200 pages of testimonials and wins from people learning and growing with AI automation. So I absolutely love this because it's just inspiring to see people get results.

It's inspiring to see the people not just learning about AI, but actually implementing and automating their business. So that's really inspiring. That's what keeps me going. You can also get our AgentOS system inside there.

So if you go inside the classroom here, then you go to new daily updates. You'll find the AgentOS system along with the last update date, the video tutorial, the zip file to install it. And we also add new daily tutorials based on what's actually useful. So we go really in depth with some really powerful courses and that sort of thing.

We have a token minimization playbook. If you like local models, we actually have loads of training on local and how to use it. And then you can find all of our best trainings right here. If you're a complete beginner, you can go from beginner to expert here.

And if you want to jump on weekly coaching course where you can share your screen, ask questions, get help and support in real time, you can do that here. And then inside the map, you can meet people in your local area who are actually building out agent operating systems like this. So you can connect with awesome people and you can direct message me anytime too. So I hope to see you inside there.

Link in the comments description or just go to the AIprofitbond.com. Thanks for watching.