AI Business Manifesto Live - September 3 2026

All right. Hello, hello, hello everyone. Uh, sorry I haven't been around the last couple of days. I don't even know if, uh, you're able to see Harry. I guess it depends on which system you are on. I'm trying to just get Instagram up. There we go. So Instagram is up too. Cool. So this is Harry, uh, Harry Lockwood. Why don't you introduce yourself, Harry?

Hello, I'm Harry Lockwood. I'm Rich's right-hand guy, even though I'm on his left right now. Normally I'm on the other side. And, uh, yeah, I help Rich out with all the AI stuff, so I'm here to talk to you guys and help you out.

Very cool. And so we have Sean. Let's see who else. And maybe only Sean right now, but that would be, uh, that would be understandable since, uh, we've been kind of no-showing recently. Um, the, uh, so could you please share your workflow for creating a skill? Well, I think we both do it kind of similar. I'm gonna let Harry answer that one.

Yeah, so, uh, we are a big preacher of don't get caught up in the technical side of it, just describe the outcome you're looking for. Um, so let's say you want to create an email inbox Sorter skill. So instead of asking, you know, what's the best way to make this skill, what's, what's all that, I would just have a conversation with your Claude. Okay, this is the outcome I'm looking for. My email is a mess. Every day when I come to my inbox, I want it to be already sorted and ready for me to look at what actually needs my attention. So what's the best way we can accomplish this outcome? And Claude will be able to help you construct the this skill better than any AI course, any YouTube video out there. And as long as you just describe the outcome and you and Claude are aligned on the outcome, it will be able to create the tool that accomplishes that outcome.

Yeah, then generally what we do after that, after we create it, is we have certain skills that we use in follow-up. Like, um, we have this one called the Prompt Architect, and what that does is it kind of audits the skill that you create, uh, in order to make sure that like 30% is instruction, 70% are guardrails and constraints and things like that. Does that help you, Sean? And hey, we got some more people joining us, so say hello. Don't be a lurker, like give us a— throw us a question. This is my peer and, um, AI partner in crime, uh, Harry, and, uh, we're here to answer questions. And, uh, I figured since I haven't been here for a couple days because we've been busy working on a few projects, um, created a really cool app that I'm looking at right now, um, that is on a touchscreen that like monitors my system. So that's what I'm excited about. Curious what you guys are excited about or what questions you have. And, uh, Harry, I don't know what you're excited about.

Yeah, I see, I see Mark in the chat, Connect the Dots alumni. Good to see you, Mark. Hope you're doing well. Um, but yeah, we, we started dabbling a little bit with GrokBot. Uh, Rich and my friend Donnie has been very big into that, and it seems like, you know, just a simple explanation, it seems like it's the best arms and legs of your AI system. Like Claude and Codex are still the best brain, but in terms of, you know, pure execution and just letting— just simple user experience and letting your AI system go and accomplish objectives. GrokBot seems to be, uh, an all-in-one interface. So that's something that we've been looking into. We have a request to share the new app.

Yeah, we'll see if we can. Um, the— so Mark had a question. He said, when is the necklace coming out? We've, you know, this is a limitless necklace, and, um, it's continuous recording, and we are a fan, but so was Meta, and they bought the company and took it off the market. So this will only work for a limited period of time. And so what we've done is we've gone to China to build better ones, um, and they're on their way. Like, it's expected this month, towards the end of this month, yeah, we'll get the first batch. The first batch, we're only getting 100, and, uh, once we get Our team, everyone on our team has one, uh, plus a few others that like are obligated. Then we go down the list of like maximum profits, then connect the dots, then we'll probably run out. And, um, then from there, uh, we'll be demoing. And once we're certain everything lines up and checks out correctly, then we'll be buying thousands of units. Um, but that's the progression there. Now let's see if I can share this. Uh, let's see if we can share. Uh, we should be able to. It's actually— yeah, let's see. The rail. Yeah, so this is—

Nice.

This is a touchscreen, this little thing that you see on my screen right now. Um, it is that— the screen is those dimensions. And it's touchscreen. So when I touch, like now, like these are the current jobs that are going on on my system, right? And what needs my help, right? Like this project is stalled, but right now this is more in demo mode. And then I also created another screen. Let's see, where's back? There we go. The paper circuit. And this monitors the imprint, but then below I also can look at what's going on specifically with the imprint, like what's being captured. If I'm recording something, I can turn the mic and switch scenes and different stuff. This relates to my streaming system, right? Anytime I have an idea on either page, I can actually just click here, hold the capture. And so I, um, I bought this small display as like, I thought it'd be really cool, but I didn't realize that it's only for PCs. So none of the software worked for it. So I asked Codex, what could we build that would be really cool? And it told me this, and I thought it was kind of cool. So I'm working on it and actually trying to make it even better. But, uh, there we go. So that's the actual, uh, monitoring system that sits right below my main monitor. And, uh, that's what it looks like, Sean. And so what other questions do you guys have?

Yeah, so I see Sean said, uh, you can use the Pocket. It's a speech recorder. So Rich has been using this Limitless device for, uh, like a year and a half now, 2 years, and then it got purchased by Meta, and so they're no longer going to be available on the market. So Since that point, Rich went out and we bought every other AI device, AI recorder, and none of them have lived up to what the Limitless is. The Pocket is good, but you need to toggle it on to record as opposed to this. Like, you just put it on, it's recording all day, and it has like a direct connection back into your system. And if the Pocket was as good as the Limitless, and if there was something as good out there, then we wouldn't have to make our own. But we tested everything and nothing was as good.

So we had to, uh, and the primary reason why nothing was as good is because there are no other, uh, completely autonomous, just keeps, keeps recording, right? Once you turn it on, it just stays on and stays recording. And, uh, because of that, you generally— the best stuff you never remember to turn stuff on for, and we want to kind of avoid that.

Yeah, and our, uh, our Neff is going to be open source, so once we give it to the community any feature that they like, that they suggest, we will be able to easily create that feature and send it out to everybody. So it's gonna be an exciting project.

The, um, yeah, so I'm gonna answer a question, but Harry, I'm gonna ask you to talk people through, um, this, right? Like, so I'm gonna tell Harry, just, I want you to answer this question.

Yeah.

All right, but don't answer yet.

Okay.

Um, so there's this question, right? How are you incorporating GrokBot into your LLM agnostic systems? So, um, that's something I was thinking about last night. And so far, right, I'm just trying to get the, um, the GrokBot fully set up. And the issue is, is that I also have a Perplexity computer. I have like a bunch of other systems out there and making it all kind of like bring together. And one of the reasons why this was pressing on my mind was that the GrokBot wanted to assign a full channel to a project that I actually have a full channel in the Perplexity computer on. And, uh, so at some point, once I've gotten the GrokBot fully set up the way I want it, and then I'm going to take a little bit of time to rewire kind of the highest level of my system to make sure that everything that's being generated in GroqBot is being captured and mirrored somewhere in my Obsidian vaults or somewhere in some file system. Because in the pursuit, right? And so in the pursuit of making sure that my system remains agnostic, but yet also doesn't fall behind, I need to actually make sure that these teams and these playbooks that eventually get generated in these, like, channels all find its way into my system so that not only can the Grok bot work on it, but any other AI on my system could work on it as well and know exactly where the project currently is and what would be the next step. And currently that's not the case. Haven't spent a single moment on it. Because I haven't even fully gotten the GrokBot set up. But like, how I went up to set up the GrokBot last night was first I asked Donnie, because he was using the GrokBot already, like, what got him so excited. And he was talking about a video that Mike Filcine, uh, created that he watched. I didn't watch the video. I loaded it into AI and told it that Donnie was really excited about it, why, and how do we get set up similarly. And so it created a file on that. Then I had it go through my entire system itself and say like, okay, I want you to understand what are all the challenges that I have to be effective, and I want you to incorporate that in. And so then it did a bunch of work on that. Then I had it— I asked Harry to do some research. And what did you research?

Uh, we, we pulled down like Nate Jones and Nate Hurk Got it.

People that we follow.

Yeah.

Uh, for AI. And so that was a markdown file. I added that to the kind of overall, like, processing. And then after all that was done, it created a plan, and then that plan went to the Grok bot, and then the Grok bot started working on it and creating all of the sub-agents, the sub-channels, and things like that. So, um, it's a great question, and it's something that we'll have to do. But I wanna get the system working first, then from there, actually, you know, then I'll actually integrate and get all that information. So, okay, so that answers that question. Now Harry's gonna talk to you about— The memory one. Yeah, the memory one. And I knew that that was you, Lisa, the TOC expert. All right, so Marcus asked this and Harry's gonna answer this.

Yeah, so, uh, you know, similar to, uh, what I said before about the best way to create a skill, you wanna just describe the outcome. Like, how do you wanna collaborate with your team? What is it? What does that ideally look like to you? And then having that conversation with your Claude, you know, most people, they wanna be able to work on the same project without stepping on each other's toes. If somebody, uh, if you're all connected to one area and somebody has a good suggestion, You want to build off that. In the same sense, if somebody has a poor suggestion, you don't want that to affect the master copy. So if you just describe exactly how you want to collaborate with your team, Claude will be able to suggest a bunch of different things. We personally use GitHub. And so the way that works is we created a shared GitHub server, and GitHub is free. And then everybody's AI system is connected to that GitHub up here. So then you can pull down the existing copy. make changes, and then when you want to submit a change to like the master copy, there's an admin who approves it. And if it looks good and like the rest of the team should have it, you just approve it, and then now the whole team can access it. So, uh, that's— GitHub is probably the best, but there are tons of different ways, uh, to do that. So just having that conversation with your quad, um, is the best way to start. But, you know, we use GitHub.

Good job, Harry.

Thank you.

The— all right, let's get rid of this question. Uh, trying to figure that out, uh, right now. Like, I think we're gonna reduce the frequency. It'll be a couple days a week, but it'll probably be Harry and I together, uh, because why not? Uh, Lisa is the expert in theory of constraints. Great to see you.

Um, Oh, I texted you the link for the waiting list if you wanna send it in the chat.

Okay, hold on. That's just my system. Um, where did you— you sent it to me?

Yeah, I texted it to you.

Texted it to me? All right, so let me grab that. Hold on. There we go. Uh-huh. Mm-hmm. Copy. Here we go. Paste. If you wanna be on the waiting list for the best pendant that's ever been made, then you will see a link pop up on your screen if you're on Facebook or YouTube. I don't know if you'll see it actually on Instagram. I guess I can also just put it in Instagram though. Here we go. And, uh, so there we go on that. All right, so Sean, do we know what the cost is?

I think like around $200, right? That's— yeah, that's what we're, uh, figuring out right now, but around $200.

Around $200. And, um, all right, so let's see, what does Jim have to say? Oh, well, glad you said that, Jim. Thanks for sharing. Um, all right, let's see. Can you share the new app, please? Of that app that I was showing?

Yeah.

Um, not fully built, like, out yet. I'm turning it into a DMG file. Um, but we've got, you know, a bunch more to wire, and then I got to figure out if I wanted to give it to some friends, like, with a monitor how I do it because it's very tied to my system, like with the exact stuff that my system works on.

Yeah.

Um, so we'll see. But anyway, let's see if there's any last questions. Uh, if not, we're gonna call it a day and we'll just, uh, you know, the trick is, Harry, leaving people wanting more. Um, so I wish there was another question that we didn't have a chance to answer, but that is not the case. I think we hit all the questions. So thanks for being here. It was like a 16-minute video. Harry and I will be back tomorrow's Friday. Holy crap. Yeah, so we'll be live tomorrow. Why not? Um, unless something tragic happens. Uh, so barring any tragedy, uh, we'll see you guys tomorrow. So whatever we say when we end—

To higher profits and beyond.

Schefren over and out.