The FIRST Automation You Should Build in Agent OS

Today we're going to be looking at what you should do with an agentic operating system and what's the first thing you should build out. So, you know, once you've got the groundwork done, once you've got the mission control dashboard in place, maybe you've built in a couple of agents like for example, Saban has built in Hermes agent and that's responded now. What is the next step that you should go for? And this is based on a question inside the AI Profit Board. And if you're not familiar with the AI Profit Board, basically I answer questions inside here daily with video tutorials like this. And Saban, who just joined recently, is an absolute amazing member. So thank you very much for this question. Says, I managed to install the agentic operating system. What is next? And they say, I'm not too much technical. I installed the agent OS. It's working on local host, which is great. Well done. I also installed Hermes and added an open root API. Nicely done. Hermes is responding. What is the next step to go for? Great question. So let's get straight into this and we'll start answering exactly how to think about this stuff. So step number one, I think with any automation, whether that's the agent OS or any sort of automation in general, is look at where you spend your time and that will tell you the next step. Now that is personalized to every single person in the world, right? So for example, what I spend my time on is not going to be the same as what Saban spends her time on or their time on. So what I would recommend is you look at your time, where you spend your time, and then you automate that next. So for example, we actually had a question from Mazzy earlier today. And Mazzy was asking like, how do you automate ads with the agent OS? And we actually built those automations for them, for Mazzy. But my point here is like, they know exactly what they need to automate next, which is meta ads reporting and ads in general. Now, if you're not sure what to automate next, maybe you don't know where you spend your time or what you're working on, etc. Then I would just for a week, record your time day to day, hour to hour, and figure out, okay, what are you spending your time on? What repeats day to day? And then how can you automate that? So that is step number one, right? Now, step number two, with an agentic operating system, you start building the task that you identified in step number one. So for example, for me, when I was spending a lot of time on SEO, I built that into the agentic operating system. So I built, for example, the keyword research section, because I was spending a lot of time on keyword research. And instead of like going through my Google search console manually, I can easily find keywords inside the research section. The same, for example, for generating SEO content, I can type in a target keyword now. And add my case study and then generate the content and publish them directly to my website. So my point here is like anything that you can think of that repeats daily, you can pretty much automate with this system. If you're not sure how to get started with that, what you would do is you go into, let's say, Codex. And you can do this with any AI agent harness. So it could be DeepSeq harness or Claude Code or Codex, whatever you prefer. And you describe the workflow that you want to automate and then start building tests in it. So if we look at this example this morning, we said, can you add a section inside the agentic operating system that allows me to generate new ads, including creatives and the copy for ads so we can easily generate new versions of ads. It'd be great if you could look at our existing stuff and then use that to create new ones that are just as powerful. So we literally built it in. It took about six minutes to build that workflow. And now we have a workflow where we've automated a task we used to spend hours on. And that's built into the agentic operating system. So if you go down to the ad studio here, you can see that we've automated that workflow. So step number one, you figure out what you need to automate. How do you do that? You look at where you spend your time. And you just focus on one task per week. So focus on one automation per week. Don't try and do everything. Otherwise, you will feel overwhelmed. Step number two, you start building out that task. And then step number three, start testing the automation you've built in step number two. So for example, if we have a look here, we've built out this automation, right? And so I can now test it and I can see, okay, do I like it or do I not? So if we go inside here, we're going to say, okay, what are we advertising? We're advertising SEO. Who's the audience? Website owners. We can add in details of existing ads, but I'm not going to do that right now. Should be optional, I think. And then we can add the creative direction. So I'm going to say, try a different hook and simplify the language in the copy, right? And let's just see if we can generate the ads or not. So you can see here that it's given us an error. Perfect. So we've found an error. And now what we need to do is anytime we find any problems with the automation we've just built, we go back and we say to the agent, hey, this is an error that we've got with the automation you've just built. Can you fix it so it actually works? Test it yourself as well so that you make sure it works and I don't waste my time testing it, right? And so you see how in the testing is probably the most important part because a lot of these automations are not going to work first time around, right? Most of the time when you build a new automation, it doesn't work first time around. So you have to test it and give the agent that built it feedback and ask the agent to test itself so that it can keep iterating and improving every time you use it. So the example prompt will be this one. And then you can just take a screenshot and give that to your agent that's building out the automation for you. So if we go back to codex here, you can see that it's beginning to test this out, taking a look at it, seeing what it can improve and then testing out. So we have the automation that we've built in the middle. We have all our chats, previous chats on the left-hand side. Then we have the actual tool that we've built on the right-hand side. And so it's just going to start debugging from there. Now, the final thing I would say about all of this is that it never stops looping around. So you loop upwards. And what I mean by that, for example, is like week one, I started with the memory system actually for the agent OS. And then I started building out the SEO system and then the video agent. And then, for example, the image generator and the notebook, the Gemini notebook connection. My point is that every week, you're going to be spending your time on something else. And every week, that's a new opportunity to build a new automation that can improve the quality of your agent operating system. So when you implement this three-step process, which is really simple, you just get better and better over time. And so just to recap, step number one, you figure out where you spend your time. Step number two, you start building that into your agent operating system. Step number three, you start testing it. And then each week, you just loop around. And you improve and you improve and you build something new. And, you know, this never stops moving around and around. Final thing that I would say is like, if you ever do need help with anything like this, I can build it in for you. So for example, Mazzy, who asked earlier for us to build the ads integration into the system, we did that for her today just to help them as much as we can. So you can see here that we've got the new system set up and fixed right here. So thanks so much for watching. That is a great question from Saban. So thank you very much. And if anyone's watching and you want me to answer your questions like this, help you as much as I can, give you a detailed video tutorial and the steps, step by step, in case you just want to automate this stuff, feel free to check out the AI Profit Boardroom. Link in the comments description or go to theairprofitboardroom.com. Inside the community, you can ask questions, get help and support in real time. You can get personalized video tutorials like this. So for example, Mazzy asked this question about Obsidian recently. We create a video tutorial for them as well. And then if you don't have time to watch a video tutorial, we just break it down into steps like so, right? And the reason that I do that is just to help you as much as I can, give you as much support as you possibly need to win with AI. On top of that, inside the AR Profit Boardroom, you get weekly coaching calls so you can ask questions, get help and support, share your screen, meet other members, see what they're working on. Inside the classroom, you can get access to all of our best training. So we've just updated the beginner to expert course here for 2026. In the first week, you will learn how to automate anything, right? We actually literally cover in the most simplified way possible, that's very actionable, how to basically build and automate anything with AI. And if you want my full agentic operating system that I've shown you today, you can grab that inside this section. We've actually got a full tutorial on how to use it as well. And we add new trainings all the time on how to help you as well. And you can also direct message me. You can ask questions. You can connect with other community members. We have a map here where you can meet people locally. And this is all inside the AR Profit Boardroom. Link in the comment description or go to the AI Profit Boardroom.com. Thanks for watching.