Full transcript
So today, we're going to be answering some of the latest Q&As about our agent operating system, how to build them, how to use them, how to get the most out of them, and also how to optimize them in a simplified way so that you can make sure that you're getting the most out of them. Now, the thing to note here is, for example, we have a memory system working across all our AI agents. So for example, as soon as something like Quen 3.8 Max drops or any new model like this, we can plug it in. So if we go to the Open Code section here, we have Quen 3.8 Max plugged into this system.
We also have, for example, GPT 5.6 code over here, which is GPT 5.6 plugged into Cloud Code. And we can switch to Quen 3.8 like that in like one single click. Plus, every time something new and useful comes out, we can plug it into the system and get the most out of it, like Deep Seek as well. So with that, let's get straight into the questions from our AI Profit Board and community.
If you want to ask me questions like this, feel free to post inside the community. Link in the comments description, or just go to the AIProfitBoard.com, and I'll help you as much as we can. So let's get straight into this. Actually, what was really cool was someone created a weekly digest for the AI Profit Boardroom powered by Gemini Spark.
So they actually created like a breakdown of everything we've covered recently, all the best topics and what people have been talking about and that sort of thing. It's amazing, actually, how much we cover inside this community. So Gary says, coding app. Good day, everyone.
Good day to you too, sir. I have a quick question. I'm a total newb for starters, no worries. I currently have Cloud Pro and Chat GPT.
I want to create an app for my company, just a simple app that lists everything, et cetera. How does one begin? So it's pretty simple. The way that I would approach this is you just go into, for example, Cloud Code, or you can use Repl.it as well.
And basically just describe the app that you want, then you test it, then you iterate it. So it works in three simple steps. We're at the point now with AI, where you can basically just describe what you want and get it built. And then the main point is like, just test it and make sure people actually find it useful.
So for example, if I was going into Cloud, I could just type in what I want, explain. That's exactly what we actually did for the AgentOS was describe exactly what we want to build. So I didn't code any of this. We just built it from scratch with AI.
And then you just want to test it and build it. Got another question here from Kurt. So Kurt is asking about the AgentOS setup, which is this system. And by the way, it's inside the AirProfit Boarding if you want to get ours.
And Kurt says, in the intro video, you know, Julian talks about Cloud Code setup, the AgentOS. I'm currently using Abacus and want to migrate over to Kimi. Do I have to subscribe to Cloud to get this done? Or can I use Abacus or even better, Kimi K3?
So you can use Kimi Code to set up the AgentOS for you. I would actually recommend Kimi Work. Kimi Work is a desktop app, and it's basically like Cloud Code or Codex. You can just give Kimi the zip file from the AgentOS, ask it to set up for you, check that it works, and it's as simple as that.
And that way you can migrate from Abacus and start using the AgentOS. So if we go to Kimi here, what I would do is go over to the classroom, grab the AgentOS zip file, which is down here. And then from there, we're going to go over to Kimi, add the zip file, and then ask it to install it like that. That's as simple as it is.
Got another question about outreach. So how do you improve outreach deliverability? So there's actually a few tactics before you start sending out the email to warm up your account and improve your email deliverability when it comes to email outreach. So these are the settings that we use.
I personally do, well, my team does a lot of link building outreach for my SEO agency. And so this is something that we've tweaked and tested for years. First step, you want to make sure that you have DKIM and DMARC set up on the email. If you're not sure how to do that, you can just ask Claude or any other AI agent, they'll be set up.
Make sure that you have a Google Workspace account. That typically tends to be the best for outreach, and you want to make sure you're actually using a domain, not a personal Gmail for the outreach account. Also, don't use your main domain because if your email deliverability score drops, that's bad for everything that's marketing related in your business. So instead, what you want to do, for example, is if your website is juliangoldy.com, then you want to use an alternative domain to do the outreach, like for example, juliangoldy.com.
Basically, an email address that if the email deliverability drops on it, it doesn't matter, you can start again. And then gradually warm up the account. Now, you can use something like Lemwarm for this. This is something I used to use back in the day, and that just gradually warms up your account.
The other option that you have is you can just send, like, five emails a day for the first week, then 10 emails a day, and just keep an eye on your open rates and make sure that your open rates don't drop below 20%. If they drop below 20%, that's a sign that the email is not being delivered. And then when it comes to the actual outreach email, just make sure that you keep it conversational, it doesn't feel like AI, keep the subject line super short. So MR says, hey guys, I'm back.
Took some time to deal with real life and procrastinate. No worries about that, happens to everyone. And so MR says they want to create some videos about AI and start content marketing that way. I would say, just get started.
You know, there's nothing stopping you from doing one video a day and just seeing how it goes. Just by creating the videos yourself and doing that 30 day challenge, you will learn so much. So that by creating a video every single day, you might not get amazing reviews, but it will take time and you'll eventually get there. That's how I started.
So I would just create one video a day. I had a 30 day challenge with one of my friends, and my views were still really bad after the first month, but the good thing is I'd learned so much that eventually the views grew gradually over time until the point where it actually snowballs. And we have a full course on content creation, which I'll link to. I would definitely recommend implementing that.
Got a question from Harry as well, which is, you know, I'm currently using Hermes agent and I wondered if there's a guide for the agent OS. So we actually have a one hour guide inside the our profile boardroom here. And this is a one hour guide that just walks you through if you want to go really in depth on this sort of stuff. And then you can see, we update the agent OS every day.
You can get the zip file to install it and just go from there. Let's see what questions we've got now. So Hawk was saying that he is using Unreal Engine to make an app with the agent OS. And this is pretty cool.
So Unreal Engine actually works with Hermes agent and this can create like some amazing 3D graphics and basically just automates Unreal Engine for you. It's typically, from what I've seen, it's typically for games and stuff like that. But if you wanted to, you could actually use Unreal Engine with Hermes agent. That's a pretty good option.
Got another question here from MacGyver, which is how do I turn a website into a skills MD? Great question. So what you could do is you go up to Hermes like so, and you can use the forward. So if we type in forward into our agent OS with Hermes and then we can give it a guide.
So let's just grab one of the agent OS website like this. We can grab that, plug this in, and we just type in forward slash learn and then give it the link to the guide. Hermes will start thinking. And basically what the forward slash learn feature does is it looks at a page or looks at a guide.
And then from there, it will actually create a skill based on the guide that you give it. So let me show you an example of that from one we've made earlier. So we said forward slash learn gave it access to a guide and it said learned it and saved it for future sessions. Here's what the guide is about.
Here's the main things I've learned from it. And then here's how it maps to you. So basically it looks at your memory and then figures out how that topic is relevant to you and what you're doing. And then from here, it actually saves it as a skill.
So this is a skill.md file. And so you can see where it's saved and how to load it. So for example, this skill that it created is forward slash skill, agent hyphen, operating hyphen system. So that is the learned skill that is created based on the guide we gave it.
Here's another example. So that one that we just tested a second ago with DeepSeek, and by the way, DeepSeek with agent, really, really good. So we said forward slash learn, here's a guide and it's gone. Done, learned the guide as a new skill.
Open code hyphen Hermes hyphen Kanban. And then it's like, here's what it captures. And it talks about the architecture, the full step set up that she talked about inside that guide, the three things to be aware of. And then it shows you where the skill md file is saved.
So we've got it right there. And then it shows you what it's linked to relevant as well. So that's a really easy way to create a new skill with Hermes agent. And you just use forward slash learn and give it the guide link.
Now for the forward slash learn feature, I think that Claude have created something similar recently for a skill, but I've not seen anyone nail it as well as Hermes. I think Hermes have probably done that the best. So if you're gonna teach an agent a skill, I would go with Hermes. I think that it's probably the best at learning as well.
Now we've got another question here, which is which AI do you use to make modifications to your agent OS? So I actually use Fable 5. So when I'm making modifications to the agent OS, like for example, the other day I was adding Hermes MCP, I literally just explain what I want to build and Fable 5 will just build it into our agent OS. The same, for example, with Hermes open code, or for example, tail scale, like literally just explain what you want to Fable 5 in plain simple, easy to understand words.
And then it's set up for you. It's really, really simple. It just takes a long time to test and iterate. That's the main thing.
It's like, once it's built, you have to test it, make sure it works. I'll spend about three to four hours just improving this agent operating system. But again, if you want to get my setup that's ready to go, you can get it inside the Air Profitable Community. Jay says, if I set up an agent OS for the manufacturing company I work at, should I use three models or something like Google Gemini or something else?
Personally, if you were limiting yourself to one single model and bear in mind, like the whole point is to be able to use multiple different models. So Claude, Hermes, Antigravity, Codex, Kimi, et cetera. We even added Quent 3.8 after it dropped yesterday. If you had to pick one, I would go with Claude and use Claude CLI with Fable 5.
The reason for that is like, it's just the most reliable, fastest AI you can use. And out of everything that I've built with, it's the smoothest method that you can use, the smoothest AI that you can use for building this stuff. So that's basically all the questions. We've answered every single question inside the community.
We've answered all the other questions as well from other people. So feel free to join the AI Profit Boarding if you want to get this full set up. Link in the comments description or go to the AIProfitBoarding.com. Inside the classroom, you can grab the agent OS over here and you can see the video tutorial, the last update date, the zip file to install it.
And then we add new relevant, useful guides, space mark drops, and I test everything inside here. Also, you can post questions inside the community and I personally answer them every day. The great thing as well is like there's 3,800 members inside here, which means there's always people online ready to help and ready to answer your questions. Plus people are getting amazing results like you see with Michael.
So loads of great people getting awesome results. And if you post inside the community, I'll personally answer this with a video tutorial like you've seen today as well. Inside the classroom, you get access to all of our best trainings as well. So if you're a complete beginner, you can go from beginner to expert over here.
And then inside the calendar, you can jump on weekly coaching calls, ask questions. We have four coaching calls a week. So you can share your screen, meet other members, ask any questions that you have, and also meet everyone else and see what everyone else is working on. Inside the map, you can meet people locally near you who are building with AI agents, and this is all available inside here.
We also have over 207 pages of wins and testimonials and reviews from people inside the Air Profitable Boardroom. So I know, like if I'm non-technical and they're non-technical and we can both use agentic operating systems, then you can too, right? There's absolutely nothing stopping you. So feel free to get it.
More episodes