How to Use AI the Right Way For Business in 2026

So a lot of people talk about using AI. They're just not using it the right way. They're using it like 2023 ChatGPT. I'm going to show you the right way to use AI this year and beyond. I'm also going to dive into real workflows in terms of me using it inside of, let's say, Slack. But first, I'm going to start off with a chart that you have to take a look at to show the different levels of using AI, just so you know where you're at and where you should be going to. This whole concept over here You have open loops, end-to-end workflows, and closed loops. Okay. So what this means is open loops. These are people when you're, let's say you're working with somebody, you're basically asking a lot of questions. You're asking a lot of follow-ups and things like that. You're going back and forth all the time. Right. And maybe some stuff you have manual follow-up, you know, and then you, some stuff, the work gets lost and then, you know, stuff just doesn't get done. Now I would actually say that you can even say this is a good comparison when you're using like a ChatGPT or a Claude just to search, for example, maybe you're using to find recipes, come up with different ideas and you're just using it purely like a search engine, right? That to me is still very much an open loop and it gets messy. I would say that this in a work environment is ad hoc work and things end up fuzzy, right? But it's still fuzzy for you personally using it that way. Now, when you use AI to help yourself build end-to-end workflows, so for example, let's say you want to have a travel end-to-end workflow where you input and say, hey, I want you to help me find a travel vacation. In Italy. Here's my budget right now. Here's where I prefer to sit on the airline. Here is what I prefer to be on the water. Here are the types of activities I like doing. So you add all these inputs inside these steps, inside of workflow that you typically would have, and then you would have the AI execute in these steps over here, okay? Or inside this area. Now, the whole idea with the end-to-end workflow is that you will input what you want as a human being, and then assuming that you have the process, maybe it's a skill.md file, Then what happens is the AI thinks in this middle process over here, and then you get a deliverable to review. And so when we say end-to-end, it's this end to this end over here. And in the middle, really, I mean, some people can even say it's middle-to-middle right now. A lot of solutions are end-to-end right now when it comes to AI, or again, you can call it middle-to-middle. Really, it's— I feel like it's more or less kind of the same thing. But once you know the process, then you can repeat it over and over and over. So the whole idea here is that you have a mapped path, there's limited learning, and then repeatable process, right? So again, most people I would say are operating with open loops. Maybe they're just using ChatGPT Claude. You want to get to this level over here where it's actually doing useful work for yourself, such as booking travel or such as scheduling things for you. And then, but you, you're gonna have to work with the AI to kind of define the core steps or the core preferences that you have. And then the whole idea here is that you are just, you know, editing these processes, these skill.md files over time, right? Because the AI reads the skill.md files. So, but the, the holy grail is you want to get to a situation where you're building closed loops, okay? Where these workflows that maybe you have over here are being inserted into these loops, and then you might have different workflows over here. And so you have input, it'll execute, you get It'll measure. But the whole idea is you see these little robots over here. These robots are— they're getting the input and then they're also recursively self-improving over time. Okay. And the whole idea, like there's Andrej Karpathy's AutoResearch, for example. There's a lot of people trying to build on that. And the whole idea here is that once you have— you're getting signals, you know, what makes sense. For example, you can say from a sales standpoint, right, maybe you have this closed loop over here where you have a virtual agent that handles sales calls for you and it learns over time What calls tend to convert versus which calls don't convert. And it starts to learn from that over time, right? And so that's an example that you might use for work. But even if we go to the travel example, too, maybe it learns over time on preferences that you don't like. And maybe there's so many different things that you're doing during your trip, right? And then you're gonna have to give this thing feedback over time, because you're basically the customer. But this works both in a, let's say, a work environment, and also a personal environment. But you want robot stuff, you want robots to do human stuff, you want humans to do, right? And so I would challenge you to You have to go beyond Claude. You have to go beyond ChatGPT instead of just using it to search. You can— sure, you can use Claude Cowork, which will— you can definitely set up some of these, these end-to-end workflows. Then you want to move over to Claude Code, or if you're using ChatGPT, you want to move over to using Codex, for example, to build things. And these are both very strong. Now, what I would challenge you to do is, is to go beyond that, because you have to think about how you can have an autonomous agent assist you like a true assistant, right? By the way, if you want to grow faster, you need to have a single brain unified intelligence sitting inside of your chat. So it could inside of Slack or Teams, but you can see right here, this bot's working and it's creating ad creatives. It is doing data pulls from Meta, from Google. You can pull your SEO data. So imagine having all these data connectors. You can ask whatever question you want, get the data pulls a lot faster. Your team can see it as well. They can choose to execute, and then you can run your other specialist agents that you have inside. So we have ad creative agents, we have email infrastructure agents. There's all these agents that can do a bunch of things. And the whole idea is they're all playing together. They're playing with your team as well. That way you're going to be able to just Move a lot faster and then grow a lot faster. So check it out. Just go to singlebrain.com. singlebrain.com. We'll see you inside. So the way I like using my agent is actually using Hermes, and I also use OpenCLAW agent as well. And then I use it on ChatGPT, the latest model at the moment. And then I fall back to Claude when I run out of credits or when I run out of tokens, right? And so that's for another whole video. But all that to say is I use it inside of Slack because I like to have it collaborate with my team. I also use it inside of Telegram as well. By the way, I just made a video, my last video, which was talking about the best ways to use AI in business, right? So you can check that one out. What I would say is I took a transcript from this one and then I just said, hey, you know what? Give me the best YouTube packaging for this video from a headline and thumbnail standpoint, 3 total packaging. And I'm like, pretty not a great prompt, right? And I got the right way to use AI in business. And I got a thumbnail. I got thumbnail ideas over here and all of these look pretty good. I'm actually good with it. And so I just sent this over to my team and that's how I'm using it. But so when you think about using this for content creation, it's fantastic. And this probably would've taken me more time, the packaging. To think through. And I was like, this is already pretty good. And I think it took my instructions from inside this transcript. So then the other thing is oftentimes I'll find interesting things on X and maybe I want it to be my thought partner. So I use the X API for this. And then you can see I'm pulling, I'm like, hey, how's this going to help us? How's this going to help us over here? Hey, can you find us the latest version of this thing that we have inside of Google Sheets? And so the way I'm using it is I'm using it to build like YouTube content. For example, I'm building these, these different di— these different diagrams. I'm using it to ask questions, to be my thought partner at work. I'm using it for, let's say, coming up with investment ideas. So it's a really good thought partner for me. So I'm going to give you another example. So I read this long-form X article and I'm like, man, this is actually a really thought-provoking piece on when knowledge is cheap, insight is everything. Jevons paradox applied to Torah learning. So he applied it in a religious context, right? I was like, well, this is cool, but I'm like, wait a minute, I'm going to tie it back into AI and more so AI bottleneck. And AI bottlenecks is also investing as well. So this is not financial advice. This is just research that I've been doing. recently. So I've been looking at Lumentum, SanDisk, and AXTI. Let's just say this is part of the AI economy. Okay? So if you don't have these things, the AI buildout will stall. Okay? So I'm working with it. I'm like, okay, well, it's thinking with me. It's like, if the Jevons paradox read is true, the market is probably underestimating second-order AI infra demand. So I'm looking for it to validate or maybe invalidate the way I'm thinking and for it to spar with me. And so I'm like, okay, well, which names are the strongest here? Well, Lumentum is the strongest of the 3. They sit in optical networking and photonics. AI clusters need insane east-west bandwidth. So it's talking about all this infrastructure stuff. So SanDisk is memory, obviously. AXTI, I mean, this is one of the ones that have this, I think it's indium phosphide, which is one of the key components of making these chips. And so his take, or Hermes' take, bullish for Lumentum, bullish for what, cyclical for SanDisk, and then potentially high upside but messier for AXTI. Very bullish for broader AI infra economy, especially power, networking, optics, memory. So all that to say is one, And I spent probably, I don't know, 10, 15, 20 minutes reading this article. And then from other research I've done, I've kind of daisy-chained this research, these thoughts that I've had, and I've organized it for my Hermes to take care of. And I'm using Frontier Models here, right? The latest version of ChatGPT. And then from there, I'm getting a response and I can take whatever I want. Like, I can choose to go deeper on this, or I can choose to kind of just move on, or I can choose to ignore this, right? Really, the power is all in my hands. And by the way, in some cases, look, I'm constantly working with— I'm using /goal to get a lot, a lot of multi-threads done when it comes to working with these agents in parallel. So I just find it easy for me because this is my surface area where I can do work with myself. I can do work with my colleagues as well. And I love Slack particularly. You could use, you know, something else if you want, but I like the threaded replies over here. I like the ability to be able to, you know, open these in new windows. I can, I can, you know, do make upgrades I want to certain commands if I want. Like, there's just so much I can do with this, and I feel like the sky is the limit, right? And I like when I can dictate using WhisperFlow and all this. So you can certainly use Telegram, and I certainly use Telegram as well, but I just find that when I'm using— you can use Discord, I think, too. So you can use— you can certainly use Discord, or you can use Slack. But I find that when you're using Slack or Discord, you have more optionality with these autonomous agents. So take that for what it is. I would challenge you to think Instead of— maybe some of you are, you know, in open loops right now. Maybe some of you are building end-to-end workflows, but I would challenge everyone to get through these closed loops, to think about recursively improving these, or at the very least, building end-to-end workflows, right? So once you learn how to build end-to-end workflows for yourself personally, you can then carry over. That's a really valuable skill, and then you can carry that over into work, and then people are going to see more value in the way that you do things, because I'm telling you, that skill is going to be in massive demand. So no matter what you do, learn how to do end-to-end workflows. Learn how to get to closed loops, because I'm telling you, the vast majority of people are going to stay with open loops and you are going to run circles around them. So look, if you enjoyed this video, make sure to check out this next one over here somewhere on how you can use AI better for business. So we're going to go a lot deeper into business with real examples. Check it out.