How I Grow Revenue With AI Loops: The Three I Run Every Week
This content was created by Eric Siu. This content was uploaded by Eric Siu. This is a YouTube video titled "How I Grow Revenue With AI Loops: The Three I Run Every Week" (published July 10, 2026). The content appears to be a presentation/talk where the creator explains how they use AI-driven loops to grow revenue, outlining three specific weekly AI loops and how they operate. No transcript or speaker information is provided in the metadata, so participants are unspecified.
Chapters
I use, maybe not all, maybe just 3 to begin with, but AI loops I use for work to help grow revenues faster. Because all this stuff, all this talk about, oh, I use this stuff to make it my chief of staff, or you use AI like you use ChatGPT, that's not interesting. What's interesting is if it helps you grow your business faster. And for whatever reason, still to this day, it's July 9th right now, Still, people are not really talking about it. I don't understand all these content creators right now. How are you not talking about how it makes revenue for your business? Oh, well, maybe it's okay. People are at different stages. Maybe they don't have a business, right? Maybe they're solo, which is totally fine too. But I'm going to show you real-life implementation of these AI loops because people have been talking about loops for a while. But what's interesting to me is yesterday a recruiter reached out to me, like someone that is like a placement recruiter. And he's like, yeah, I use AI for sourcing talent and all. I was like, oh great, fantastic. What are the loops that you build and how did you build them? And he's like, well, it's not like from a machine learning perspective, blah, blah, blah, blah, blah. He's writing, it's like 4 paragraphs, wall of text. And I'm just like, this guy has no idea what he's talking about. Okay, so I'm going to first do a refresher on what a loop is and how it's a little different than what an agent is or what a /goal command is. And if you stay till the end, we'll even share with you the Loops Creator skill that we have. And you know what? I'll tell you where to go get it afterwards, even better, because we have even more skills for you there. So you're going to have to stay till the end for that. And I might forget, Max, you might have to remind me or someone might have to remind me in chat. But drop in chat right now where you're coming from in the world and we're going to get started here. And let's first go through a recap of what a loop is. Okay? So I actually have this nice diagram for you all. Let me share my screen. This is the difference between an agent versus a loop. Everyone's talking about loops right now. Again, loop agent has— there's prompts that you can have. A lot of people, they're not even using prompts anymore. They're too good for prompts, or /goal command where you tell it to do something and it does it for an extended period of time. For example, right now, and I'll show you in a little bit, I've had a /goal command running for the last, I think the last 60 hours or so, something like that. It's still running right now. And you can use different tools. You can have different connectors, right? So you can connect in your CRM, you can connect in your revenue intelligence, like your phone calls. You can connect your different project management tools. The more tools that you have, the more helpful your agent becomes, like Hermes, for example. And you can also have answers and artifacts in there. So you can have PDFs, you can have Documentation. And this is good. Agents are really good, don't get me wrong. And I use a lot, was using my agents this morning. They're really good. They're extremely helpful. Now loops, on the other hand, the main difference, because when people are talking about loops, everyone's like, oh, the creator of Claude Code, the creator of Open Claude. Oh yeah, it's all about loops, all about loops. And then everyone's talking about it, but nobody's explaining it well. And then through a lot of studying over the next, I don't know, 1 to 2 weeks. I say a lot of studying, really just me thinking about it every day. Not a lot of studying, but I realized the main difference is that a loop is able to record the learnings and it's able to compound on those learnings. People talk about recursive self-improvement, but in my mind, a loop is something that is able to look at what the target state is. The example I like to use here is that Let's say you want all of the sites, all the pages on your site to be 50 milliseconds. Okay, that's a target state that you're defining. Well, maybe then you want your loop to be able to observe the current state. You want to be able to see what the base level is at. Maybe the base level is 150 milliseconds. The gap right now is 150 minus 50 equals 100. 100 milliseconds is the gap. Then the loop will continue to work and you can have a slash goal inside of a loop and you can have it run for an extended period of time. And then you can learn, stop, or escalate. So this is what we're talking about here. We want to build operating leverage. And I don't think enough people are talking about it in this way. And so that's why I decided to make a skill that's tied to it. And I workshopped this a little bit. I shared a lot of these X articles. I looked at what other people were doing with their loops and I made my own skill with it. And then I realized that I'm like, okay, well, if you're going to record the state and you're going to record the learnings, where do you store all that? And so I'm like, can we just— I was just talking to my Hermes making this Loops Creator skill. I'm like, Do we just put it in a repo? It's like, yeah, repo's fine. I'm like, okay, we're just gonna put it in a repo then. So at the end of this, I will share with you where to get the Loops creator skill that we have and also the repo template and a whole host of all of our other skills. It's basically a skills marketplace. Okay, so we'll get to that in a little bit, but I just want to share this with you all in terms of what an agent versus a loop looks like. Again, a loop is something that can create operating leverage. And an agent's good for bound and execution. So that's what I would say. And I'm going to quickly look at where you guys are all coming from, by the way. So let's see, we got Montreal, Sean. We got Pinehurst, North Carolina. We've got Austin, Texas. We've got Spokane. Amazing. Thank you for that, Bob. I'll take that as well. We got Minnesota. Bill from Phoenix. Okay, great. Orlando, Florida. Okay, awesome. So, and I'll answer you. Keep asking questions as we go because I can see questions coming in from YouTube. I see questions coming in from the webinar as well. Kevin, one of our ClickFlow customers coming in as well. By the way, one other thing I'll say. So Kevin is a user of one of our products called ClickFlow. ClickFlow is an AEO SEO tool and there's an API for it. There's an API in MCP. Guess what? When you think about your agents over here, when you think about connecting tools, you can connect, let's say you're using an SEO tool like Ahrefs, you can use the API there or Semrush. And then you can use ClickFlow or you can connect in with your Google Analytics 4. You can connect in with your Google Search Console. You get connected in with, for me, when I talk to Max on this call over here, I'm like, dude, I talk about so much every single week and there's got to be at least 3 or 4 content spikes that should become content on our website for AEO and SEO purposes and he should draft it out. And then using my friend's company, my friend has a bunch of WordPress businesses. One of them is called WP Vibe. So if you're using WordPress, there's a WordPress MCP there. So imagine It's looking at all the content that I'm putting out there. It's looking at what's genuinely differentiated. And then it'll use something like an Ahrefs to look at, hey, is this keyword, are these topics actually trending upwards? Are they not? Are these perhaps content gaps that we should attack? Then you can use a content writer like a ClickFlow. By the way, it's free. You can use the API MCP for free for a little bit, test it out, see if you like it. And then what you can do is you can use an MCP with WordPress and then push it over there, draft it. I wouldn't say just let it YOLO and let it do its thing. But that's an example of an agent that runs. Now, what I'm going to move to next is I'm going to show you the loop that keeps this going. Because the problem right now is I'm talking about an agent here because the problem with an agent, you got to keep running it over and over. It doesn't learn. You got to keep repeating it. And it's like, that's like a solid B player. A B player employee is like, you got to tell them what to do. An A player employee is like, they just solve the problem and then they compound and they keep learning. I was actually on a standup call about 30 minutes ago And I was talking to one of our designers and I was like, who, like, you're really good at AI right now. Who taught you all this stuff? Right? And I wasn't expecting him to say me at all. Him. He taught him all this stuff. Good people don't need to be told this, right? And so I would just encourage you all to say this. I had a friend call the other day. She's like, every day I get strong with this stuff, my tolerance for BS goes lower and lower. And you all should feel the same. So if you're working on a company right now and you find that there's a lot of inertia around you, go find another company, right? But if you're running a company right now and there's a lot of that BS around you, you got to drop the hammer, right? And you can drop the hammer in a nice way, but that's all I'm saying. So I digress. Let me show you the AEO loop here. This is the first loop that I'm going to show you. And John, maybe I can tie this into Claude and Ubersuggest MCP. We'll come to that in a second. So I'm seeing Neil later today. We're going to be recording Marketing School too, by the way. Let me move my share to my Codex. Codex is my daily driver. I like Codex even though I have Claude Code as well. I think you all can see this. Okay, great. This is my GEO AEO loop. Maybe I'll put this in the area that we're going to have a section for loops for that skills marketplace I'm going to show you as well. The way this works right now, this is working for a very long time. All this here, I'm going to say, watch. Hey, I'm doing a webinar right now. Can you explain in one sentence what this actually does and how high leverage this is on a scale of 1 to 100? And that's a little thing while this thinks too, by the way, I always like to ask the question. I want my AI to push back on me. I want it to be my thought partner. So 1 through 100, how high leverage is this? I want to know how genuinely differentiated is this or is this just like a Like a BS thing. And oftentimes I'll be sharing my, whatever I'm reading on X, I'm just constantly sharing it to my Hermes or maybe sometimes I'm sharing it to my Codex over here. So what does it say? Look guys, I don't even have to work. I just— this turns SEO and AEO work into an operating loop that finds page opportunities, drafts or updates content, quality checks it, routes it for human approval, and then tracks results over time. Leverage 92 out of 100 if we connect it cleanly to publishing and analytics feedback. How does it work in one sentence? What connectors does it have? I'm going to keep working here. What I'm actually doing right now as I'm going back and forth, look at all this other crap I have on the left sidebar. Every day as a business owner, I think about all the bottlenecks. If you guys have read the book, The Goal, highly recommend that you read it. I highly recommend that you push your team to read it too, at least your A-players, because then they will understand what the theory of constraints are. 'Cause every business has constraints, always. Every business has constraints. What does it do here? It pulls signals from ClickFlow, Ahrefs Brand Radar. Ahrefs is a great MCP API, by the way. GA4, GSC, so Google Analytics 4, Google Search Console, Hermes Content Ingestor. Max here knows that we have this Hermes Content Ingestor where it's pulling my content every single week. Local site and template evidence turns them into update, consolidate, delete, net new page actions, drafts the work, runs QA. Stops for human approval, then tracks 7, 14, 30-day results. Okay, well, how is this a real loop and how did the Loop's creator actually help build this and how helpful was it? So think about this. If you're running this for AEO and SEO and it's running every single week, imagine if it's pumping out 50 pages for you and then you have an editor that's spot checking these and maybe you approve 20 of them. So you have 20 new ones coming per week. Or maybe if you have a travel website and you have a lot of pages on your travel website, you can go a lot more aggressive if you want to. And then you can probably have a loop that's spending time just focused on AEO optimizations. Do you have an FAQ? How's your schema markup doing? All these things. Oh, you can probably— all the things that you do right now are a loop. Come on. If you're not setting these up right now, You are, I think you're wasting time in terms of how you work. So, excuse me. So it is a real loop because each run has state evidence, scoring, action decisions, human gates and follow-up measurement, observe page content signals, choose next URL action, draft or prepare packet, evaluate quality, duplicate template fit, ask for approval. Guess what? If your SEO agency or your SEO sucks, fire all of them. You don't need 'em anymore, right? Yeah. I'm sorry. You don't need the ones that aren't good. You need the ones that are good to run. You need the ones that are good to run this stuff. You need the ones that understand how to compound themselves. You need to have the ones that understand how to create leverage out of their lives. Honestly, if I were coming out of college right now, I'd be having the time of my life because I played a lot of games growing up and I'm like, dude, I think I'm going to be a failure. This is when I was a kid, by the way. I'm like, man, I think I'm going to be a failure because all I want to do is play games, but this feels like a video game. Actually, Max, people can't hear you, but Max, does this kind of feel like a video game right now for you? I'm going to see what Max says. Go ahead. For sure. Max says for sure. I don't know if he's just appeasing me, but I think he's actually telling the truth. When life feels like a video game to you and you wake up, you look forward to it. And then you realize that you only wanna hire people that work like this. Okay? So anyway, all that to say, we're gonna come back to this in a little bit, but my point is you guys can all see this. It doesn't have to be an AEO or SEO thing. How does this apply to social media? 'Cause I'm sure all of you watching this right now, the wheel should be turning in your head. How can I make this work? What is the most annoying thing right now that I can make into a loop? Now I'm gonna move to the next one to give you some more ideas. I have a CRO loop over here. Okay. So this CRO loop, okay. So like, what does this do in one sentence? And tell me how it's a real loop and tell me what connectors it has and tell me how high leverage this is on a scale of 1 to 100. You guys see how I'm doing that? Same thing over and over. In fact, go to YouTube later, take the transcript from this video, okay. And just say, hey, how do I skillify everything Eric is talking about? That's how lazy you can be now. Be smart, but be lazy. All right. So Jack says, yeah, weekly scan my website traffic. What's the workflow for making the SEO loop? Well, you got to stay till the end for that because we will share the repository, more so the marketplace on where you can get our skills and our loops. All right. So let's see, let's see. Where do you get the results of your Codex loop? So basically the way this works right now is you can see that there's a little clock next to this thing. You can do the same thing in Cloud Code as well. You can do the same thing with your Hermes. You can set up cron jobs and you can have the desktop app and you can look at where all your skills and where your cron jobs are. If you're using Hermes, I highly recommend using the desktop app. For someone like me, I get lost a lot because I have so many things running in parallel. But this is a real CRO loop because it runs scheduled live site and GA4 readbacks, records state artifacts, compares conversion rate against baseline and target decision rules, and escalates, keep, win, lose, iterate. Connectors are GA4 service count, public site checks, and Codex cron local repo plus Vercel rollback contacts when approved. Because it turns— so this is 86 outta 100 leverage because it turns homepage CRO into a repeatable decision system but still lacks randomized A/B split. Okay, watch. So I don't like an 86, right? Because when you're Asian, an 86 is considered an F. Okay? So I want you to put together a— I want you to do a UX review on this And I want you to put together a team of expert— I want you to put together an expert panel of UX/UI people that are best in class and also conversion rate optimization people that are best in class to figure out what we need to do to improve this, to make this a 95+. So I want you to continue to score this until it becomes 95+. And then in one sentence, tell me the changes that you would make. And let's go from there. So what I'm doing over here is I'm saying, oh, what did I say? I want you to do a UX review. So that is pulled from Gary Tan's, he has GBrain and GStack. So he has CEO review, UX review, finance review, all these different reviews that you can run. So these are skills that I'm pulling in here and I have it. Look, look, look, 3 skills. I forgot I had these. Okay. Apparently, we have an expert panel one, we have a paid CRO one, and we have a design review skills as lenses. And we're just gonna let this work. But my point here is that every single week, this is gonna run a new test. And it knows the baseline number is 1.80% right now on the conversion rate for singlebrain.com. And I said, hey, just raise it by 30% in the next 4 weeks. This is something that was a full-time job before. Right. in terms of having someone that's on top of your CRO. Not anymore. Same thing with my AEO SEO piece. Not only is it scanning content ingester, it's scanning my website too. It's scanning for opportunities. Maybe it's also scanning the web too. You can use— you can— sorry about that. You can also combine that with other skills like Matt Van Horn has this, it's called Last 30 Days and you guys can look it up. But the Last 30 Days skills is really good because you can just say, hey, Hey, use the last 30 days skills to look at what's trending in marketing or AI or business. And in fact, I used that today for Marketing School 'cause I'm recording in a bit with Neil and it came up with some pretty good ideas. In fact, I'll show you those ideas in a moment. And let's see, let's see what's happening over here. It answered me a little too quickly, so we'll see what happens. Let me pull up Marketing School for today. I'm going to show you the topics that it came up with. This is in fact something I can make into a loop too. If I show— here, let me just share my screen real quick. I'm going to move and then maybe I'll answer some questions as we go along. Let's move over here. Great. Let's see. Notice today, Is the 9th of July. All this is from the last 30 days still. Okay? So sorry guys, I don't know why the coughing is coming out now, but your attribution dashboard is lying. Cold email reply rates are collapsing. LinkedIn wants you to own— wants to own B2B creator discovery. Software is becoming marketing. Okay, that's pretty good. Your brand is invisible in AI search. Okay, guess what? This is a— I think they gave me 10 ideas, 15 ideas, something like that. And then I gave some ideas here too. And I said, hey, I threw in some ideas. Also add in the hook. I want you to add in the hooks here. And I added in the hooks. What if half of companies need new CEOs? So ex-Google president, 50% of companies will need new leadership. 95% of AI transformation is failing. Why are 95% of AI transformation— okay, this is all good. Because then now I don't need to think about what are the hooks I'm going to have for short-form content. Okay? I don't need to think about what's interesting because Last 30 days, we'll look at the entire internet and look at what's interesting. Guess what? I can plug that into my AEO thing. So you can just, every single week, it's not like you're done working. You're constantly adding more and more to the loops that you have. And you're also probably deprecating the loops that don't make any sense to keep. But if you continually do that, and this is how we teach our people internally, we do this pod of one apprenticeship right now. And I'm going to come back to Codex in a second. But we do this pod of one apprenticeship. Why do we do this? It's where I'm doing a group training with the people who are AI pilled. Most people are AI pilled, I'd say, in that meeting. And then we are trying to make one person, we're trying to give one person the ability to do 10 different things. And so a week ago it was more so about building the right skills, having someone else use the skill and then make a Loom around it to verify it. And so that's the first piece. But now we're moving on to loops. So how do you make a loop? How do you verify it? Excuse me. Okay, so that's that. That's how this works. But hopefully you guys are seeing how all this stuff you can actually daisy chain them together. And that's why it's very important to have a skills marketplace as well, because you want to be able to share the wealth. There's no sense in me doing this all on my own if I can't share this with my team and I can't help my team get better. It is actually a failure on my part if I become way stronger and my team becomes weak. They don't get stronger. That means I'm not a good leader. So as much as possible, your job is to bring the people who want to come along and create the conditions for success. And there's going to be a contingent of people that don't want to come along and they're going to self-select themselves out. But that is the way the world works. And I was talking to a friend of mine the other day. Her business does about $14 million a year. And she was like, again, every day I get stronger with this stuff, my tolerance for BS becomes lower. And that's how you should feel. Okay? So let me move over to go back to Codex and then I'm going to show you. Okay. So specific 95+ changes. Okay. Change the hero to name the buyer and outcome directly. For example, deploy managed AI revenue agents for sale. Okay, great. These are all the changes it's recommending. Then these are things that I might just let it run next time. Let's just ask it. I'm actually wondering what's going to happen. What happens here? If your test this week succeeds, then what do you do next? Or if it fails, then what do you do next? I'm on a webinar right now and I'm showing people how this works. Do a good explanation here in terms of Explaining your logic in terms of how you would lift our conversion rates by 30% in the next four weeks. Okay, so while that works, I'm going to show you one more thing. This is I'm going to make this into a loop, but this has been running for the last two days and 14 hours. Okay, so 60 hours. This thing's been running for 60 hours. It's a slash goal command. It's a It's building our TalentOS recruiting loop. Let me just see what this is. I'm going to tell you what I did here. Bring the TalentOS recruiting loops up to date across this documentation over here. Then inventory completed work, reconcile existing changes, implement safe local changes in order, validate with tests and stop before external writes. What does this do exactly? With recruiting right now, because recruiting is a big bottleneck for us because we only want to hire AI-pilled people. By the way, if you're AI-pilled and you're looking for a new gig, and you think you're that damn good, we would like to talk. Just go to singlegrain.com/apply. Again, singlegrain.com/apply. So looking at this over here, some of the best people that we've gotten have come through sourcing. And we want a machine that can basically source through LinkedIn. It can also source through emails as well. And it will follow up. It'll also optimize the messaging that it's sending to people. It'll also look at the roles that we're prioritizing every single week because it changes for us. Quite frequently. So right now you can see the platforms that it's touching right now. It's looking at Team Tailor, which is our ATS, our applicant tracking system. There's LinkedIn. It'll look at Juicebox, which is our AI kind of sourcing tool, Gmail, and it'll look at other candidates that we have and it'll start to spot patterns. And you want a system, a recruiting system that continues to get better. Guess what? You no longer have to pay all these contingency fees with recruiters. Okay? You now have a machine that can do this for you. So all this stuff that I'm showing you right now, if you're a business owner right now, you can see all the problems with your business every single day. What's the highest leverage problem that you can make into a loop every single day? And sometimes it's better for you to just focus on one loop for the rest of the week. But I'm telling you, man, this is probably in the last 3, 4 weeks, once I understood how loops work, this is one of the highest leverage things that I can do. And I love doing this stuff. I'll tell you what I don't like. I don't like dealing with people problems, right? The A players on the team don't have to worry about anything because they just want to get better all the time. What I hate dealing with is, oh, I just don't have enough time. Oh, it's too hard, right? I don't know how to do it, right? You have to put your foot down as a leader and say, those excuses might've worked 7 months ago, not anymore, right? And you get what you tolerate ultimately. Now I get it. Some of you, maybe you're newer leaders, but I'm telling you, I used to care so much about being liked and being a nice leader. Nope. Nobody likes that. That's weak. All right. So that's how this is building right now. So sometimes these loops take a while to build. Again, working for 60 hours. I'm going to show you one more. I'm going to show you my Mr. Siu sponsorship one, which I showed last time, but I'm going to show you how it's getting better. And oh, actually, We have this HubSpot hygiene one over here. What this one is, it's looking for, it's coming up with suggestions on what to fix or update from a hygiene standpoint in HubSpot. Right now we're still having a human in the loop review the changes that it's recommending. Then once we iterate on it, it's going to become better. Because some of these things sometimes you have to, as you're building these loops, you have to still be the human giving it feedback. You have to constantly be iterating with it. Otherwise, if you just expect it to build it all for you with limited context, you're going to get garbage. That's that. Oh, what else? ChatGPT light loop handoff. This one's cool. Max, I don't know if I sent you this article, but this article was written about this ChatGPT thing. I read about a guy at— he's the head of go-to-market at Cursor. And he's talking about how all of his sales reps will have a brief every morning for the calls that they're going to get involved with. And then also it makes a custom landing page for all of the, let's say, the discovery calls that they're about to have. And so they can go through that custom landing page. So imagine when you get on a call with someone's doing a demo with you and they show you a custom landing page and it's clearly set up just for you. And it looks like they went through the time and effort to do that. How much more likely are you to win that deal? It's gonna go up significantly, right? Your conversion rate's gonna go up significantly because it shows that you went above and beyond. And so he does that. He preps people before the call, after, and he's made all these things. Nobody asked him to do this at Cursor. So this guy's name is George, and I didn't realize this afterwards. I'm like, wait, his name looks familiar. Guess what? This guy went to my high school and he's a valedictorian. He's much younger than me. He worked for my friend's company for a little bit. My friend has a fund with Jake Paul. And I think he does some business with Logan Paul as well too. But my friend Jeff hired this guy before and then now he's at this company. And so I'm just looking, I'm like, you know what? I might just reach out to him. But what's easier right now is I just took the post itself. Hey, which repo does it make sense to build this into a loop using the Loops Creator? You see how lazy I am? I have a Loops Creator skill. I have an article that I like and I'm like, hey, we might have repos that are related to this. In fact, I know there are as it relates to sales. I want to build out what he's talking about as it relates to what we're doing on the sales side. So it's almost like those of you that follow basketball, I'm trying to be this point guard where I'm just creating these little loops, maybe getting them 60, 70, 80% there. I'm prototyping them. And then once I feel good about it, then I can hand it off. And so I think the job of the founder or a leader now is you're becoming this point guard where you're distributing the ball quite a bit, And even when we talk to clients on the single brain side where we do these implementations for clients, these managed revenue agents, it's a very collaborative call because everyone wants to do this stuff, but sometimes you just need a little direction. And so all that to say is once you start to show people how to do this, we work with other founders that we show them how lazy we are being with this. It just basically becomes a drug. So I'll leave it at that. So I think I have, I'll show one more and then maybe I'll answer some questions here. So I showed you guys this one last time, but I have this Mr. Siu sponsorship one. Okay? So I have no idea what it's doing right now. Okay. Here's what I'll say. I have this loop that basically handles my sponsorships that come in for my YouTube or my podcast. Basically, it messages, it replies to people saying, Mr. Siu's rate is $25,000 for one post. What inevitably happens is they'll say, oh, sorry, we can't afford that. We just want to do an affiliate deal. Our budget's only $1,000. It's $500, $100, whatever it is exactly. Then my thing became smarter and smarter over time. It was like, I'm sorry, Mr. Siu is very committed right now. It's $25,000. That's all he can do. Let us know if that works. It draws a hard line. And then what started happening afterwards is people are like, okay, well, if it's $25,000, what are the usage rights? Can we put usage rights? How long do the usage rights last for? All this other stuff that I haven't been thinking about. So now what happens is this loop learns over time. It's running every single day. So now when it responds to people, it's like, hey, if $25,000 works for you, Our usage rights are for 1 year and it's for this and that. It starts adding in all these things and sometimes it'll check with me if I want to have that in there, but it's adding it in there and it's making the responses better. It's making them more crisper because it's starting to see all these patterns. All this stuff is just ultimately pattern recognition at the end of the day. For me, I don't want to deal with having to deal with this. I also don't want to hand this off to someone on my team to deal with this. In fact, someone on my team could be doing something that's higher leverage. Again, you think about all the things that annoy you right now, prioritize them, and then start to think about how you can loopify this stuff. I'm going to start to answer some questions now and maybe I'll jump back to my screen share if things pop up. Let's see. I'm going to answer some questions. Let's see. Okay. Let's start with the— what do we got? Okay, not that one. Thank you. Interactive chat. Okay, got it. Do you have a really— you want to just throw a good one up, Max, if you saw a good one? So John— okay, you can throw one up. But I know John asked one about having all this stuff hook in with Claude Code or Ubersuggest MCP. You totally can do that if you want to. And the next one would be from Bob McIntosh, I think. You got that one? Okay. So this one, could you share how you're framing to your team to step up? I find that I think my team is genuinely afraid AI will take their job. I keep saying it's about 10x productivity and not replacing them. So it's been 7 months and here's what I'll say. The last week of December, that's when Opus 4 or 5 took off and I went crazy on it. That was my Christmas vacation basically. I told everyone in the beginning of the year, everyone's going to have to be proficient with Claude code in 6 months. Now we've got all this other stuff. We've got Hermes, we've got Fable 5, we have OpenAI ChatGPT 5.6 that just came out today, SOL. All that to say is that I've noticed that the people that had that sense of agency, the people that wanted to grow, the people that have a good relationship with change and they're very adaptable, they ate it up. There needed to be not really any pushing. So we're doing all these hackathons, we're doing all this, we're doing these AI office hours and things like that. We're meeting with people one-on-one. But I've just found that you can't change mindsets that easily. People have to change it on their own. I just feel that it's your job as a company to set up the conditions for success. And if people want to come along, they can come along. The other thing too is if people have that mindset where they're genuinely afraid of it, my real question for that, Bob, is how much time are they actually putting into it right now? Because it's like there's— if they're genuinely putting the time into it, they'll realize that AI is— my friend likes to say it's IA, it's an intelligence amplifier. But if you're continually afraid of it over and over and over again, You're starting to become a liability to the company because it's no different. You could try all this different framing. Oh, this is no different than when the internet first came out and people are a lot faster because of the internet. But this time, the only problem that I see with this is that you have synthetic intelligence. And if you're compounding with the intelligence, the gap's going to get bigger and bigger. And so in my mind, it's like, hey, you set up a condition for success. If people want to come along, they come along. But if not, you have to make decisions for your company because if not, you're going to have 2 companies. You're going to have one company, one part of your company that is super AI-pilled and they're having fun every single day and they're working super hard. And I just worry about their burnout and I worry about how do we pay these people the best long-term, right? That's what I worry about. Then there's the other side where it's a different conversation. It's where you start to hear the things, oh, It's too hard. I don't have enough time. I'm overwhelmed. But then it's like, what? But I'm working so hard right now and I'm like, yes, it's my business, but I don't have— I've certainly felt overwhelmed before. I don't feel overwhelmed right now. In fact, it's more so I can't stop myself from working because it's so fun. So that's what I would say. You don't want to have 2 cultures right now because that's going to cause culture clash and that's going to hold you back from achieving your goals. So that was a longer answer to what you're looking for, but I hope that helps. What's the next one? Yeah, I think I answered that one. You can throw it up. That's fine. Okay. Okay. So where do I get the answers, the results from my Codex loop? It drops into Codex and it would do the same thing inside of your Claude code as well. You can run CLI, you can run this in the CLI as well. You can also do this inside of your Hermes agent or your Open Call agent as well. There's that. The workflow, let's see what else we got here. I'll scroll down. I forgot that I can actually pull these up myself too. Let me pull this one up. Let's show this one. We need to show it for longer. Let's do 20 seconds. Here we go. Joe asks, you shared how your AI just pulled skills and ones you didn't even remember you had. How are they set up to go in and use them? It's a skill and the skill says to go search Dojo or— Okay, so here's what it is. A lot of my skills live in my private repo in GitHub and I just connect my GitHub, my Codex to my GitHub. I connect my Hermes to my GitHub. I connect my Claude code to my GitHub and that's how they play together. I did forget to tell you guys too, if you guys want access to these skills and the skills that we have that we've shared publicly, I think I uploaded 150 of my own. But these loops, we're going to be sharing these loops in there as well. Just go to skillsdojo.com. I got the domain, guys, skillsdojo.com. I think there's a waitlist for it right now, but you can sign up for it. But our whole intention here is to have it be public for everyone so you guys can share your skills in there. Maybe if I find a skill that you guys have that I like, maybe I'll fork it. I might fork it and then I might make it into my own and I might make it for my own use cases. But we want to democratize these skills out there for everyone to use. That way it makes it easier for you to train your team. It makes it easier for your team to start to use Loops as well. Skillsdojo.com. Then let's I'm going to pull this one up from Jack. Where are you checking with all these requests from your loop? Slack? My Slack channel is— Don't do Slack. If you're going to use loops, you're going to set up all these cron jobs are even hard to keep up with. I wouldn't use Slack for that. I even tried making a cron job channel in Slack where Hermes would report in there or even Codex would report in there. It's too difficult. Again, I like using the Hermes desktop app. I like using Codex if you're going to build the loops in there. I like using the Claude Code app because it shows you, you can see it has a little blue thing, hey, it's time for you to come back to this. Or when it's a recurring loop, it'll show a little clock and it refreshes every single day or every single week or however often it's happening. That makes it easier for me to come back to it. I can see with my Codex that I have 84 scheduled things right now. I probably can cut 50% of them. You want to make sure that you pare them down. every once in a while as well. In terms of cost, let me pull this up. In terms of cost for running some of these loops, I have API usage if I need it. I have 2 $200 Codex accounts like the ChatGPT Max, so $200, $200. I have another $200 for Claude Code. Then if I need to switch to API usage, I will do that. For a Fable 5, for example, as of in 3 days, It's going to be usage-based. Let me cough real quick. Okay. Then the other thing too is to consider is this. The reason why I like using ChatGPT is because they're so generous with their plans because you basically get for $200, you get $14,000 of usage. That's what it's mapped out to be in terms of how generous they are. For $200 with Claude, you basically end up only getting $8,000. I could actually tell the difference. And there was a month where I paid about $14,000 on API usage. So I know it feels just about right. I can tell. And so they can't subsidize forever, but if you're watching this right now, you guys should just go to town on it and max out your usage every single week. You can even set up an eval loop to run through all the stuff that you have running every single day and to basically try to do a token burn before your usage refreshes. That's what I would recommend. That's something that you can work on. So let's see. Let's share this one over here. Actually, let's go to this one real quick. What AI tools are we using to edit our podcast? One, I don't have the best answer for you. I could tell you we're using Higgsfield. I can tell you for some of the B-roll. I can tell you, I think we're using Descript a little bit as well. I can tell you that there's other stuff I would like to use. Oh, we use Overlap for making the Midform clips that we have that we throw onto X and LinkedIn. So that's been pretty good for us. And then we also use Riverside. So you guys are on the Riverside webinar right now. So Riverside's been good. They have an MCP now. And yeah, so have enjoyed Riverside as well. And then there's some AI tooling within Beehive, but it's nothing too special right now. That's what I would say. And then Joe says, you shared how your AI— Okay, yeah, we talked about Skills Dojo. So go to the Skills Dojo if you want the skills marketplace that we have. It's free and we're making a really public version next week, but you should be able to opt in immediately and you should be able to get access to all these skills that we're talking about, all these loops that we're talking about. So let's see. So Anthony says, I think what you're seeing, when you're seeing progress, it becomes contagious, but there's meaningfully time cost to get to the point with the team moving things together, in my opinion. So yes and no. It's not hard for me to— I don't need to push the people that understand this stuff. If I need to push you, you're probably not a fit to work at my company, at least for me. And everyone could be different. The best people don't need to be pushed. And so you might be arguing like, well, it's not realistic for you to have the best people across the board. And I would challenge that because a lot of the people that I'm talking to from a recruiting standpoint right now that would love to join the company, they would join today if we could get them today. And that's how the way I'm looking at it now is like, oh crap, we might actually need to raise money. If I want to get all the people that I want to get right now and also take care of the current people that we have, we need to raise money because we're building the services as software thing with Single Brain. that fits into Single Grain, right? And so I don't know. I was thinking, Anthony, in the beginning of the year, okay, we're just going to train everyone up. And I've realized that that's a fallacy, that you think you could train people to be motivated, that you can train people to be relentless learners, to be biased to action. You can't train that. That's a mindset thing. There's certain things that are much easier to train. There's certain things that are much harder to train. And I think If you're going to make a transformation like that to become a relentless learner, to have a bias to action, to make bold long-term bets, to be super resilient, you have to go through a lot in life and there has to be, you have to be built a certain way. And so I thought we could do that in the beginning of the year. And I just realized this like 7 months into it, we've created conditions for success and the ones that they had it all along, it's been no problem for them. It's frustrating for me to see the other people that don't want to take it on because it makes me feel like I've failed as a leader. But at the end of the day, I think about it, I'm like, well, I can't make them change their mindset. So long-winded answer for you, but that's what I would say, Anthony. I thought I could train it all up. Nope, can't do it. So there's that. Yeah. Anyway, So here's what I'll say. Go check out the Skills Dojo and we're going to continue to talk about loops that we're building out. We're even building out this game right now that maybe I'll show you guys for a second. And then we posted this yesterday to Twitter. It actually did pretty well. And I would say this. So we have this image over here. So Fable 5 just turned my company into a video game. Now, is this live right now? Only I'm testing it, right? So here's a lesson for you. So I made this using Famo.us. You have a marketing hub, a product lab, a sales engine, and you can see if buildings are on fire, if you need to fix things for these buildings. You can also set up quests as well. You can see recent activity. There's a leaderboard, there's active quests. And so you can give people rewards, or maybe if they get a certain amount of XP, they can claim it for rewards, right? But then you could gamify your company. It will also make it very transparent who's crushing it, and who needs help. And for the people that need help, maybe they truly need help and you just need to help them, or maybe they don't wanna work. That's a different problem, right? But we're gonna be testing this internally to see what the engagement looks like. And if you're gonna roll things out, just make sure if you're rolling out skills, if you're rolling out loops, that other people, at least one other person is adopting it so it's not just a bunch of theater. So that's what I would say. And you guys should drop in chat too, by the way, as we go through this, what else you guys would like to see because we're doing these weekly webinars right now. We're also going to bring on guests. I think next week or the week after, we're going to bring on Cody Schneider and he's going to talk about how he uses AI in the context of paid media. And that's something that I don't really talk about myself because I'm more of an SEO background. So let me know what other topics would be interesting to you. Drop it in the chat. And other than that, thank you all for your time, and I hope you all have a good rest of your week. And goodbye.
