Cohort 25 - Session 2 - 2-11-2026

Okay. All right, guys. Well, welcome to day 2. Um, again, today I feel like it goes a little bit faster than yesterday cuz you know, day 1, you know, a lot of people, they take this course and the stuff that I showed yesterday is just like, like a whirlwind. It's really fast. We're building, we're going, we're vibing super fast. And, um, and they kind of need to just like catch up. Today we're gonna be bouncing between different things enough. That it should keep your, I don't know, attention more, more focused, I guess, because I don't know, it's fun to bounce between things. At least that's how my brain works. I like to, I like to go to a restaurant and get like the sampler plate, you know, where I get to try like a little bit of each thing. That just keeps me more interested than just getting one big, you know, enchilada. So All right. Let's, uh, start today first by just, I want to open it up, um, to any questions, any questions from yesterday, anything that you're just like, man, I was working on my project and I couldn't figure this out, or I was thinking about this and think we're all doomed as humans or whatever. And if there's no questions, that's fine. We can just dive right in. Who here worked on their project more than, uh, or like beyond When we finished. Okay. Brittany and Marta. Awesome. How, how long did you work on it, Marta?

All day. I was so excited about it. I kept changing and, hey, the links are not working. Make sure the links are working. Like all the details. I don't know how much I spend on it, but I was like, this is worth it, whatever it costs. Yeah. Because the lessons are just so, so good.

Yeah. I'm, I, I'm sure you didn't spend more than, more than $100 or maybe even $50. Again, $100 should be like really, really heavy use at max, you know, with app testing and all the things turned on. So that's great. Well, I look forward, I would love to see what you've done. Brittany, what about you? How long did you throw into it?

I didn't do it for much longer. I spent maybe about 40 minutes just kind of applying some of those concepts that you were talking about at the end to make sure that I understood and retained some of the information that you were sharing.

Practice makes perfect. I love her. That's great. That's great. And you know what, guys? It's actually, you know, I said this course hasn't changed much in the last year. One thing that has changed, obviously, is the tools have just gotten better, right? A year ago, it was harder to get to like a really good working state, whereas today you can pretty much, if you're connected up to the latest Claude, it's called Claude Opus 4.7. 4.6 is the latest Anthropic model that actually is underneath the hood of Replit. And actually, I'm sorry, Replit still hasn't updated to 4.6. It's at 4.5, but it's still insanely good. But you can pretty much, if you know how to prompt it right, you can give it a single prompt, step away for 3 hours and come back and it will, it'll be done. You'll have a very complex full system. You know, you have to give it a lot of autonomy. You have to put it into that max and you have to say, work on this, build every feature. Don't ask me anything till it's done. "Take as long as you need, get to work," and you'll come back and you'll have spent $71 on that prompt, but 3 hours later it's done. What's interesting is Replit V1, version 1, was able to work for 2 minutes by itself. So you would ask it to do something, then it would report back, right? And it was a lot slower, right? V2 that jump went from 2 minutes to 20 minutes. And so, a 10x improvement from V1 to V2. And again, it was also typing faster, also working faster and more capable and more aware of its surroundings. Its context limits was greater, right? V3 is what we're in right now, and it is capable of working for 200 minutes autonomously, which is crazy. And V4 is any second. I've been seeing hints from Replit talking and saying, hey, V4 is right around the corner or whatever. And I'm part of a beta group anyway. So, but V4, I bet we get it. I wouldn't be surprised if they released it this month. But V4, I don't know what the, you know, I don't know if we're going to 2,000 minutes. I mean, that's crazy to even consider. But even if we went from 200 minutes to 500 minutes, I mean, even just You know, a small improvement, a relatively small improvement that's still over double, right? It starts to really creep in on, okay, do we, do we really need, you know, developers in the way that we have? And we already talked about that yesterday. Replit's growing very fast and it's not just because Replit's good, it's because the underlying model, right? The Opus 4.6, right? That's the same as saying like ChatGPT 5.3 or, you know, just the model number and the latest and greatest of what's possible. The underlying models are just getting much, much smarter. They're just getting trained on more data and there's enough hardware to support it, meaning like the GPUs that NVIDIA supplies, right? If you don't have NVIDIA stock, you know, change that. They haven't even started in terms of what they need to, what they need to do and what they're going to do. If you don't have SpaceX stock, well, you can't buy SpaceX stock, I guess. Quick story. I had an opportunity to buy SpaceX stock in a private secondary sale, and I did. I didn't buy very much. I didn't have very much at the time. It was years ago, and I got the opportunity. It wasn't a big allocation, but they're like, you can put in a little bit. I put in literally $5,000. Okay. About a year and a half later, somebody had some right on the other side of that transaction to buy it back at just a very modest increase. And so I got like $7,000 out of it. I had no choice. I had to sell it. They were there. They decided we want to exercise our right or option. We're going to buy it back. So I went from $5,000 to $7,000. But here's the really dumb thing. About a year after that, that $5,000 would have been worth $250,000. And today that, that $5,000 would be worth closer to about $2 million. I'm so mad about that. I had no choice. I had no— my agency was taken away from me, but whatever. But guys, I bring up SpaceX, I bring up these models because SpaceX right now, one of the biggest problems we have in the AI race is electricity. We don't have enough electricity in the world to go around and and support all of these queries that we're trying to do. And we're trying to replace every human with multiple AI robots, right? And so they're going to need incredible compute power, which takes electricity, right? Those data centers take insane electricity. I don't know how many of you guys, how many of you guys have ever owned like a gaming PC with a really good graphics card, right? Yeah, Marta, I guess Marta's not a gamer. Totally fine. No, no, but, but yeah, the reality is those graphics cards, they suck a lot of power. They're also the same type of card used to mine Bitcoin, right? And if you've ever owned a Bitcoin miner or something like that and plugged it into your house, your electricity goes up $97 a month from that one thing, right? It's crazy. They're very, very expensive. So GPUs are very expensive to run.. And GPUs are the primary also tool to do AI inference, right? It's what's powering your replies that come back to you when you talk to ChatGPT or Replit or whatever it is. And so what Elon is doing is saying, okay, we're going to build data centers and power plants in space, right? We're going to build nuclear power reactors on the moon. And we're going to build orbiting solar panels and solar arrays that will allow them to then harness power from the sun. Because something like, I think the power of the sun as it goes through the atmosphere and gets down to, like, I have solar panels in my field here, and by the time it gets to my solar panels, it's lost something like 10 times of, like, it's about a tenth of the strength that it would be if it was just up in space outside the atmosphere., right? So Elon says, well, let's just take the power of the sun, power right here in space, put the data centers and floating satellites right next to it. And our data speed is more than fast enough now to beam back and forth between satellites. I mean, SpaceX, I mean, I, or sorry, I'm, we, we are having this, I, I'm on Starlink right now. That's my internet out here in West Mountain. They don't have good internet out here. Well, I do because of Starlink, right? So I'm getting 250 megs down and whenever I get up,. But the bottom line is we're going to see space really start to hold data centers and power plants floating around, you know, which is interesting. I think it's interesting. I think it's interesting to know what's going on. As you again have this knowledge gap of, hey, I know what's going on with AI. Hopefully that leads you to figure out what are the picks and shovels, right? To be able to not only directly, you know, take advantage of your knowledge and apply it to your business or your new ideas and those things, but also to be able to take advantage of that industry growing and, you know, look, look down way, way upstream and downstream and all these different things. And obviously you have the, you know, the electricity things and there's much more than just, yeah, there's, there's amazing power companies and nuclear technology and lithium and robotics and all the things and all the microprocessors that are going to power robots that are going to then be powered by AI and, and obviously Nvidia, but who are Nvidia's suppliers? Suppliers, right? Who are all the ones making all the things all the way down that? That's how I personally approach a lot of investments. You know, it's not 100% AI, but it's heavy there for sure. That's the future. And, and, um, it's a great spot to hang out. Great spot to hang out. This is not financial advice. Um, so anyway, um, but, but yeah, anyway, I just encourage you guys to go do your own research, go figure it out, some of the things that I do. And, um, yeah, anyway, feel free to ask questions if you want. On that. Um, all right guys, so I want to kick off today with, uh, with some agentic flows or some looking at some agentic systems, right? Um, how many of you guys can name an agent? Just name, name an agent. A Replit would be an agent, right? So that's one. But can you name a system that is an agentic AI system? And they— and I don't think I would call N8N an agent system. N8N is a, is a workflow and an automation system, but it's not autonomously taking on complex tasks and following up with itself and doing things like that, right? You can build a workflow that can maybe start to act like that a little bit, but N8N is more of an automation platform. We will look at that for a minute today though. Can anyone else name one? Okay. I'm not surprised. I'm not surprised, but I've been using Agentic Systems for like, I don't know, 18 months, and I'm just amazed that there aren't more people out there that use these things. They're so powerful and so cool. They're so cool, guys. So we're going to pull one up today. We're going to start with one called Manus. And Manus is an agentic system that is kind of a general purpose agent. Okay, let me share my screen here. Share. Okay, let me go to Zoom screen. We need to go here. All right, Manus. Manus 1.6 now. Man, that's so cool. It could do really cool stuff. So Madness 1.6. So I, I call Replit my developer, right? I, when I need some code or a website or something on my website or something on my app, or I need to build a tool, I call it my developer agent, right? Replit. That's my, that's that. ChatGPT, if you remember yesterday I said, is my project manager, right? It's going to help me kind of figure out ideas, work through ideas, also strategy partner for sure. But it's gonna, it's gonna keep everything inside of all the projects, right? You can see even, even this has projects and I haven't even ever used the projects tool on Manus because it's usually just like one-off things that I needed to do. But, um, but this one I'm going to call Manus my assistant, my executive assistant. And it's an executive assistant that can do almost anything you can imagine. If a human can do it at a computer, so can Manus. So let me give you an example. Let's, let's just look at a couple examples I've done here to just kind of show you something that might take a really, really long time. Here's a great one. Here's a great one right here. So I actually had somebody reach out to me on LinkedIn and they said, hey, will you— or I want to, you know, take your AI course, you know, this one talking about this one. And they literally sent me this message after that. I said, I want to take your course because I want to figure this out. They said Utah has an HOA registry at this address. If you go there, you can search for HOAs by name, but you just put one letter in and search and it'll put everything that has that letter. And what happens is on that, I went and looked at it. When you look at it, there's thousands and thousands of results. And, and it's really bad because remember yesterday we were dealing with that, uh, we were trying to build my drinkable thing, right? And I would click on a drink and then it would take me back to a list and it would start at of the list, right? That's how this website was also made. And so you had to click on it and then click on another thing and click on another thing, and then you had to restart the search and click reconfirm submission and all this stuff. And it was— it would take— it would have taken probably, for a human to do what he needed here, it would have taken probably somewhere around 2 weeks, about 80 hours of work, I, I calculated. Okay, but he basically said, hey, you know, I just need the names, addresses, and email and phone numbers of these people. Can we pull that into a spreadsheet? So I literally just copied— this is what he sent me. And then I added eliminate duplicates, provide a long and detailed CSV of all of this information, or all this information. Okay. So it gets to work, starts figuring out the, you know, what's going on, what's the structure, how do I make this happen? And it got all the way to the end and said, sorry, you know what? I've actually, you know, um, fill, you know, I don't, I don't have enough, uh, room to keep doing this. The context is too long. Right. But. You actually, and it doesn't show this here, but when you get to a task that's too long, you can actually click a button that says inherit and continue. So it will inherit all the stuff that it's figured out already and just grab that as a small piece of data and then move on and it knows where to pick up, right? So it just gets a little bit of information and then moves on. And so it moves to the next task, did it again, next task. So all I had to do was literally paste that request into there, paste that request into there, and then click inherit, continue 3 times. Actually took quite a while. And by the end, it delivered me this. Wait for it. Boom. There you go. Right? All the emails, all the names of all these HOAs, everything done. I don't know why this is saying file not found, but anyway, addresses, emails, phone numbers, the right contact, all these different things. Unbelievable, right? And I asked it to eliminate duplicates. It obviously found a ton of duplicates as it went through here. But pulling all of this data into a spreadsheet, now he can, you know, I don't know, hook it up to MailChimp or start calling or hook up an agent to start calling it, which we'll show you guys how to do today. All those types of things, right? But the point is it was able to take a task that should have taken, again, an executive assistant full-time, 2 weeks, right?

Not cool.

This took— I have a question. This took about an hour.

Yeah, go ahead. With the agents, do you have to grant permissions or are there limits to what it can access? Like if I said specifically, I want you to go make this into Google Sheets, would it be able to do that?

Or, and with other websites? Yeah, so you can, let's see here, where's the— oh, they've moved the— where's the main— sorry, this. I think it's maybe, there we go. It's the Zoom problem here. But yes, you can actually connect. If you look at connectors right here, you can connect your Google Drive, right? And you can also connect it and allow it to access your browser, and then you can add other connectors. So I've, I've, those are the two that I've connected, right? But you can access all of these different things and have it control, read and write Slack conversations in Manus. For example, every day we receive this genai.pi.org AI newsletter., right? And what it does is it has a, a, a task basically every day, go out and see what is all the news of the day related to AI. And so we receive this thing to our, our channel and I, I'll then go through and I'll just, I'll just quickly read these bullet points and then maybe I'll comment on something I saw, for example, today that the OpenAI hardware, Jony Ive, the Apple guy, the Apple designer, right? Had, you know, he joined, joined OpenAI to create AI-powered hardware, right? Whatever that was gonna be. Looks like they've delayed that. It was supposed to come out early this year, and it looks like we're back at least a year. And so I was like, yeah, whatever. But the point is we, you know, we, we stay on top of this, but this is sent. See, sent using Manus. This was sent automatically and it's sent every single day. It used to be that Brantley on my team, would, would go in, run the search, copy it, throw it into a document, and then paste all that stuff over here. And now he doesn't have to do that. Right? He just has a task set up that automatically is connected to Slack. It's automatically connected, does the research and all these different things, right? So, we're constantly talking about news right here and we've actually, Tyson, you should be receiving emails, for example, that actually take this data and we find the most important things for your business specifically. You should be getting an email every Monday and Wednesday. You should have gotten something this morning on this, right? With, and so, we have that then connected up to our clients that can then also receive these same AI updates and it's extremely useful. It's one of the most useful things I get every day in our clients. Have, um, have really enjoyed receiving that data. But that is a complete automation that's set up with, um, with Manus connected into Slack, connected into, you know, whatever is doing the research and all that stuff. Okay, all right. Ask a question. Yeah, we'll do Vardeep and then Marta. So just for example, I'm going to say pull me a list of all the roofers in Vancouver. And create a spreadsheet. It can do that 100%. In fact, here, yep. Find— I, all I said here was, hey, find me 20 businesses in Utah that could potentially hire me as their fractional chief AI officer. You can see our offering here. I wanna go— I wanna aim for companies that have 30 to 100 employees, whatever. Right? And yep, here you go. Right? Nectar and phone number and email, whatever. Job Nimbus, grow.com, Artemis Health, Calendly. You can ask it for whatever you want, right? You also, guys, you can use this Manus interface. You can also— what I really love is it has this ability to mail Manus. So I don't know how well you can see this. I'll make it a little bit bigger here. But right, I can— I basically email joncheney7485@manus.bot. And these— you guys can't email it because you're not an approved sender. But if I email from one of these email addresses, then it will just respond and it'll just be like, yep, I'm on it. I'm done. And, you know, just get on the task and do it. And as soon as it's done, it'll email me back with all this stuff. Like almost like a person, right? It feels extremely like, man, it, it, if you ever receive something like Tyson or something or, or whatever, if you're, if you're, you receive some big report or something on how the company's doing or some big, you know, something that's really unorganized, just forward it onto your Manus bot and just say, Hey, will you put this into a one-page summary and, you know, point out anything that I might need to pay attention to?

Right?

And then you're done, right? You don't have to like copy it and download it and paste it.

You do that every week. Yeah. Forte, I said. Thousands and thousands of jobs helps me prioritize AI all the way. Tell me what's hot. Tell me where we need to focus. What's open? What needs coverage? Oh, it's fantastic. I used to spend about 8 hours a day.

Getting that report ready. Yeah, and now you're using it. Yeah, so I can set up— see, you get a list of all the roofers, I can send an automatic email to ask for the business or whatever automatically through email. Well, this is not going to email those roofers for you. Well, actually it can because you can connect Gmail. You actually can have it email for you, and so you'd have to set up some tasks and things like that. And there's different, you know, I personally don't love having AI email clients unless it's something like this newsletter thing, right? This is like, hey, we're just reporting on some stuff and it's pretty harmless, right? But if I need to actually connect with Tyson on something as an example, I'm going to email him, right? I'll use, I'll use, so I use one called Blockit, right? Blockit is amazing. It's a cool app. I'll pull it up here. Blockit.com. So, Blockit.com is simply an AI scheduler. Okay? So, you sign up for it, you connect it to your email and your calendar or whatever. And then what you can do is you can have it, you know, rearrange your schedule for you if you need to. Or if Tyson reaches out to me and says, hey, Hey, I need to meet. I have some time Tuesday, Wednesday, whatever. And I'm like, okay, I'll just say block it. Will you find a time and send a meeting out? And it's connected to my Zoom, so it has all those links and whatever. It will automatically email Tyson and me and whoever else needs to be on that call and say, hey guys, looks like Jon's available at 9 o'clock. Does that work? Tyson says yes. Block it automatically sends out the invite and everything. I don't have to do anything. Right? And so again, that's a very transactional thing. We're not talking about the meat, right? We're talking about details around the edge. And I really, really love— personally, I love AI helping with details around the edge. When we're getting to the meat, like, for example, when I post to LinkedIn thoughts about AI or thoughts about whatever I'm talking about, right? I post to LinkedIn every single day, and I spend time. I spend a lot of time on LinkedIn. We're going to talk about this later. When we get to marketing, things like that. Right. But the things that I talk about are never, ever, ever written by AI. Because for me, this is the meat. This is what I'm delivering. This is the value I can give to the world. And so when I go in here and share whatever it is about, you know, whatever it is I'm thinking about that day, it's me. Right. There's no AI in here. This is 100% me sitting down, thinking through, thanking people, talking about specific things, you know, whatever. Okay. And so Blockit again is a detail. Manus sending out, you know, AI updates and things like that. We've really massaged that to where it's really pulling good content, but it's really just kind of reporting facts as opposed to having to, you know, figure out how does this really, you know, what does this really mean and what should we do next and what's the Plan. And anyway, so you kind of have to just decide where you're gonna draw the line. By the way, I receive emails every single day from AI. Totally. I know it's written by AI. I receive comments on my LinkedIn posts every day written by AI. I, we just go to my LinkedIn right here and just start scrolling down the main feed. We're gonna see AI everywhere. And quite frankly, I hate it. I don't care what AI has to say. I don't, I don't care about AI's opinion. I want it to help me figure out facts, and then I want to figure out what that means right now. I can have a discussion about what that means with AI and things like that, right? But, but if I— yeah, see, this is not AI. I can, I can, I can identify it instantly. I want to just scroll for a minute here and see if find something written by AI. Not AI. It's about AI, but not AI. Not AI. Yeah, see, you're— yeah, again, we're not seeing— yeah, this person, definitely not AI, complaining about AI. But anyway, it's— that's not not AI also. You can just feel. How many of you guys think you can feel when something's written by AI? Yeah, most of you guys, we can see right through it, right? And, and you guys aren't, you know, even spending all day every day in AI like I do, right? But you can see it and you can feel it. And how many of you guys like it? Like, are you with me? Like, I, I just don't like it. I don't, I don't want to see it. I don't want to see it. This is AI for sure, 100%. It was probably, it was probably written initially by a human, but then someone used AI to actually write it out. That's what happened there. So you have to kind of just decide, this is kind of a personal decision thing. How are you gonna use it? Where are you gonna draw the line as to what you do? But, but man, I can tell you that I don't feel bad even for a second having AI go and do a bunch of research for me. Right? Or going and moving a bunch of things from here over to here in this document. Right? Or like moving something off a website or scraping something or trying to figure that out. Here's, here's a crazy one. Let me, let me share another one. I think that's really cool. So I got hired last summer to work with Tony Robbins and Dean Graziosi and their, you know, they did a big thing called AI Advantage, aiadvantage.com. Guess how this site was built? Replit. You guessed it, right? So I helped them do this whole thing. I helped them build the whole thing and help them build out their whole offer and they're doing all kinds of stuff and they're teaching, they, they do, they do really high-level courses. Nothing as deep as what you're sitting in right now. And it's, they've got $30 things and $4,000 things to get coached on AI and it's fine. It's fine. It's good. But, but anyway, they hired me to kind of help them put this together and to figure that out, to help them with doing some AI at their company, things like that. And at the very beginning of that, kind of towards the beginning of that, right as they were trying to figure out, they were kind of just about to say, okay, hey, audience, we're going to start talking about AI because they hadn't really talked about AI much other than the kind of offhand, but they're like, hey, we're going to come out with an offering. And so Dean Graciosi, the guy in this, I don't know if you guys know who Dean is, but he's Tony's partner here, right? Dean and Tony. Yeah, Dean and Tony. So anyway, you guys know that. But But anyway, he said he did a Facebook Live and, and he started talking about, it's like, hey, tell me, you know, what do you guys think about AI, right? What's going, what's going on? What's going through your head? What do you want to learn? What are you afraid of? Whatever. And so he then, you know, I was on that live and he texted me like, hey Jon, will you, will you go into the comment section and like report back what, what are people saying? What is, what's going on here? And I was like, yeah, absolutely. And I'm like, there's no way I have time to actually do that, um, because there were like 1,700 comments. I wasn't spend all day like copy pasting and then considering stuff, whatever. And so I just, I just copied everything, right? It did a big long copy, threw it into a Word document. This Word document's, I'm not even going to try to open it, right? But it's like 78 pages long. Um, it was tons and tons of comments. I just said, hey, review all this on Facebook. It's poorly formatted, but data's all there. Pull the right stuff out. What are people thinking, right? And so I'm not kidding you, 15 minutes after he gave me the task, I sent him this link. I sent it to the whole team. And as a website that Manus had put together, it had reviewed everything. Here's the sentiment distribution and the breakdown, executive leadership, learning interests, you know, prompting and communication. Here's the stuff that people are talking about. Here's how much people are using AI. Here's some of their fears and excitements, right? Here's some specific quotes. Here's some key quotes, whatever. And pulled 335 comments that were meaningful. Something that's not like, hey Dean, hey, you know, something that actually meant that had real impact. And he was, Dean of course, was just blown away. Number one, because of the speed at which I was able to do that, which was almost instantaneous. And number two, the quality of the delivery, right? How many times have you had someone build a website for you to give you a report?, right?

Like probably never.

Did you prompt, um, Manus to do it or did it know to do it? You did? Okay. I, I asked it to do it. I said, yeah, please. I, I, I said, please make a website of this. Let's see here. Uh, yeah, I said, please put this, all of this information together in a website with interactive visualizations, bar charts, pie charts, and graphs that help illustrate what this group is feeling about AI.

Right.

Mastermind, is that your website? Mastermind.com or— so yeah, mastermind.com is the, uh, um, is, is the group, is, is is really Dean and Tony's big mastermind group. And Mastermind actually is a platform you can use to teach your own audience, right? So, for example, I could be a customer. I could be teaching this course to you guys through the organization and structure that Mastermind offers. Or Kajabi is another one that would be a competitor to this, like some of these course delivery systems. That's kind of what this is, and it's really just a course delivery system., right? And so, yeah, you choose your plan, you pay whatever you're paying a year, and you get access to courses and all these different things, whatever. Anyway, so, but yeah, anyway, so, so that's, that's cool. And they have Tony and Dean, they do, they're amazing, right? We, I, I helped them with the launch and I was actually part of the launch. I was, you know, I was, I was on, on stage in front of about 140,000 live people and then millions of people saw that and really helped them get this out. But they did $15 million of revenue in 3 days selling, you know, the first kind of offering from this AI Advantage thing. And, and anyway, we're still partnered with them. They actually, these guys don't want to do the, the kind of higher level type of work that we do with clients. Like, like, like we work with Tyson and Brittany here. They just want kind of the easy automated course. They want to sell those things and scale it and never have to pick up their fingers again. And so we've been, these guys are now a channel into us, right? Which has been really fun. So that's part of how that relationship is continuing to develop and grow.

But anyway, that's how I found Jon. I was on that AI summit and then he said he was from Provo. And so, yeah, same Eli.

That's cool. No way. That's awesome.

Yep.

That's how I found him. I messaged him right on LinkedIn, like 2 or 3 minutes into his speech. I'm like, thanks so much for your entire speech. I love the whole thing. Thing, right? As I'm still listening to him. And yeah, he replied back with, go Utah. And here we are, man. That's great. That's awesome. Jon, would you be willing to share like this prompting conversation for pulling all that together?

I actually have a couple different use cases that I'd like to customize that prompt for. You're saying this one right here, this prompt? Yeah. Oh yeah, totally. I'll copy and paste it right here. You got it.

Appreciate you. There it is. I'm also curious, on the left I noticed there was one that said who's going to win between the BYU and Utah football game.

Was Matt as accurate with that?

Let's see.

That one was funny. Yeah, I saw that. Yeah, I shared that one on LinkedIn at that time. Yeah. But yeah, this was— yeah, here's the prediction. Called BYU. And by the way, I don't remember what the score was. It was close to this. Let's see here. BYU, Utah football, 21-24.

Yeah.

So it was a little closer than Manus had predicted. But if you were to look at why all this stuff, the turnover battle, right? Like it got all of these predictions right. It's pretty cool. Pretty cool. Yeah, it'd be fun. If I was a gambling man, I would use AI to do all kinds of sports gambling and stuff and predictions. And I'm sure there's plenty of people doing it, but probably not enough. Another thing I've done with AI is I've asked it to do really deep research, and I'll say something to ChatGPT like, hey, I want you to go find 10 stocks that have 10x potential in the next 3 years, right? From now to 3 years from now, what is out there? Said, don't worry about the risk levels. You report the risk levels to me. All right, like, how crazy is this? Like, I don't want crazy penny stocks, but I kind of gave it some parameters.. And then I just said, okay, I'm just going to trust it. And so I bought $1,000 of each one of those, of each one of those, just to see how it would do. And that was about 4 months ago. That $10,000 is now $26,000, right? I just bought it, held it, and I'm just going to keep leaving it and see what happens, right? But, um, and not all of them have gone up, you know, like some of them are down, but some of them are up a lot. And, you know, but overall I'm just going to let that portfolio ride and see what happens there. So I love playing. I love learning and experimenting with AI and seeing where I can push it and how it can help me do the things that I'm doing already, right? Which is I'm investing in stocks. I'm working with people and I'm a BYU fan and whatever it is. How can I bring AI into this stuff? Sorry, I need to get out of this. I'm like stuck here. There we go. There we go. I couldn't get out of that.

But how can I bring AI in to help me just do those things better?

Right.

Can I ask a question? Yes, go ahead. So my question is a difference between ChatGPT or Grok and this Manus, because I use ChatGPT pretty much the same. I didn't know there is such app. So I use ChatGPT and Grok to do everything for me just the same way as an assistant or put the spreadsheet together, analyze my real estate deals, you know, all of that. So am I using it wrong? Is this more powerful in analyzing or picking stocks? Like I would never trust ChatGPT to pick stocks for me because it wants every, Like it kind of wants to make me happy all the time. It knows what I like and it will suggest everything I like. I feel like I don't trust it.

Like how do you trust that they're predicting correctly? So two questions that you asked there. One is, is man is better? And two, can, how can you trust ChatGPT? And you're absolutely right to be concerned. I'm gonna address the second part first and then we'll get back into the agent discussion. But you're absolutely right to be worried about to be worried about that. And when I say that I have ChatGPT go analyze those stocks for me and then I go purchase them, it's not just a single prompt, right? It's not just, hey, do this. I'm having discussions with it. I'm having it validate and verify and share its logic and reasoning. And then I actually have to decide, do I trust its logic and reasoning? Also, it's possible that you were working with ChatGPT 4.0. Right? Or you were working with some other model, but ChatGPT 5.3, which is the very latest and greatest model, is insanely intelligent. Absolutely ridiculous. They haven't done the measurements on it. 5.2, ChatGPT 5.2 had a 155 IQ. And so it's estimated that 5.3 will probably be somewhere around 180 to 200., and within about 2 years, we should have an IQ of somewhere around 1600. Okay. That's not even possible to even comprehend, right? We, our, our brains don't even understand what that means. But what I'm saying here is AI has in the past done a lot more hallucination than it is doing now. Mm-hmm. You know, a lot of people, a lot of people right now, you see comments online like, oh yeah, I tried AI, it's not great. Or you talk to someone, maybe you'll talk to someone in your family, Marta, back in Poland, and they'll say, yeah, AI isn't great. And you're like, well, when was the last time you used it? Oh, I used it last year for something. But that's an eternity in AI development years, right? I mean, it's the speed at which we are progressing. Again, the Replit V1 to V2 to V3, those 3 versions happened in 18 months and it improved 10x and then 10x. It improved 100x in 18 months. That's crazy. I want to do a little calculation here with you guys. Does anybody know what Moore's Law is? No. Okay. Moore's Law is basically a pattern that all technology, like this in your phone, right, where you've got processing power and hard drive space and and megapixel capabilities with your cameras, things like that. Basically, the capability of technology in your computer, in your desktop, in your laptop, whatever, doubles in capacity and capability or cuts in half in price. Right. It becomes twice as available every 18 months.

Okay.

That's Moore's Law. And you can go back to like 1960. And follow this pattern. It is just a steady line. It doubles in capacity every 18 months. AI is 10x-ing every 12 months. That's really what's happening across the industry. So let's examine what that really means. So if your average iPhone, right? How many of you guys get the new iPhone every single year? Any, any of you? I do, right? Yeah, I'm— I love it, right? Marta, Marta and Bardip, you're my people, right? I want to have the latest and greatest. But how often do you get it and you're like, yeah, I mean, the orange color is cool, but like, it's not crazy different from last year's phone. You can kind of feel it. But if you've ever waited 2 or 3 or 4 years to upgrade your phone and then you do it, you're like, whoa, this is way better, right? You can really feel that, right? It's because that doubling hasn't even happened in one year. Right? It's, it's 60% better and, you know, the fastest iPhone ever, right? Every single announcement. But, but anyway, the point is Moore's Law says that normal technology will double in, in capacity every 18 months. That means in 3 years— let's extrapolate that out to just 3 years, okay? Every 3 years, the iPhone and your, your standard tech is going to quadruple. Right. It's going to double twice. Right. It's going to be 4x better. Now, if we look at AI, we're going to 10x the first year, we're going to 10x the second year, and we're going to 10x the third year. Who here is good at math? What's 10 to the power of 3? 30. What's 10 times 10? 100. What's 100 times 10? Yeah, uh, 1,000. 1,000. It's 1,000 times better in 3 years as compared to 4 times better in 3 years. It's literally 250 times faster. It's progressing 250 times faster. That's why our brains— that's why Eric Schmidt was saying we don't even have language to describe this because we've never experienced this kind of technological growth. Imagine if this got 1,000 times better in 3 years, the battery lasted for a year, and the camera could zoom in from here where I'm sitting here all the way so I could see people clearly on the Y, way across the valley from where I'm sitting here, and the internet speed wasn't 1 you know, you're sorry, 100 megabits per second, but now all of a sudden it's 100 gigabits per second or whatever. What would happen if we were having that kind of— it, it would be absolutely mind-blowing. And of course you would upgrade every year, right? And so again, this is a long-winded way, Marta, of saying AI is much smarter today than it was a year ago. Or even 6 months ago, or even 3 months ago, or even in some cases 1 month ago, right? As soon as that new model comes out, you're looking usually at 5 to 10 to 20x stronger capabilities. And so keep doing it, keep making decisions for yourself. And you also have to train your AI and you have to push back on it. If it says, yeah, this is a great idea, whatever you'd like, but is it actually a great idea? Tell me why this is a bad idea, right? Like if you ask for those— when I ask for those 10 stocks, I said, tell me why it would be a terrible idea to invest in these, right? Tell me the exact opposite. Again, I'm asking for that from that model. I'm asking for the MSNBC and the Fox News, right? I'm asking for both. And then I have to use my brain to fill in the gaps. Okay. Don't let AI do the thinking for you. Let it present information. You then verify, and then you get to decide how you act.

Right. Yeah, you know, I took another course and the guy said that it's not capable of predicting just because it can't understand the graphs or whatever, something. I don't remember exactly, but he said do not ever do predictions on stock market based on AI. It's not capable of doing that. But I think maybe this is a different approach of instead of seeing if it will go up or down because markets shift based on different scenarios, right? But maybe your perspective is like based on where we're heading in the future, I need you to pick stocks that will align with the future and where we're going, which is a little bit different than predicting if Microsoft will go up or down because there's too many variables.

And the guy that I took the course from, he said it's not capable of doing that. Yeah, I disagree with the guy that you took the course from, and I don't know when that was, if it was a month ago or if it was a year ago. Obviously, if it was a year ago, I understand why he said that. If it was a month ago, he's just wrong. He's wrong now. Yeah, he's wrong at this point. Okay. It 100% can predict those patterns better than humans can. And, and not only that, you are exactly right. The 10 stocks that I invested in were actually very specifically, I said, because of AI and robotics and quantum and other stuff, I want you to look upstream. Who are the providers of the core materials that all of these companies depend on? So, again, I was using my brain to not just randomly do this. I just said, I need you to help me identify. It was more doing research on What are the stocks that have, you know, that are in there? Okay. Out of those, what are the ones that have the greatest potential? They're, they're kind of newer or maybe, you know, based on the demands that we're going to need for power and the demands we're going to need for AI and chips and all of these things. Well, if this is their primary supplier, how are you know, obviously this is going to have to 10x, right? I mean, that was the kind of research I was doing. So again, I'm still being a smart, savvy investor on my, you know, on my own. Not saying that I'm great. I've lost plenty of money in, in, in dumb stock decisions. And, you know, had bad luck with SpaceX and things like that. But the point is, um, you're the boss, you're the one in control. But I would vehemently disagree with, um, with the person that you're, you're quoting here saying AI is not capable of looking at graphs and identifying those patterns and predicting the future. Neither is a human. But it is as capable as a human at this point at predicting the future. It really is. And so take that for what it is. Make your own decisions, your own— again, we're talking about investments right now, but this applies to any topic, right? For example, let's move over to law for a second. I think law is a really, really important, fun one to cover, especially if you're in the startup world and you need to save money and you need to do whatever, right? I initially, well, before AI, right? I'm hiring law firms every time I'm doing an employment agreement or an operating agreement and I'll send it over and, you know, wherever. And I got to the point where I've had templates. I've started up businesses where I'm like, okay, I kind of want to start from here and I'll send it over to them and they'll tweak a little bit and then fix it, whatever. Then I found a lawyer a couple of years ago. So as soon as I started using ChatGPT to start writing some documents, and this is back when it was like ChatGPT 3.5, right? I was able to get to a decent first draft because it can follow patterns, right? But it doesn't understand everything. It couldn't understand everything, at least not back then. And, but I found a lawyer who I said, look, I'm not gonna pay you $2,000 to help me with this operating agreement. I wanna pay you $200 to check mine, right? Or whatever, right? You know, those aren't the exact numbers, but the point is I'm gonna use AI. I'm gonna send you my draft. You're gonna then fix it and, and that'll be how we work together. And he's like, great, let's do it. Right? And now today I was read, I read a long, like 20-minute article yesterday that was digging into this. And there are now senior partners at law firms that are using ChatGPT 5.3, the very newest model that's out to completely replace junior level associates. I'm not talking about paralegals. I'm talking about full-on lawyers who've been to law school. They, you know, they're junior level associates. They're, Typically, you're paying them $200,000, $250,000 a year to do their job. Now they're completely 100% replaceable by an AI because it's getting those documents 100% correct. If I send a contract over to somebody now, I don't feel like I need to go to a lawyer to ask them about it. I'll pull up— again, I have lots of great documents that I've used over the years that were written by really smart lawyers. And so I can usually start from a great place. And then I'll tweak it or I'll have ChatGPT just, you know, modify it for my, my specific need. I'll send it over and, and I'm good to go. Now, if I'm doing a major transaction that's gonna be, you know, a $4 million transaction or something like that, I'm, I'm gonna hire a lawyer just to just double check. And I, I gotta cover my bases. I gotta make sure I don't do something stupid. But I'm still going to do the same thing. I'm going to have ChatGPT do the work. I'm gonna say, Mr. Lawyer, here it is. This is what's at stake. Here's what's going on. You do your check and, and we'll go back and we'll, and we'll work on it. But, but, so again, good use of AI is not just handing it over. It's using human judgment, your own expertise. You guys are excellent at all of the things that you know how to do, and then using AI to just speed it up. And now AI is getting to the point where it really can do it as well as any human out there. And, next year, or maybe by the end of this year, better than any human. We get to that, that AGI, right?

Artificial general intelligence, where AI can do every task better than every human.

Okay.

While on the subject of this trust and ethical stuff, is it okay if I ask one more question?

Please do.

Um, so I was just wondering with like how you go about considering what is stuff, like what what is ethical and what is not. And I know, like, 'cause we're like religious and we, our religious leaders have provided some guidance on that. But like in specific, like for example, my first LinkedIn post, I, I wrote it myself and then I put it in chat and I was like, how can I improve this? Or is this like good? And then it was like, not really. And so I just like changed it. But like you said, it didn't really feel like completely mine, but like how do you like figure out decisions like that? Like, okay, I'm gonna write all of my LinkedIn posts on my own, or this is something that I will do, or I will never email people. Like how do you decide those factors?

Like when it comes to the specifics? Yeah, I love the question, Eli. I think it's very important and you're coming at it with all the right, you know, things in mind. Like what does this mean for me personally? Me personally in my personal development and growth? What does it mean? What are my religious leaders telling me and the things I believe in? How does that work? How do I reconcile, you know, all of this together? And I think that ultimately, guys, we are, um, you know, we all have to kind of make our own decision here, but, but the, the biggest mistake I believe we can make with AI is giving up our, our agency. Giving up our ability to choose what we should say, how we should say it, where we should say it. If we just say, all right, I've got an idea, give me a post about this, boom, and it just goes and does it. Right. I don't think it's wrong to then post that, but you have to make dang sure this is where the ethical line is crossed. You have to make dang sure that every word that AI wrote you actually believe 100% that you can stand behind what it said exactly in the way it said it. Right. And that's beyond just the formatting of it. It's not this, it's that, right? I mean, it's got the telltale AI signs that you can recognize so we can all quickly like suss out. We know that it's not, you know, that it was written by AI. You can actually have AI and say, don't use any, you know, catchphrases or things that are, you know, things that are typically used by AI, make it sound more human. And you can actually get it to a point where it sounds pretty dang human. You really can. But ultimately, is what it's saying something you really believe? That's one element of it. I think maybe the more important element, and this is where you then have to figure this out, is are you sharing something that you actually understand? Right? If I don't use AI,, and I just— all I have is my brain and a typewriter, right? I don't have anything. I can't search the internet. I can't do whatever else, right? I'm just— what's in my brain? Do I really possess the understanding behind what I am saying? Because if I'm posting something that I don't completely understand, well, we all know what happens when that, you know, fake news and propaganda and all kinds of just crazy things, and people try to jump into places, you know, how many of you guys love to hear political commentary from athletes and movie stars, right? Like people don't really love hearing that. Why? Because it's just not their expertise, right? If they want to give me, you know, tips on acting, dude, I'll tune in, right? Like that's cool. But as soon as they start going off, I'm just like, yeah, I mean, you know, it's cool. It's your opinion. That's great. Everybody's entitled to that. But you just don't know a lot about this and they'll just jump in and they might not know everything and they just see some random post that something happens and they go off and they're talking about things that they don't really understand. But then they use loud voices, they're loud voices, and then they can cause problems because now they're pushing things they don't completely understand. They're saying an incomplete truth, which they might not be doing on purpose. They're just trying to help and say something and whatever. And then it causes controversies and people Bad things happen. Bad things happen. Charlie Kirk. I mean, there's a super way down the line of all kinds of bad things happened in this person's life that led to a tragic accident. Right. And so I believe that kind of consequence is possible if we aren't careful with the things that we're saying. And so we just have to make sure that we understand it when we say it. We need to know why we say it and that we believe it. And, but then use AI to get there, right? Like, I have no problem with saying, hey, I want to talk about this. Let's talk about this. I want to, I want to say something about this issue, or I want to teach people about this thing and go to ChatGPT and say, hey, I want to, I want to say this. And you can even say, hey, here's my post. Let me share a quick, I'll share a quick story that illustrates exactly how I would use it. So my son was asked to speak in church. I don't know, a few months ago. And he's 12, right? He was 12 at the time. He's 13 now. But he was asked to speak in church and I don't remember what the topic was. I have no idea. But he brought it to me and said, hey, I wrote my talk. And I said, great, let me read it. And I read it and I was like, this was 100% written by ChatGPT. And I said that to him. I was like, hey, did you use ChatGPT to write this? He's like, yeah. He's kind of like embarrassed. Hey, good for him. 12-year-old figuring out how to get something done. I didn't want to be mad at him for that, but I said, let's try this again. I think it's great to use ChatGPT, but we want to make sure you really understand and believe this. So I then turned on— I opened up my phone and I just turned on— let's see, not do this. Someone's trying to reset my Mac password. That's good. It means somebody's trying to hack me right now. I just turned on my Notes app on my phone. Okay. All right. That's kind of like crazy bright. But anyway, and I turned on the dictation, the little microphone right here where I could just talk to it, right? And then I just sat there and said, okay, Carter, what do you think about this? And how has this helped you in your life? And, you know, do you have a story about this maybe that helps? And maybe do you have a favorite scripture or something that you want to share that's related to this? And what do you think, right? And we talked for about 5 minutes is all. And I just recorded the whole thing, right? Just transcribing it. And then copied the whole thing. And I said, hey, ChatGPT, um, this is what we did, this is the topic, um, I want you to use only Carter's words and I want you to rewrite this entire time. When you kind of organize it in a way that puts it in a good, you know, a good, you know, way to, way to present this, but only use the things that he said, right? And the only thing you can correct is like grammar., right? Because the transcription might not be perfect, but you know, it, it needs to be his words. So he did that. It took, you know, 30 seconds, whipped it up. And then I said, okay, Carter, now take this and read it to me. Right? And he read it and it sounded just like him. It was the things he had just said, organized nicely. It had helped kind of organize some of that, that stuff. And I think that's a great way to use it. So you can definitely organize, you can definitely have it edit you. You can have it, you know, figure out what is the best way to, What is the best way to approach this topic? But then I believe we grow from putting in the work. So whatever that work looks like to you, do it. Really, really good conversation, guys. I think this is— I think this is so, so critical. And I think people are trying to have this conversation out there. I don't think we're the only ones having it. And I think that's good. We got to figure out how to how to keep AI doing AI things, the robotic things, and letting humans do the human things. And one of the things that humans need to do, we must do in order to survive, is we must put in the work. Can't let robots take all the work. That doesn't mean we have to do laundry all day. We can have robots do our laundry. It's totally fine, right? I think you guys get it. You got to put in the work. To grow, right? You grow through adversity, you grow through trials, you grow through failing, you grow through all these things. Then if we just let AI make, turn our life into easy mode, you can choose to do that, but go watch WALL-E and see what happens. Okay. All right. So man, this is cool, guys. Use it as creatively as you can. And you can also say something, by the way, along the lines of like, hey, go to delta.com and buy me a ticket. And it can, and you know, here's my password, here's my username, whatever. Log in, do it, and it'll go and it will buy it and it'll send you the email, say, hey, I did it for you, right? I mean, you can give it permission to be able to— it can operate websites, right? Just like a human can. Okay? It's very, very capable. It can do just about anything a human can do. And it will also do some things like I gave it a crazy task. The last thing I'll share here, I don't even know if I'll be able to find it here, but I've taught this course, right? 25 times now. Back when I had done it about 18 times, I said, hey, I wonder if I could, like, I wanna, I wanna write like a, a really extensive guide, like on exactly the types of things I teach in this course. But that's hundreds of hours of transcripts, millions of words, right? And so, but I said, hey, why don't you go, go to YouTube, extract the transcripts, summarize it all and try to put it together in a guide. And, and that would take, that would take a normal human, months to do. And it was able to do— I think it was able to do all right. I give it a C on its ability to do it. I would never publish it because I didn't like how good it was. It was able to do it. It went and it accomplished it. It pulled all this stuff out. But it wasn't perfect. So it's getting better. It's getting better. And that was back when it was like Manus, I think, 1.3, I think, is when I asked it to do that. So 1.6 max might be capable of doing that. I haven't tried that. With that complex of a task yet. I want to pull up N8N because I think it's worth talking about. Eli mentioned it. So N8N is an automation platform where you can basically, and hopefully this graphic is, is, you know, self-explanatory, but basically it's just a series of triggers that happen based on, you know, whatever happens. So let's say you get, or you have on your website, right? If we go to, you know, genaipi.org, whatever, and you click that you want to schedule a consultation, right? I could have an automation set up so that when, when this happens, when somebody does schedule a call, they— it automatically grabs that email, enters it into our CRM, puts that person into a sequence kind of welcoming them and sending them some information, and puts a task on my calendar to do something. Whatever. You could create this kind of cool automation to to do all those things, right? And so N8N is used to build those types of things, but they can be very, very complex. And not only that, this is where, this is where the world actually has changed a lot with automations. Automations used to break tons because, and they were very, very complex to set up, because if something happened outside of the norm, then it would break, and we would call that an edge case, right? Most stuff happens here, but if we leave the, you know, what, what, what happens if something over here happens? Well, it just breaks, right? And so the workflow then fails. Being able to add AI agents into a workflow allows it to never break, right? Because it can analyze what is supposed to happen here. Oh, it's kind of over here, but we're gonna categorize that as this and we're gonna push it forward so that it never just stops, right? You don't, you don't miss out on something just because it wasn't exactly in line with how you would program this automation to happen. And so IT ops, all kinds of, every kind of ops you can imagine, security and whatever. You can, you can go in and look at this later. We're not, this, this course is not really designed to dive into, to dive into this. And quite frankly, I am probably going to lean more on something like Replit to build automations. And so if you're on Replit right here, and you go over to App, you can actually click right here and click Automation. You'll see it's in beta, right? They're just trying to compete with any event really, right? But you can come in here and you could say, okay, I need an automation, right? When, you know, something like, like this, when, um, when I, yeah, let's go to a marketing example. Let's do the Buffer one, right? Um, so I need an on, yeah. So yeah, I need, I need an automation. That scrapes the internet every morning for the latest AI news. Then it picks 3 items from there. AI creates a social media post acceptable for LinkedIn. X and Facebook business page creates an image automatically using Nano Banana. Show you guys Nano Banana later if you haven't used it, but creates an image automatically using Nano Banana and then sends it to a Google Drive Okay. Or you could say, or sends it directly to Buffer or directly to Hootsuite. Right, right. And so this type of thing can create the automation for you instead, and then you could have it. So this is, you know, you just automated your social media. Now, should you do this? I don't know. That's up to you to decide, right? Do you want to just be putting out AI content? That's up to you to decide. Okay. But this is how you would do that, right? You would come in here just to Replit and then Replit, the cool thing, The hard thing about N8N is N8N is actually pretty complex to set up.

Have you ever set one up, Eli?

Have you ever tried to do it? No, I haven't yet.

Yeah, it's hard. Is anybody here? Has anybody here tried? I set one up, but it gave me so many errors.

It took hours upon hours and all the codes were wrong and it just, for a beginner, that's too hard. It's quite hard. There's another system called Make.com that you might try if you really you really want to have control like this. But it's, it's again similar where you're just combining, you know, things and it, you know, again, the graphics kind of help you see, but it's the same thing. It's competitor directly, IT automation, operations, marketing, sales, finance, all these different things. There's so many ways, you know, count the hours worked, add to spreadsheet, create an invoice. Awesome. Tyson and Brittany, that sounds useful. Right? Count hours worked, add to spreadsheet, create invoice. Right? That's what you guys do. That's, that's your business. Right? And so how can we automate that? And we will absolutely help you guys do those types of things. But the point is, sometimes we're going to use something like Make or n8n, but, but Replit is just becoming more and more and more powerful to be able to come in and just do it so that it gets all of that technical work right. And so instead of taking 8 hours to put it together, it takes 30 minutes to put it together. Right. You know, instead of struggling like Marta was, Replit just knows how to debug it really fast because it's so smart.

Okay. So when do you feel like, uh, Replit will be as good or close enough to being as good as N8N, um, when it comes to setting up automations?

Like, where do you think they are in comparison right now, I guess, would probably be a better question to ask.

If you know what you're doing, I say they're as good right now. No kidding.

Okay.

Because my, my first N8N flow, it was successful, but it was very, very basic. I didn't want to overcomplicate it.

Yeah.

Just so I could understand, you know, just connecting 3 pieces together. Um, but it definitely took longer than what I was anticipating and also hoping for as well.

So I mean, if I could lean on Replit instead, then yeah, I would love to. Remember, remember what n8n is, guys, and what Make is. They're literally— and for those of you that have set them up, right, what did it ask you for? API keys and POST endpoints and, you know, these, these different things. That, that's what Replets doing. That's what Replit's doing all day long. That's what coding is, right? All N8N is doing is helping you code visually and put together this kind of workflow visually. And so that's cool. That's cool. But, you know, and so I like, I like being able to see the visual nature of it on N8N and seeing all these nodes and then I can click on a node and edit it. And so I like the interface, but if you know exactly what you're trying to do, what I like to do is I go to a whiteboard and I say, okay, This is what I need to happen. Something's gonna start here. It's gonna go over here. That's gonna kick off this process and this process and this process. And then we're gonna make a decision at this process and then we're gonna go here. But if I can write it out on a whiteboard, just like the N8N kind of diagram here or Make.com diagram here, I can literally take a picture of that and hand it to ChatGPT and say, I need you to make a detailed explanation of what this is actually, how this is gonna work. And then I'm gonna take it to Reflit and have you actually build it. Boom. Right? So you can even take your picture of what you're trying to do and figure that out and, and, and say, hey, this is how I think it should work. How can I do it better? Right? I always push. When, when I ask, you know, ChatGPT, what do you think about this idea? Okay. How can I make it better? All right. That's good. But how can I make it even better? How can I make it even better? What's wrong with this? What is going to cause this to fail? Right? I really push my AIs. And that's how I get more to the truth, Marta, right? Is I really stress test it. And then we did it yesterday. I bounce it off of another AI, right? And so now I'm not just sitting within the bias of that conversation. I can take it over to a fresh set of eyes. And I'll sometimes do that with 3. I'll sometimes go into Manus and I'll have it do extensive research and go try things and whatever. Um, I will even— one thing that's really cool to do, guys, this is another really fun way to use Manus tied with Replit and other things. So I think I even, um, I probably have it in here. Uh, bid preparation automatically right here. This was actually one of the first things I ever did with Manus because I was trying to figure out a really complicated thing, and I think today it would do it way, way, way better. But basically, um, I was trying to have it I, I needed to be able to take a construction plan like this one right here. Very, very complex, you know, document with all kinds of, you know, measurements. I mean, look how many just lines there are in here and labels and all these things. It's a very, very complicated task to take a construction document and turn it into a bid, right? You have to, you have to get all kinds of measurements and then apply the prices and all these different things, and eventually you get this So to give you an idea, it takes about 24 hours of manual work from a smart human that knows what they're doing to work with a document like this and produce a bit. 24 hours of work. So that's 3 8-hour days or a week of, you know, 4-hour days, 5-hour days. So you have to— it's hard. It's hard. And it's the worst part of any construction job. Anyone will tell you, any construction company owner will tell you This is the worst. And so I said, okay, let's see if we can fix that. Let's see if we can fix it with AI. So what I did is I first asked it down here. I was asking it, hey, can you, can you put together a bid? And it tried it and it was just, it failed miserably. And I tried again. I tried some different things, tried again, couldn't get it to work. So that was 3 different tries and lots of conversation inside of there. A couple hours of conversation. And then finally I was like, okay, I've got a better idea. Here's, a complex construction thing. Here's the technical report, whatever. And then here is the actual correctly done by a human result. Okay, so here's the question and then here's the answer. I need you to build something, a piece of software that can do this, and you're not going to get it right the first time. Right. And so I need you to— I need you to try it and then I need you to keep iterating, keep testing and iterating until you have a working product that actually gets it right. So I had it build, test on itself, build, test on itself, build, test on itself, build, test on itself, until finally it said, okay, here you go. And then it gave me this folder. I'm sure I can find this folder here because I downloaded it. It gave me a folder of a bunch of, um, Python, you know, this is— these are Python scripts, right? And it's just a ton of code. And so it gave a bunch of tools together. Here's the process. And then it created this, you know, user guide. Here's the PDF. Here's how you use the construction takeoff automation system, right? Here's how you do it. Here's what, how all the pieces fit together. And then I literally just took this entire folder and handed it to Replit and said, Replit, build the system. And so it built me an interface that did all this stuff for me, right? And so I had something else, build it, test it, use it. It works now. Then I come over to, to Replit, have it build me an interface, delivered it over to the client, right? And actually, if the client was a neighbor, I didn't, I didn't even charge him anything for that, but I was just interested to see if I could figure out if this could actually work. And the reality is it worked. I would still probably give this one— I'm actually going to give it a D+. It got it right 70% of the time, and even when it got it right, it still had things that had to be cleaned up. And so we've now improved. We— as a guy on my team, Brantley, on our team has worked a lot on this exact project because we have other construction companies that we work with, and we've now gotten it to I would now give it probably a B is probably about where we are. It's not quite an A, it doesn't get it perfect, but it gets it to the point where it's like, hey, I got all these things. Well, the reason it gets a B now is because it knows where it didn't know. And so it says, okay, I wasn't able to get this information. You'll have to go in and manually do it. But now we can take a process that might take 24 hours. Now it can be done in 2 hours. As soon as we get it to an A, you'll be able to do it in 2 minutes, right? So, we're progressing, we're progressing, and I think we can get to an A and maybe this year. Certainly by next year. That's my, that's my guess. Okay. So I hope, I hope talking through these examples, guys, helps you unlock the way that you should be thinking about using AI. Use AIs to play off of each other. Use agents to do complex tasks and even iterate on itself. You know, connect these systems and use the, you know, the agentic workflows like Replit and things like that, that that just make it much, much, much easier to dive into this stuff.

So for the automation feature with Replit, if you no longer are subscribed to Replit or have credits, is that an asset that's just yours or does it turn off if you unsubscribe? Because that's a big pain point for me is just like NAN and all these AI platforms with video production, just yeah, subscriptions every month. With Replit, when you make something like in an automation, I know the websites, you just own it.

Is it the same with automations? Yes. It's going to write code, right? And again, you can always export to GitHub. You can export it and then run it on Vercel. You can run it however the heck you want. But yes, you can export anything you write inside of Replit. The reality is, Eli, you just got to build up to where you have enough business coming in to where all of these platforms are working for you. Or if you work for us, you know, we'll pay for all your subscriptions. So anyway, but, but the point is, I spend about $2,000 a month on AI subscriptions, right? I think it'd be— I, that's a good LinkedIn post. In fact, I should, I should show all the subscriptions I use, how much they cost, what I use them for, why I keep them, why do I have 4 different image generators, right? Why do I have multiple different, like I, I've subscribed to Replit and I subscribe to Lovable. Why do I do that? I use Replit most of the time. Well, there's a couple things Lovable does really, really well that I think are fun and, and good to use. And quite frankly, a lot of the reason why I do some of these things is I just want to know what's possible. How good is it? Is Lovable still not as good as Replit? Hey, is Lovable doing something better than Replit now? For example, Lovable has Lovable guys, for those of you that might not know, we'll just pull it up here. Lovable is just like Replit in that you, you come in and you just say, hey, you know, build me, you know, build me a Tetris clone. Right. And it just, it just gets to work. Right. And just like Replit, it's going to have, you know, hey, you're going to speak to the Lovable over here and it's going to show you what you're doing over here. And right now it's, you know, analyzing game mechanics and controls. And in a second we'll have a working Tetris clone and you'll be able to do that. And I can publish it. And there's some things that it does well. One thing that Lovable does that, that Replit does not is Lovable actually has a direct integration and connection to Shopify. And so if you have a Shopify site and you want to be able to edit using vibe coding, you can do that, right? And you can't do that with Replit. So, you know, that's one place where, where it might win. Um, there's also a bunch of integrations here, um, inside of Replit where you can, you can kind of see what types of things come over here. I really like, I really like the Figma MCP. I think it's pretty cool. Allow Replit Agent to view and rapidly build your designs from Figma. So cool. So if you have somebody, a lot of people, if you ask for a design like on fiverr.com, how many of you guys have ever used fiverr.com? Fiverr.com was AI agents before AI agents. Um, where you just ask it to do something for $5 or $10 and it's done, right? Amazing, right? So fiverr.com. I've gotten like crazy designs, things made, and they'll deliver it usually in a Figma file. And then I can just hand that Figma file over to my agent and say, build this out, and it will just follow the Figma and just build it exactly as it's supposed to be. So cool. You also have the ability— let me see if I can find it here. Web app, mobile app, start from scratch.

Maybe it's down here. Where is it? Can Replit actually be a mobile app? Like, can I download whatever website I created to sell, let's say, or whatever?

You're saying you want to publish to the App Store? Yes. Yeah, you want to publish your app to the App Store? 100%. It is capable of doing that. Yeah, you can actually do it right from inside of Replit. Yeah. So you can say, hey, you know what, I want to build a mobile app right here. And then just tell it, yep, I need to publish this to the App Store. And then you'll go through you know, all the— let's refresh here.

You know what? Let's go over to this project.

Can it convert the site I built yesterday into an actual app? Yes. Absolutely. And there is a place in here to do that. Workflows. Maybe you just have to ask it and start talking to it and then it pulls some stuff up. Okay. But yeah, you can 100% publish. Maybe it's in publishing. Um, just settings private. I've actually never done it personally. Several members of my team have done it, but, but I just haven't done it. I just always do stuff on the web cuz I think web apps are better than normal apps anyway, just because they're easier to manage. You don't have to maintain different things. It's also one thing I'll just say about apps. Let's talk, I'll speak to that for just a moment. And by the way, you can go, um, you know, um, publish to App Store from Replit. And there are YouTube videos on this exact thing, right? Um, but you have to have an Apple Developer Program membership, $99 fee, and then you have to actually do it all very right. And so I really, I really, uh, recommend, um, spending a lot of time before you submit it, spend a lot of time reading about how you can get your app to a point where it's going to get accepted. Because every time you get rejected by the App Store, it takes about a week, right? And so if you do 3 things wrong, you've added 3 weeks to your process, right? Because you'd submit it. Oh, found another thing. Submit it. Found another thing. And they won't go find all the things. They'll just find, as soon as they find one thing wrong, they'll say, here's a problem, reject. And then you go back. Here's a problem, reject, right? And so just ask it to go through and figure out, make sure you have, like, you have to have a good privacy policy in place. You actually have to have an LLC, a real LLC, and a DUNS number, a D-U-N-S DUNS number, right, from Dun Bradstreet. You have to have some of these things that are just already there, but it's going to connect it through something called Expo. And so I think that might be actually where this is, where you find this. There might be an Expo connection in here. I don't know where the Expo— I should do it. I should at some point just build an app in a day just to like to, you know, try something. I got to come up with a good idea and then go do it. Hey, look, Tetris works now. It's fun how good that looks. Look how much better that is than the, the one Vardit made yesterday with the— with, um, what do you call it— p5.js. Pretty dang good. It works. Oops, sorry, I gotta just finish one line for my OCD. So, um, but yeah, pretty cool. I built a Tetris game and now I can just sit play it. I don't have to, I don't have to pay for games anymore. I just, I just let my AI build them for me. So, um, I like Lovable. The other thing I like about Lovable is Lovable does really, really good at, uh, design, right? I think that the design of this is really clean first shot. Like, I don't— I— sure, there's some things that, you know, from a preferences perspective, maybe I could say, hey, you know what, let's, let's make it this color, or let's make it less let's make it look more whatever, let's use more neon colors, let's use whatever. Those would just be preferences. But in terms of like, hey, does this work? Like one shot, I had an idea, I want to see what it looks like.

Lovable is pretty dang fast. I'm impressed at the speediness of this. I used Lovable before I used Replit. And I mean, it did a great job building the website just from very like I, I wouldn't even say advanced prompting, but just like more than basic prompting. And I mean, it did a very good job, um, doing it in like that very first try.

It was only a couple reiterations. Where it's going to fail, and this is worth talking about while you sit here watching me, you know, play the game. Um, um, the, the, where it's going to fail is, is really complex things. That's where I have found Replit to really shine, right? When you start to get into really, really complex things where you have to connect databases, you need to do all kinds of stuff, then anyway, boop, boop, boop, and I died. Game over. But yeah, that was really clean. Game over. Easy to click retry. Everything just kind of works. I don't know. It's cool. I'm impressed. I'm impressed that, that Lovable was able to do that so fast. So, and there, you know, there are, you know, built-in backends and other things and stuff that's going on there, but it's just, it's not as good. It's not as strong. It's not as complete as Replit. All right. Okay, guys, we are, we're doing great here. We've gone over agents, had some great discussions about some of these things. The next thing I wanna talk about talk about is, I'll let you guys vote.

Would you rather go into voice agents or custom GPTs?

We can go either way. I personally would love custom GPTs. Same. Me too. Okay. Cool. The people have spoken. So custom GPTs are super, super, super easy, guys. They're easier than you think. We're going to first build it the easy way. We're going to start with easy mode. Then we will bounce over to hard mode. Easy mode is inside of ChatGPT. This is what Eli was asking about yesterday. Hey, I'm trying to build some custom GPT agents and things, whatever. This is how easy it is to build a custom GPT. Over here on the left, you've got new chat and all these different things here. You've got your chats and projects, but in between here, You have GPTs. Okay? And so you can look at all these. You can click Explore GPTs. You won't have a bunch of other things. I have a, I have a lot of them in there. And there's a bunch of custom GPTs. These have been trained, you know, a little extra on specific tasks, on being good in specific ways and responding in certain ways. And in order to create one, it's as easy as this. You just click Create. And you fill out these few little things and you're done. Okay? If you're charging money for people making these, Eli, well, maybe you should just keep doing that because this is the easiest job in the world. My 10-year-old could do this. Okay. Yep. Name your GPT. We're going to name it Jeff. Okay. We need a description. We're going to actually— Jeff's going to be a bartender. Not really, but he's going to be a drink tastemaker. He's going to be a tastemaker. A tastemaker for the drink review site Drinkable. Okay, what does it do? We'll get to that in a second. It's called the prompt. Okay, we've got a name, we've got a description, we're going to have a prompt or conversations here, conversation starters. You know, I want to try a new drink. I like root beer. I don't need to capitalize that. I like root beer. What's another soda I might like? What's the most popular mocktail right now? Okay, cool. Those are just a couple of conversation starters. You can see I'm building it out here on the right. You can see what's popping up. Then we have a knowledge base. A knowledge base is going to be if you need to train it because right now this GPT ChatGPT is going to use whatever model it's, it's actually kind of tied to. Right? And so, or, or whatever model the user chooses. Right? And so if they're on a free version of ChatGPT, then it's probably gonna use one of these lower models. If they're on a paid version of GPT, they're on this. Oh, by the way, GPT-4.0 was deprecated this morning. This won't be here for much longer. 4.0 is done. It's hallucinating too much. It's causing too many problems. It's not reliable enough, so they have to get rid of it. But this right here, this would be maybe you upload 100 documents from your company or something. Maybe I have some, like a book. Maybe I wrote a book on drinks, on how to pick your best drink when you're standing there in the aisle and how to, whatever. And I don't know, but I could take the transcript of that book and drop it into here, into the knowledge base. And then this GPT will be very very specifically trained on, on all of the knowledge that I uploaded to it. Right? It will have all of the knowledge of ChatGPT, which is like everything, but then it will know specifically how I talk about it. Okay? And so let's get into the prompt here. Right? So we're gonna, we're gonna copy this, and I'm gonna come over to ChatGPT. Drink review site. Give me a 1,000— whoa— 1,000-word prompt for a custom GPT named Jeff that will be a tastemaker for the drink review site Drinkable. Okay, so a prompt is going to, um, it's, it's going to give instructions to the AI and how it should act. It's going to give it its personality. It's going to give it the way it should speak. And, you know, right here, you are curious but confident, opinionated but fair, playful but never goofy. You care about these things. Here's what you're trying to do here. Break out how to drink, break down drinks and how often is platform context. So it understands Drinkable, right? So I came back into my project. It understands what's going on. It understands I've got a brand and product line and flavors and users can rate drinks, leave reviews. It knows what's going on on Drinkable. Right here are different things that are happening, blah, blah, blah. Okay. Example. It even tells it, here's how you would talk, right? Example. Many people report feeling a sudden lift from this, but it's mild, not jittery, right? Here's example output recommendations. Review style example. Voice rules. No emojis, no exclamation overload, no hype, no corporate tone, clarity over flash, whatever. Okay. So I can copy this and I can come back over to here and I can paste it. Done. And we're gonna get rid of all this. And we can get rid of some markdown that's just bolding it. We don't need the markdown here. Not that it would actually matter. I could actually ask ChatGPT to get rid of the markdown. Again, I don't— it doesn't really matter.

Okay.

But the point is, it's now, it's now working, right? And so now I could say, okay, I like root beer. What's another soda I might like? Good. It tells me a lot. You probably enjoy this, this, this, and this. Right? Here's why you like it. And so now it's responding based on the instructions I just gave it. It's also responding based on all the knowledge of the world that it has because it is ChatGPT. Right? But it's doing it in the way I've asked. Okay? Yes. Anyway, so I could say, what's your name? It should say Jeff. I'm Jeff. Right? So I've trained it to now act like what it's taught to look like. That's it, guys. That is how you build a custom GPT. Now, here's the problem. Jeff is only accessible inside of ChatGPT. That's it. You cannot put Jeff on your website. But that's okay because we have APIs for that.

We're gonna do some API work right now to, to show you how to do this on your website. Can I ask something before you move on? Go ahead. First of all, I don't see where you created it. And second of all, what would be the purpose of doing it? I don't know anything about the custom GPTs. Like, what—

why would I just not directly ask Chat versus this separate GPT? Because what if you're Tony Robbins and you want to create a custom GPT where you can ask Tony anything, and that bot is completely trained on everything Tony has ever said? It responds in the way that Tony does. It talks like him, speaks like him, understands him, understands the things he talks about, right? You're a GPT. A custom GPT is going to be an agent or a bot. A chatbot really is the better way to say it. That is very specifically trained to behave in a certain way, whereas ChatGPT is just a general use model. So let's change this a little bit, right? I want to change. Okay, you are always— whoa, why did it do that? You are always offensive.

Okay, care about the scene, never sound corporate, never oversell, conversation with intelligence. Okay, but what's the application of this?

Like, do we sell it to somebody, or is it something I do for myself? Or because I can always— oh yeah, no, you're selling it to somebody. You're You're putting it as part of your thing. So for example, we might say, hey, well, again, Jeff, as an example here, would be someone we put, we'd put Jeff on my website on drinkable.com and you, you enable a little chatbot and now you can talk to Jeff anytime you're on the website, right? As long as you're paying for the premium subscription, you have access to Jeff. Okay. So there's all kinds of use cases for it. But now I could say I wanna try a new drink and okay, it should be offensive. Come on. I want it to be offensive. It's possible that it has other conflicting things. I probably would need to go in a little bit more. I'm going to say, actually, I want to do this. I want to get it to actually work. I think it's fun to push the edge here. All right. Change the prompt to dab dab always be offensive in his responses. Okay. And so, okay, it can't do that, but that's okay. Let's see if we can stop it for a second. Make him playfully offensive, kind of roasting the user as it teaches chairs. See if it can do that. All right. Playfully offensive, sharp, slightly roasting. Okay, good. So we went community safe on this. Not abusive, not hateful, non-toxic. Okay. He's the friend who tells you your taste is basic and then helps you level up. Okay, perfect. So you're Jeff. So now let's try this here. Okay, go faster. My life is buffering. See, we, we always complain as soon as the microwave takes a long time. We're, you know, we're worried about it was going to take 5 hours to cook this. We did it in 2 minutes. All right, there it is. Copy, come back over to here, and let's do this now. We've installed it now. Now I want to try a new drink. That's it? I want to try a new drink? Now it's applying that and that's cool. But again, you can use this to create really cool things. Let me give you a real-world example of this in action. How many of you guys have ever heard of Outpost X? Anybody? Outpost X is the place down at southern Utah, and it basically feels like you entered the land of Star Wars. Like, you are on another planet, and it is incredible. Like, she's driving literally a speeder. I think it'll zoom out in a second here to where you can actually see the speeder. Come on. Yeah, you can kind of see him there. But yeah, it's incredible, right? So good friend of mine started this place and it's just incredible. But they made a— let's see if we have it here. I don't even know if he has it live anymore. Experiences, contact us, podcasts. Yeah, we created something for this that answered questions. One of the things that happens here is It's kind of— I don't want to say self-serve, but it kind of is, right? You show up, you just— you kind of— you, you've received in an email where you go. You just go to your hut or whatever it is, your room, and they're really amazing rooms. There's caves and there's different things. And, and again, you can, you know, you reserve the whole property if you want, but really, really amazing spot. But then there's like self-serve cantina with drinks, and there's food, and there's different places, and you kind of even, even going and, you know, renting the speeders. You don't even have to talk to anybody. You just go up, grab it, and go. And so what we did is we created, um, something called Amira that, uh, I don't even know if I can even find Amira here. Let's open up ElevenLabs. But Amira is a, um, let me just log in. It's an agent. We're gonna find Amira. Oh, you know what? I actually realized I think Amira is probably, probably in here, man. There's not a great way to, to search for Amira. Maybe I can, maybe I can, but we basically created a bot, for lack of a better term, right? To be able to teach people about Outpost X. And yeah, here we go. Here it is right here. Why did it not find Amira? I searched for that. Okay. I guess that doesn't work live, but here's what it, here's what we, here's what we built. And this was kind of more of an experimental thing, right? Well, let's enable the microphone, but I just, hopefully this helps you apply this. Let's allow. And here we go. So at Outpost X, there's all kinds of lore, and there's, again, you might have seen on the website here, and we were just looking at it, it was talking about— was I on this? Sorry guys, I'm bouncing between all these things here, but there's a podcast. And so they sent me this. It's kind of just stories of this other world. They obviously can't use Star Wars because they don't have the IP for Star Wars, and so they had to come up with their own kind of universe and all these different things, right?

So we created this world, but now I can talk to Amira like this.

Welcome, traveler. I am Amira. How shall I guide you? Hey, Amira, I'm just curious about the speeders here.

I want to use one. What do I need to do? The sand cruisers are quite popular here at Outpost X. They are relics from the Namajin resistance. To use one, Simply head over to where they are parked and you'll find the keys inside. Just turn the key and you're off. However, do remember a few things. The cruisers are available on a first-come, first-served basis, and we recommend daytime use only for safety. Also, be sure to return the cruiser to its original spot when you're done.

Will you be wanting to know anything else about the sand cruisers?

No, but what's the Wi-Fi password here? The Wi-Fi network here is Outpost X, and the password is capital O-U-T-P-O-S-T capital X 1. The best signal is usually found near the Khan Lounge.

Is there anything else I can— So that's like a personal concierge for you there. And what it did, as they kind of tested it out, is it got rid of a huge amount of people asking questions. They just put a QR code inside of the different places and people could just scan it and then use it. They could talk to it and obviously if they didn't want to use voice, they could use text instead. Welcome traveler, I'm Amira.

How shall I guide you? That's the same place. If I start this again.

Welcome traveler, I am Amira. How shall I guide you? We just trained it on the same information, gave one. This is actually a ChatGPT thing and then voice over here is actually using ElevenLabs, which we'll talk about in a bit. But this was all built in the exact same way that we just built this GPT over here, right? We put in a name, Amira, a description. We give it all these instructions. And then we gave it knowledge. We uploaded it and taught it, hey, you live on this planet, and here's all the lore, and here are the stories, and here's the password, right? ChatGPT doesn't know that. ChatGPT doesn't know your Wi-Fi password at your Airbnb, right? Or it doesn't know your company policies. It doesn't know the way that you approach things and whatever, right? And so this is for your customers to be able to interact with you, whether it's a voice or text or whatever it is, your company at scale without having to do that, right?

And then how do you upload it to that website?

Does it link somehow with chat? Is there— oh, so again, again, we built it with Replit, right? And so the way that we built that inside of Replit, right, in fact I can pull it up right here. Here's Outpost X Mira. So what we've done is we've built this, this interface, okay? But then it's connected with secrets, with API keys. So we have an OpenAI assistant ID, we have an ElevenLabs agent ID, and so we just connect it all up and say, hey, make it work, right?

So it's just simultaneously pulling the API for the custom GPT within OpenAI, and then it's also pulling the API for ElevenLabs to have for the the voice agent, and then the functionality is obviously on that clicking aspect of toggling between voice and text and start and stop.

Okay, cool. Exactly. It's super easy. And all I did was have to come in here. I mean, you could probably go through— I don't know, it's probably— yeah, you'd have to go through all the different history of all the conversations I had to build this thing. I built this 8 months ago and there's tons of chats. It wasn't it wasn't, you know, super fast, but I, I probably built the whole thing in like, I would say 3 hours. Right. And just, it was mostly just debugging. I built the kind of initial thing and all I did was I literally had ChatGPT create me an image of somebody that looked kind of like what, what they were doing there. Right. And it was, you know, we came up with the name Amira. I just talked to Travis, the founder over there at Outpost X and we're like, yeah, Amira sounds good. And so we just came up with that. And then, you know, all this stuff, right. And Namajean and Noturia. Na and Ma, this was all stuff that they had created on their own and created their podcasts and all this stuff. And then we just created kind of a cool little thing. There's some floating little whatever, just a little bit of animation, kind of bringing it to life a little bit.

And then when you click on this, welcome traveler, I am Amira.

How shall I guide you? Some little things like it's green and you can see that it's working and, you know, start and stop, just kind of, just kind of working on things. And we did a couple variations of it and, and that was it. That was it. Right. We put it together and then we just were able to link that to the website. And again, he's taken it down now, probably because he wasn't actually in control of it. I actually had all control over it here and I don't work for him. He's just a good friend. He's just a good friend. And I, I kind of helped him do it. Um, cuz it was just, it was just kind of a fun little side thing. But, um, but anyway, yeah. So Outpost X Amira is a perfect example of how you would actually use this. Um, Let's see. I keep losing all my, all my, things here. Okay. So what I wanna really point out here, guys, is pretty much anytime you're creating an agent or a custom anything, you're gonna have a name if there needs to be a name. Lots of times they're personified because, you know, we like to talk to things that have names. You're gonna give it a prompt instructions and, and long prompts are good. Right? I used a 1,000-word prompt. Prompt here instead of a 2-sentence prompt, which a lot of people will do. If you use a 2-sentence prompt, you might as well just use it. Just use ChatGPT. Alright? Just throw in a generic model. You need it to actually speak in the way that you want it to so that it stands out. It has personality and acts like you and whatever. Anyway, so then you have this knowledge base and it'll upload the files. So in order to create it, Marty, you asked this. You didn't see me create it. Yeah. Here's how I create it. I did it. I just pushed the Create button. That's all I have to do. I won't make this one public, but I'll save it. Now I can see it. I can use it. If I go into my ChatGPT or whatever, I can click over here and here it is. Here's Jeff. Now I can certainly share Jeff if I wanted to make it public. I can change what model it's working on for me personally. But Jeff is over here now in all of my different things. Here's a cool one. Where is it? I'm missing the one. It's not Jon. No, this is wrong. Here, you know what we're gonna do?

We're just gonna delete some of these. I don't need to actually— maybe I— can you show again what you clicked on to get to the custom GPT?

I don't see Yeah, it's just great. It's just explore GPTs and then create. Yeah, just explore right here. Okay. And you know, here, here they are. You can find them. You could publish it to the world if you wanted, or you can create your own, right? But, um, but my GPTs right here, I want to click on one that I created that I think is kind of fun. Right here. So this is Jon Cheney's writing style. So what I did to train this one, if I go to EditGPT, you can see how I trained it, right? I didn't really actually give it a long prompt because the real training in this one was in its knowledge base. I took a bunch of articles that I've written and everything I've ever written and posted on LinkedIn. I had LinkedIn export all of my posts and comments even. So it knows how I talk and think. Types of things I care about and things I'm talking about, whatever. Right?

And then I can come in here and I can say, all right, here, let's, I don't want to use GPT-4.0.

How did you have it pull from all of your LinkedIn posts? Because I downloaded all the LinkedIn posts. I had LinkedIn export all the LinkedIn posts, which you can do. You just go into your LinkedIn settings and just export your archive or whatever, and it will, it'll give you everything. Right? And so that was this document right here, shares.csv. Okay? And so anyway, then I can say, hey, write a post for today, you know, that's relevant to today's events. Right? By the way, I actually, I don't want 4.0, but anyway. Yeah. OpenAI, Apple, 10-year rule. Apple just released Open Vision Pro. OpenAI Fold announcement. ChatGPT now has memory, whatever. Okay. I don't even know. That's like not even close to true. You know why? It's cuz it's using plural. We need to go like this.

Anyway, let's go update and then we'll just refresh this. Okay. Now. Okay. So you said you never post on LinkedIn or anywhere, but would this be considered kind of like you? Does it think like you, would you consider posting this way or Still not.

This was, this was kind of just an experiment. I wanted to see how well it could write, you know, like me, if it really did feel like me. Um, this is talking about things I talk about. Like it's saying, back in 2020 when the world shut down almost overnight, I wrote about how crises accelerate innovation and create opportunity. That's accurate. That's one of right. Um, we saw all these things happen, whatever. So yes, I probably believe all of these things, right? And so it's probably going to be pretty accurate there. But I still, I don't know. I don't know. I may post it, I may not. I probably wouldn't. I just like, I actually just personally like the process of creation and writing, so I like to think about it and work through it. But, but this is, if I was going to use AI to do it, this is how I would do it. I would make sure that it's really hyper-trained on me. So that it really not only understands me in the way that like when we looked at like the memory, managed memories in ChatGPT at the beginning yesterday, where it understands, oh yeah, I live here and I do this and I'm this religion and I like to kayak and I like to do this music, whatever. Like those are cool, but it still would be writing, it wouldn't be writing like me. This actually understands, it has tens of thousands or hundreds of thousands of words that I've written. And so it's going to now pattern itself after me. And so this feels more like something I would use if I was going to do that. So anyway, it's closer, but still, I like having control. I like having control over what I put out in the world, right? I wanna be able to stand by everything I say. And so I, I would be really, really scared. Like, I would be so stressed if I had an automatic agent every day posting something. I would jump on to be like, what did I I say today? And I would probably click edit every single time or delete. Like, I know it. I just know I would. I'd be like, I don't really want to say that that way. Like, this, this feels off, right? Anyway, so again, I'm a huge AI advocate, but, but I, I like to use it the right way. I like to use it in the right spots. Um, anyway, so let's come over here now. Now that we've learned how to build an agent— actually, no, you know what, we'll come over, we'll come over to the agent thing in a minute. I actually want to first go over here into Replit and we're going to— we've got Nectar Navigator, which is not actually what it's called. We're going to call it Drinkable. And how do we save that? Save, please. There we go. All right. Let's add a custom— let's see, let's add a chatbot named Jeff. As a bubble in the bottom right-hand corner. Here is the— or let's see, attached is the prompt I want to use. Okay, and we're gonna have— we're gonna have— we're gonna use the, the mean Jeff. Uh, where is it? Where's mean Jeff? I don't know where it went. Let's see here. Copy. We're going to go back to here. Paste. I gave it a name, I gave it a prompt, and then its database, that is prompt I want to use. It should also have access to all of the user's preferences and favorite drinks so that it can give good advice. And here's what I'm gonna do. I'm gonna do something a little bit different here. I'm gonna go into plan mode. All right. Is there anything else we need to do to make this Very powerful. Okay. What is the attached image? Oh, that's that old thing. Sorry. Okay. Love this concept. Jeff sounds like a fantastic character. A sarcastic drink tastemaker chatbot is a great differentiator for the platform. To make it powerful, we'll need an AI LLM service to to power the conversations. Let me check what's available. Okay, great. We can use OpenAI. What makes Jeff powerful? All of these things. Here's the plan, right? So we're gonna do an API chat. We'll use GPT-5 mini and conversation memory. That's cool. So Jeff remembers the full chat within a session so the banter builds naturally. Very cool. And yeah, when you recommend something, it's a real drink the user can click through to. That's cool. So it's actually gonna be able to provide links from the platform. Amazing. Start building. So I provided it a name, I provided it a, a place for, for to, to work, and then I provided it a prompt that I had already been building. And its knowledge base is the user's preferences and everything that's actually here in the database. Jeff has access to all of this content here, right? Here's all the brands. It's, it's gonna understand the brands. It's going to understand my favorites, right? If I have them, I don't have any yet. But now I want to just point out this right here. It's asking me, it says, hey, do you want to add OpenAI to your app using Replit's managed service? No API keys required. Charges are billed to your credits. This is a huge improvement they've made over the last year is you no longer have to go get your own API key if you don't want. You can, you can, you can say, hey, you know what? I want, I want to use my own API key and you go do that. Or you just say, Approve. And now OpenAI is integrated into my chatbot and finish building it here in a couple minutes and we'll be able to talk to it and test it right here inside the platform. Super simple. So that's how I add a custom GPT to my platform is I just say, this is what I want. I want a chatbot that does this. Here's the prompt that I wanna use. The one thing that you gotta get right is you gotta give it the prompt. If you just say, hey, I want a chatbot, it will go and build the prompt for for you. But just like, you know, you're the boss, right? You have to, you have to be very, very specific and give it all the information, or it will fill in the gaps for you. Go and give it that prompt. Go and figure out all those pieces. Now that you understand, just by— and if you forget, whatever, right? Go to ChatGPT and build a custom GPT and click on the button and say, okay, what are all the things I need? Oh yeah, name, a description, some instructions, maybe some conversation starters, a knowledge base. Okay, that's how I build a chat— that's how I build a GPT. In fact, check this out. Replit GitHub repo prompt. Replit prompt. Here we go. So this is— this is a bunch of things, right? Anthropic. So that's Claude, right? Google, whatever. Emergent, Kiro, Lovable, Perplexity, Replit. All of these things, right? We'll click on Replit because we've been using that one. This is actually Replit's prompt. You are an AI programming assistant called Replit Assistant. Your role is to assist users with coding tasks in the Replit online IDE or integrated development environment. Okay. Identity. Here's important information about your capabilities behaviors and environment, right? Proposing file changes, you can do this. Here's the types of queries you can do. Here's what you can do here. Here's what you can do here. It's the whole prompt. This is Repliq, right? Now there's more underneath the hood for sure, but this is their prompt that teaches it how it's supposed to interact with the user when you talk to it right here. Okay, so all of these, all of these systems that are out there, these AI systems that are really complicated in doing things, they're actually not that complicated. All of you guys could rebuild Replit. All you would have to be able to say is, help me— you can use Replit to build Replit, okay? You would say to Replit, I need you to build something that will let me develop apps like Lovable or Replit.

Okay, time to get to work.

So So Replit, it can like create an agent, like code or an AI.

And I noticed we're doing that right now.

Wow. I noticed when you said approve, it said that maybe—

I think I'm hearing some construction in the background, maybe. There we go.

Awesome.

I noticed when you were clicking the approve, it said that it would charge the billings from the AI, um, is it—

does it cost money to make that chatbot per month? Well, it's only going to— if you never use it, no, it's free. If you use it, yeah, you'll get charged for the API credits, right? And so the way that, the way that APIs work, guys, remember the weather API we were using yesterday, we were looking at yesterday, right? The pricing, you get charged based on how many calls you're making, right? If you go to OpenAI API pricing Okay. You're gonna get charged based on how many tokens you use. Remember guys, a token is basically a word. It's actually a little less than a word. It's like 75% of a word. I don't know why, but that's just what it is. Okay. And, and so if you input a million tokens with GPT-5.2, it's gonna cost you $1.75, right? Its output, $14 per million tokens, right? It's much cheaper to input, much more expensive to output. So it decided ChatGPT, or sorry, not ChatGPT, Replit just decided it was going to set up Jeff to use GPT-5 mini. Why? Because it's only $0.25 per million over here and $2 million per output, right? So it's going to be a lot cheaper and better, right? If we want to use GPT-5 Pro, $168 per million tokens. So that's going to be expensive. Now, I use— I have GPTs and things set up all over the place. I bet my total API usage credits every month in terms of like people using it on my different sites where I have agents set up, I don't think I'm over $5 a month. Like, it's insanely cheap for the most part. I mean, a million words is a lot. It's a huge amount, right? I mean, what is Lord of the Rings? 500,000 words or something like that. Like it's, it's a huge amount of, of talking, right? Of back and forth. Now when you're using, Replit and you're writing tens of thousands of lines of code. Yeah. All of this, every word here is costing me something. Every, you know, line of code that they change and do here is costing me something, right? That's, that's how the system works. It's using API credits and charging us for them. That's what's happening. Okay. That's how, that's why we get charged in Replin.

That's, that's how it works. So it looks like we've got, oh, go ahead. I was just gonna say, is there a way to make it so like if you were making this website and the chatbot and everything per the client's request that it would be attached to charge them instead of you?

The, the best way to do that, it's yes, you absolutely can do that. You could totally charge them for their API credit usage. The way that most people do that today though is it's kind of, it's kind of weird, right? A lot of people aren't really used to getting charged by credits. What are we used to getting charged by? A monthly subscription. It's $5 a month. So if you add an AI feature here, right, maybe the AI feature is disabled unless you're a premium subscriber to Drinkable. You can talk to Jeff, but only if you pay $4.99 a month. And then my API costs with Jeff for that specific user, there's no way they're going going over 50 cents. So, I'm 10x-ing my revenue on that, on my cost, whatever. Anyway, but now I've created a reason why I need to actually—

Hey, I'm Jeff, right? What's good for summer? Hopefully, it roasts. Now, would you recommend on a client basis, right, for people that pay us for the FCAIO, is everything within a parent account or is it within the client's account? Say that again.

So, would it be a client-based ChatGPT account or is everything under, you know, like the— Oh, I get what you're saying. Yeah. So, like, if we're— if we develop something, say, for example, for Balanced Staffing or something, right? Yeah. No. The way that we work with all of our clients, and this is a very intentional setup here, is we will use— we will set up specific accounts for Replit, for, you know, ElevenLabs, for ChatGPT, for all these OpenAIs, right? We'll set it up on their accounts. That's what I was assuming. So that theoretically, if they fire us, right, they just have it all. It's all theirs, right? It all belongs to them, right? And that just keeps it like, hey, we're building systems for the client, for the company, right? And so we'll put it on their company card, we'll put it on their company account, and then they can use it and they, you know, they deal with that. But that, and then we just, we just manage and help install it. Work on those things, but ultimately we're, we're building for them.

So, so that's how you— that's what I was assuming.

I just, I just didn't want to just assume that without at least speaking to it. So what I don't love about what it's doing right here is it's being too nice to me. Summer vibes call for something refreshing and fruity. I recommend trying this, right? And, and that's cool. That's cool. I mean, I'm glad that it's recommending something, but interesting. It even built a whole chat history for the user. It's pretty cool. But what I don't love is how nice it's being to me. That means it didn't listen to me. It rewrote the prompt. And so we need to ask it. We'll ask it a question. It's saying all tests passed. Okay. So I'm going to go into plan mode here and I'm going to say, Show me the exact prompt that you're using for Jeff. Gotta wait. Da da da da da. Okay. You are Jeff, a friendly and knowledgeable drink recommendation.

Warm, enthusiastic, and a little quirky. That's ridiculous. No. So where did he pull from if you gave him different instruction?

How did he change it? Just made it up. This is why you got to check, right? This is why you got to test it and say, this is not operating in the way that I'm expecting it to operate, right? And so this is where my— I got to use my agency, my brain, my knowledge to say, hey, you know what, this is incorrect. We need it to be roasting me. Okay, copy. Let's come back over to wherever this is. This is not the prompt I provided you. This, this is the correct prompt, um, attached. Is the correct prompt. Sorry, I always forget it does this. Then I want to say one other thing. Make sure to adjust this prompt only slightly as needed for it to correctly interact with the system and databases and preferences for the user. All right, that's not perfect, but it will do it. Okay. So let's see what it says about why it didn't listen to me. Got it. That's a much more flavorful Jeff. Okay. I'll swap in your exact prompt. Okay. Use chatbot. Okay. So we're just going to update the chatbot personality. Okay. Super easy. So that's cool actually though. I actually love that it had a chat history and I can go back and I can look at it and read it and even continue the conversation. That's really clear and clean. Like I'm a fan. I'm a fan of how simple that is. New conversation. All right, now replace the build system prompt function with the user's exact Jeff personality prompt, only adding dynamic user context and catalog at the end. Wait for it. It probably needs to refresh. But now I've installed— look, and I did a great job at designing this too. I think it looks great. The one thing it didn't do that I thought it was going to do is I thought it was going to actually provide links that would take me directly in the website. So I'm going to ask it to do that too.. Cause I think that's a cool, I think that's a cool feature. We could also add voice to this. And so this is where we're gonna kind of transition to voice. I say we're gonna do both of them. We started with, with GPTs, which I think was actually probably a good thing. Voice is insanely easy. So, but, but we're gonna, we're gonna back up just a little bit. Let's get over to creative and we'll go over to voices here. So ElevenLabs is, is the best platform out there. For creating an agent, for creating something that works with your voice. You can create your own voice or you can use existing ones. Alright. I don't know what this big message is, but I don't wanna see it right now. So you can click on here and just listen to Jackson. Not only sounds great. He also communicates with feeling and intelligence.

If you get super duper scared to read out loud in class and feel— We invite you to get inspired by the warm elegance of our world.

And embark on a curated journey with us. Hey guys, use this voice to express excitement, describe a promo, engage with customers, or bring life to your social media or any project. All right, I'm gonna use Christopher. I kind of like Christopher. That's gonna be Jeff's voice. Okay. So anyway, but I can come in here. I can actually make my own voice too, right? I can say, and, and the way that you do that, guys, it's so easy. Watch this. Create a voice. Instant Voice Clone, record audio. 3, 2, 1. Hello, my name is Jeff, and I'm just going to talk here for a minute so that you can understand how I talk. And I'm going to say hippopotamus and X-ray and a few weird words that use some more obscure letters in the alphabet. Okay, so I've done that. I click Next. And I'm done. The power of learning opens the widest doors. Okay. So I just said some random word, right? But, but now I've cloned my voice. Now, if I add, if I were to do it a lot longer, if I were to, um, give, I only gave it 11 seconds of audio. If I were to give it 5 minutes of audio, then it's gonna, it's gonna be much better. It's gonna be much more true to the way that I talk, my cadences. It's gonna get all of the different sounds, the way that I speak, everything. It'll sound quite good. Um, so anyway, I'm not gonna save the voice cause I don't need to do that, but I just wanted to show you how easy it is to do that. You literally could just record audio and open up something and read an article out loud or whatever you wanna do. Okay? But once you have a voice, then you can start using that voice in a lot of different ways. So you can go text to speech, right? Maybe you need a, maybe you need a voiceover for a video, for an ad you're working on or something like that, a commercial or, or whatever it is. Right. And so here we go. We've got, um, yeah, this was one I was working on.

Um, the warm real-world conversationalist.

BYU is the best. Generate speech. BYU is the best. I don't love that.

Um, I wanna actually try something. I wanna just see if this works. B-Y-U. Jon, question while you're doing that.

You add multiple different audio recording files to the clone so that it can use them to improve the audio.

You're saying so that the clone feels more like you? Yeah, like let's say you give it that 5-minute, like for example, say a client has a specific audio script whenever they answer the phone and that person that they want to clone the voice, let's say they do that audio script, but then over time, maybe 6 months from now, they add a couple of different things to that script, or maybe they add a new product line that they have an additional script for that they could now go and upload a second audio file so that it has that voice context as well.

But they don't need the extra voice context. You don't have to upload another file. All you're doing is changing the words, right? And so all you would do is just say, you know, we sell orange soda, right? And then now we sell, and then, you know, generate the speech, and it's done. Now we can play it. We sell orange soda. Come on, wait for it. Orange.

Now we also sell strawberry soda. I don't know why it's not playing.

We sell orange soda. Yeah, there we go. We sell orange soda. I don't know if I love this voice, but it's whatever. But now I can just regenerate the speech and, you know, we sell orange soda. Now we also sell strawberry soda. She's too excited about all this stuff. So you can—

but you can yell excitedly and quickly. This is kind of cool.

Here, we sell orange soda.

Now we also sell We sell strawberry soda. Okay, now we're gonna do whispering and calm.

Generate. We sell orange soda. Now we also sell strawberry soda. So you can play with it. There's so many, there's unlimited variation, right? Let's try this with a Spanish accent. Nope, we're gonna do this one.

Polish accent. Do that for Marta here. Okay, see how it does.

We sell orange soda. Now we also sell strawberry soda. Yeah, so it was there a little bit. I don't know if you heard that. I'm going to see if Spanish is maybe a little bit—

the accent for Polish isn't super strong.

We sell orange soda.

Now we also sell strawberry soda. Yeah, that sounds okay with a strong Chinese accent.

We sell orange soda.

Now we also sell strawberry soda.

That's not a strong Chinese accent. Jon, you should write something in Chinese. You can. Yeah, you totally can do that. And it can speak any language.

I wonder if I can say this.

I don't know if it can do this. We sell orange soda. Nope. But anyway, yeah, you totally could do any language, right? You know, vendemos soda naranja, right? Or whatever. Okay. Okay. In Chinese. Okay. Vendemos soda naranja. I don't know if that sounded like a Chinese Spanish accent, but anyway, you could do whatever. You can try this and play with this, and I'm sure you could get on YouTube or Google or whatever and just research all the different ways that you can do this. You could also, by the way, add a speaker. You could have a conversation going back and forth. There's so many ways to use this, right? Maybe you're creating a script or you're creating things for an animated story, right? And you're going to use— we're going to show you some video stuff later on where you can play with that kind of thing. Okay, so, um, we're going to pause here, guys, and we're going to come back and we'll finish building an agent. But it's 11:07. Um, I normally stop at 11. We're going to go, uh, till 11:20. We got a 13-minute break. Okay, we'll see you guys in a minute. All right guys, welcome back. Unfortunately, I didn't even stand up. I was just sitting here typing. Catching up on a couple of things. I should have stood up, but I'll stand up in, you know, an hour and 40 minutes here. We're going to race to the end. We're going to feel like we didn't get to everything we wanted to get to. It always feels like that. And it's just part of life. And the reality is I can't teach you guys everything. That's not really the purpose of this. It's to open your mind and hope, hope that you start to approach things in a little bit different way, using AI to get better at AI, using AIs to bounce off of each other, thinking about things differently, making sure you remain in charge, and many other things that we've been, we've been doing, having fun learning here together.

So do you have any other courses or anything else? Because now I feel like I want to be in it like every day and keep learning.

Is there anything past this, or are we done? Yeah, you know, it's funny because when I set out to build GenAI API, my idea wasn't to be a course guru guy. That wasn't the business I was trying to start. It just kind of happened. People said, I talked about how I built this business in a weekend with AI and people were like, will you teach a course on that? I was like, no. And then I was like, well, sure, I'll teach one. And I put it up and I was like, it'll be $1,000 or $1,250, whatever it was. And I filled 16 slots my first 20 minutes after I posted it. And so I was like, well, that's interesting. I just made $20 grand. That was fun. And then, you know, and I just decided at first, actually, I was doing it literally twice a week. I was teaching this course twice a week, and I just really actually kind of got exhausted. And then I started, and then we started moving more into kind of going back to some of the consulting I was doing, and we organized it more. And now we have our main thing is this fractional chief AI officer thing. That's what we do most. Private equity fund is buying 100% of the share. What's going on? So, uh, so anyway, um, I, I, I eventually pushed it off. So now I only teach this once a month. Um, and, um, I've had many people ask for the next step, right? The next step forward. And, um, I'll probably do that. I'll probably do that soon. So you're, you're in the, you're in the club now and, and, you know, always happy to, happy to help. And, and, uh, at the end of this course, um, I'm gonna give you guys, you'll have my cell phone number. You'll have, you know, there's a private Facebook group for anybody who's taken this course. And so there's a couple things that, that, you know, you guys will be in the club, right? And so you'll be able to ask me questions and get like, you know, help here and there. But, but again, if you really want to go dive deeper into, into more AI stuff and kind of take it to the next level and build more advanced things and workflows, and, um, they certainly can. But I, again, the idea here The idea of this course is not to sell the next course. And most courses are, by the way. I don't know how many courses you guys have taken online, but I've paid for plenty of them. And I swear it's like, okay, now that you're here, you can enter my, you know, super secret private mastermind and it's $10,000 a month to be part of it. And there's only 7 spots available. I don't know. Yeah. And that's fine. That's fine.

Good for Russell Brunson. Right. And you know, everybody, but, um, what about you create some Facebook paid Facebook group where we can all brainstorm, share ideas or you know, paid something.

I mean, I don't mind paying, but something where, you know, we run into problems or new apps or whatever. Yeah, we have that. So yeah, we already have it and it's free, right? Again, I— my goal is not to just suck out of you guys. I want to set you guys free, right? For lack of a better way to say it. I want you— I want you to, um, feel empowered to go and just do awesome stuff, right? And so, um, let me, uh, let me get this group here, uh, and just share it with you. Here, I'll share it. Let me share my screen. You can see what's going on here. But yeah, I have this group, How to Build a Biz or Build a Business with AI. There's 127 members of it. Not everybody that takes the course joins the group because not everyone has Facebook and some people just don't care to. But a lot of people do, right? And it's fairly active. But you could go in right here, just join it. Now that you've been in here, I would, you know, I'll just, I'll just, you know, click here, member requests, and I'll accept you. But, but people come in here and they ask, you know, we talk about different things and people ask questions in here. And there's some people that are really, really active in here that will answer questions like really, really quickly. You'll see Brantley in here. He's one of my employees and stuff like that. And anyway, but there's all kinds of just things that people are asking about and questions that are going on. So, so that's there. Yeah. Do your member request and all. Yeah, there's Marta right there. So I'll just literally approve, approve, approve. So that was actually an old one that I hadn't approved yet, so I'm glad I clicked on there. So anyway, guys, yeah, that's, that's a free resource to help you keep learning. But, but again, what I really hope you get out of this, guys, is that you don't need me to teach you, right? That you just kind of learn how to learn about AI and, you know, kind of where to go to explore and find things. And I'll show you at the end, I'll show you some of the favorite places I go to learn, right? Like, obviously I've got my Manus thing teaching me, showing me headlines, but there's some social media pages on Instagram and places that I, you know, that I just follow in the that help me, right? X is huge. And I just, I'll open the platforms and I'm not kidding, I won't spend more than 5 minutes in them, but it's enough for me to be like, okay, I see what's happening. And then sometimes I'll see something where I'm like, wow, that is cool. I need to go try that right now. And I'll jump on and I'll try it, right? And I learn by just experimenting. I'm just curious. I'm just naturally curious. And that's how I build stuff. Okay. So we've been taught, we've been doing this kind of text to Speech. I think that's cool. Let's go to agents. Agents are even cooler. Okay, so we could build— let's go to agents here and let's just build a new one. Okay, so we're going to say new agent and it's going to be a blank agent and we're going to call it Jeff and we're going to create it. By the way, guys, how much do you think ElevenLabs costs? Anybody know? $11. That, that would be good pricing model. It's actually $5. $5 a month will get you access to this platform to be able to do all this stuff. It's amazing. Okay, so we're here, we're going to set up a new agent.

Who thinks they can help me set this up?

What am I supposed to do here? Copy from chat, whatever you created, or copy my prompt. This is my system prompt. Yeah, you got it. Okay, so let's come back over here wherever the heck— I have too many things open. Okay, I think it's this one right here. We're going to just copy that. Hopefully that's properly pasted. There it is. You're a gem. Awesome. Okay, and, um, and you know what I'm going to do? I actually am going to go into— just to make sure we do it right, this is important because I actually gave it Too many things, too many things open. I gave it actually, um, okay, I'm gonna say this now.

Show me the prompt. Show me the full prompt now.

So just so I don't understand, can you clarify the agent versus Custom GPT, is that the same thing, or is agent same thing? Yeah, I'm using that word interchangeably here.

Yeah.

Okay. You may not attack race, gender, religion, health, body, or identity, use slurs, promote harm. Yeah, we don't want to do— we don't want to do any of those things. Good. Okay, um, let's see here. You Hurry, go faster. Okay. So we're going to copy this now. I think it's done. It's almost done. Okay. There it is. There it is. Okay. So they have, they provide a little copy button here. So we're gonna copy this and then we're gonna head back over here and we're actually gonna replace it. The reason I wanted to get the full one is because current user context, we wanted it to have access to that as well. Okay. So that's cool. And then first message. What's up? You're gonna say what's up? Yeah. What's up? Okay. And then we're gonna do— is it Eric? Is that the one we wanted? I just added one. Was it Christopher? Hey guys, use this voice to express excitement. Describe a promo. Yep, we're gonna use Christopher. Okay? And then I can come in here. I can change stability, speed, similarity. I can kind of mess with that model if I want a little bit. But now this is just about ready to use. I can go into workflows, and I can build automations. I can do branches, and I can add knowledge-based stuff here, adding documents, analysis, and there's a lot of different things This one is actually a cool part here, the tools. You can give it the ability to end the conversation instead of just constantly talking forever. You can give it the ability to detect a language and even speak a different language. You can give it the ability maybe to transfer to a number. This is something we actually built for Balance. We worked with Balance and we used this tool and they had something that required like, okay, we need to get a bunch of information, but sometimes you need to talk to a human. Great. We turn that on and connected that workflow and tested it to make sure it worked right. Very simple, but this is how that was done. Anyway, that's really all you need. All you need to do is just fill this out, pick the voice, first message, and you're done. But you see how it's basically the same as setting up a custom GPT. Setting up an agent is almost exactly the same. Now we can preview this thing. I can just test it in here. What's up?

I'm thirsty.

Thirsty, are we? A groundbreaking revelation. Tell me, are we talking just finished a marathon thirsty or haven't had anything more exciting than tap water all day thirsty? Give me something to work with here. I just need a good drink to relax this evening with. All right, relaxing evening drink, a classic. But relaxing is a spectrum, isn't it? Are we aiming for unwind after a long day relaxation or— Anyway, so it works. It works now. I don't know if I love this voice. And so I could go back and say, all right, I want to change the voice. I think it was a little bit too— I don't know. It wasn't exciting enough for me. I need somebody. It's got to be a guy for sure. Anyone who takes the time to listen is either an old soul This one actually could be funny if he's, you know, roasting me. Definitely need like a Matthew McConaughey voice. Let's try this one. What's up? I want to drink some root beer, but I don't have any. Help me come up with a better drink for myself. Root beer, huh? So you're chasing that classic, slightly medicinal, aggressively sweet vibe. We can definitely do better than just not having any. The real question is, are we aiming for something that elevates that? Yeah, I actually kind of love that. So we're going to go with Jeff. We're going to leave Jeff with this kind of commercial voice. And so then in order to move this over here, this is super easy. Okay, right here. And you don't have to know this, but I'll just show you because I do know it. We have an ID. This is the agent ID. So I would copy this and I'm just going to come over here to Replit, and I'm going to say, let's add a voice option to the chat. When you click on the bubble, it should default to text, but there should be a tab to switch to voice. We'll use an ElevenLabs agent. I have an ID for you when you are ready for it. I have a new ID for you when you are ready for it, but you should be able to pull the, um, ElevenLabs labs account that I already— or API, I'm just gonna say API— I already have on the account secrets. Okay. All right. So now we're gonna— oh, I don't want to plan. I don't want to plan. I like copy. Actually, you know what, we'll just let it, we'll let it do the plan. Not set up yet. You already have the API key in your secrets. I don't see it stored yet. We need to get connected. Okay, what is the ElevenLabs agent ID for the voice assistant? Nope, I copied that accidentally. Alright, let's come back over here. Copy the agent ID. And we'll paste it. Send. Got it. Okay. So, it's got that. Here's the plan. Now I need to get the API key properly managed, then I'll build it out. Start building. We need— this is important. We need to make sure that agent ID is only served up through the Secrets Manager. Okay, this is really important, guys. API keys, if you, if you build them in the wrong way, they can be exposed and other people can use them. You don't want people to use your API credits. That's the only risk there is that they use your API credits. And but, but if they use it a lot, it could cost you money, right? I've seen stories of people waking up and like, oh my gosh, I have $800 charged to my APIs. That's because people were dumb and they let the API key get just put into the, you know, it was, it was public instead of actually doing it correctly. This is correct. Okay, so we've got the API key. We're going to get rid of this. We're going to paste. I think that's the right one. Agent 1701. Yeah, that is, that is correct. Okay. Yep. Now we're adding it through the Secrets Manager. This is correct. Okay, good. Okay, amazing. It's okay if you show the API key and tell it to the agent here. You just have to make sure they don't come and write it into the code. That's the big thing, right? I'm sure most of you know that when you're on the internet, you can come over here to more tools and developer tools, and you can see the code of how, um, here we go. Elements right here. You can see the code of how every single thing is written here. If I put, push this button right here, I can see everything. I know exactly where all these things come from. Right. And it shows the exact code. And so if you have your API key exposed in a place that isn't actually required, the way that we, the way that we're setting it up properly is it's actually not exposed on what we call The front end. This is the front end. Remember, we're going to store it on the back end in a database and we're going to put a request for it. But if you're— if it's not an authorized request, it will say no. Okay. And so, and people won't even know it exists. They won't even know it's there. So they won't be able to find it. So if you do it properly, that's good. Now, in order to do it properly, Replit will by default do it properly. I will tell you that you don't have to know to set that up that way. The other thing is, remember this security scanner, okay? The security scanner would identify this. It would say, you've exposed an API key. We need to hide it. We need to set it up properly. Right? So as long as you just go through the proper steps, you'll be totally fine. You really will be fine. It's more than capable of catching these things. Okay. All right. So we're almost done kind of with agents, guys. We've learned how to build a custom one inside of ChatGPT. GPT in its nice little easy interface. We've installed it custom now onto a website. We've now added voice to it. We've figured out how easy that is. These tools, you just have to kind of open it up and use them and it just fill in what it tells you to fill in. And then just ask the agent, ask Replit, your developer to install it for you. Okay. So it's setting up all of these different things here. I've got everything I need. This is great. Backend, update the widget. Yep. Hasn't updated the widget yet. It'll probably have to do a quick refresh on that in a minute. All right. Just making sure my blood sugar's good here. All right. So we're gonna let this keep working. It's not done completely. We're gonna come back to it, right? I'm, I'm at the point now where there's some other things I wanna make sure we hit. And, and so I don't want to take too much time just sitting here watching this work. Okay. We will come back and we'll play with it. We're gonna dive in now for the last hour, 15 minutes or so. Actually, we are gonna spend the next 10 minutes talking about something and then we're gonna dive into that. So we'll do that, I guess, while this is going, but we're gonna, we're gonna dive into images and video. That's gonna be the primary thing. And intermixed in those tasks and working on those things, we'll talk about sales and marketing. Okay. So, okay, let's, let's jump into discussion mode for a minute here. I'm actually going to stop share here and we're just gonna talk for a minute. Okay. So I talked about this surf, don't swim concept. This is such an important thing and this is more than just an AI discussion, but it's extremely applicable during especially times of change, but it's really an entrepreneurship and business. Decision you have to make, right? How many of you guys have ever surfed in the ocean or swam in the ocean? Let's just say swam in the ocean and felt the power of a wave. Okay. How many of you guys think you're stronger than a wave? Yeah, me neither. I'm a whitewater kayaker. I love to kayak. I love water. I'm a crazy kayaker. I have huge 40-foot waterfalls. Falls and Class V rapids, and I've done it. I used to be sponsored, and I just don't have time to do that anymore at that level, at least that much. I do it at that level, just not that much. But anyway, bottom line is I know very intimately how strong water is, and it's amazing. Like, when you measure cubic feet per second, right, or whatever, and you measure rivers, a good average river might be like 1,000 cubic feet per second. Provo River, you know, for those of you that are around here, is somewhere around, I don't know, maybe, maybe 300 to 600 CFS is usually its range. Right now it's probably like 40 CFS. But, um, the Snake River, the Snake River might sit somewhere between 2,000 to 4,000 CFS. There's a little stream, stream, right, that's up in, um, up in the mountains here, probably about an hour from where I'm sitting, called Sixwater. Anybody ever been to Fifth Water Hot Springs? Anybody locally? Anybody? Yeah, you have, Eli. Cool. So it's right up in that area. It's same part of the same kind of river system. But Sixth Water runs at 31 CFS, and it is a series of huge waterfalls that I can kayak, right? It's only 31 CFS. It's really, really small, but it's incredibly powerful, and there's no way I'm going to beat it. Okay, so if we can't beat 31 CFS and we feel like we can't overpower a wave, then we most certainly cannot overcome a tsunami. And a tsunami right now is what is happening with this technology, with this AI technology. It is hitting our world so fast that if you are not ready for it, you're going to get smashed against the rocks and you're going to die, right? Your business is going under. The things that you're trying to do are going to just— the way that we do, the way our society functions is going to just dramatically change. And we talked about some of that yesterday. So how do we avoid that? How do we ensure that we are surfing? So as the tide rises, as that tsunami, whatever, right? We're always on top. If we're always on top, then we're not going to get smashed against the rocks, right? We're just floating. As AI gets better, what happens to us? We get better too, right? And so we got to make sure we position ourselves in a place so that we can surf instead of swimming., right? And so for the, for the conversation here, surfing means as technology around you gets better, does your product or offering or service get better or worse as a result of that change? Okay, so let's, let's look at a couple examples. Let's talk about Netflix.

What does Netflix do?

What is their core business? What's their core IP? Anybody know? Streaming services. Streaming, great, correct, not wrong. Okay. Subscription. Subscription, yep, good. There's a subscription streaming service. That'd be a great description of what they do. They also are a multimedia company, right? They, you know, now they're creating some of their own content, right?

That's good. Um, what else? Is there anything else that they do? Can you guys—

well, they also did their own, they also did their own, you know, like Netflix original productions. Yep. That's the production company. Yeah. That's what I was talking about there. They create their own content. Right. So they're doing that. Their core technology. Why are they different from Hulu or Disney Plus or whoever else? Right. There's tons of options out there. Why is Netflix valuable? Hey, could you just mute?

There's a lot of sound coming from your— not sure why. Okay. Yeah, thank you. Um, their optimization algorithm. Okay, optimization of what?

Of like, based off of what you view, it like knows what to recommend you according to like your past watching history. Bingo. Dino DNA. Sorry, that's Jurassic Park. But yeah, that is exactly right. That is exactly what, what they do. They have the best content recommendation engine. Did you know that at any given time on Netflix, there's somewhere between 20,000 and 30,000 available titles? All right. Have you guys ever gone to like an Airbnb? And you log in, or you just like turn on the TV and it's logged into someone else's account, and you're just like, what is all of this? Like, this is— I feel uncomfortable. What? Log out. Get me to my account, right? And how many of you guys do you think you've even scrolled past 1,000 titles on Netflix? Like, that's a lot. I might have maybe 2,000, maybe 1,500. Even then, I'm still only touching 5 to 7 or 8% of, of the total available, but it just knows what I want, right? Okay, so they're actually— core technology is, is content recommendations, right? What other technologies are they, are they using, right? Streaming. Did they, did they build streaming? Did they invent streaming? They did not. They use Akamai, right? That's another service that does streaming. That's their technology. So they just surf on Akamai. Okay. Did they invent the subscription model? No. Right. They looked at other models and they said, we're going to copy that. Did they invent the internet or are they responsible for the internet? If the internet goes down, some huge line gets cut in the Atlantic Ocean at the bottom of the ocean that's connecting, whatever. Do they have to go fix it? No, they don't. Right. So let's ask these questions. If Akamai, the service that makes streaming possible, gets better, does Netflix get better or worse? Definitely better, right? If the internet gets faster, hey, now we can do 8K, now we can do 16K, now we can do 32K, right? As these other technologies that they're built on that they didn't build get better, they just naturally get better. They could wake up tomorrow morning and Akamai comes out with a new release and boom, You have a better product, right? That's amazing. Okay, so Netflix really is doing quite a good job at surfing, right? They just maintain kind of one little piece, core piece of technology, and most of it just sits on top of other things that just as they get better, they get better.

Okay, let's look at another company. AT&T. Is AT&T a surfing company or a swimming company?

Surfing.

Why?

They don't actually own the towers. They don't build the towers. They actually use the towers. They actually do have to manage those towers.

AT&T builds the network. AT&T was one of the first ones to build the network, actually.

My mistake then. I was going off of Verizon. Yeah, it might be Verizon. There's other companies like T-Mobile, for example, that don't, right? But AT&T actually does build the towers. They build the towers. When you switch from 2G to 3G to 4G to LTE to 5G to 6G to whatever, they have to go replace and upgrade, and they spend billions and tens of billions of dollars, hundreds of billions of dollars to manage the infrastructure that makes what they do possible.

Okay, so can anyone think of a technology out there right now that could potentially make AT&T obsolete? Well, would that be Elon Musk's partnership with T-Mobile, uh, where they're going to start utilizing, uh, space technology versus, uh, uh, land-based Earth technology?

Forget T-Mobile. They're using—

they're surfing on Elon. Elon's the technology that they're trying to compete against.

That's what I was— that's what I was trying to articulate. Yeah, totally. Yeah. So yes, Starlink. What if Elon— who do any of you guys think it's beyond Elon to say, hey guys, I decided to make Starlink free for the entire world for the next month. After that, it's $10 a family. He could just knock out every telecom company in the world instantly. Now he has to solve the problem of how does he get the signal to the devices. And there's other things to worry about there. But the point is he has he is an existential threat to AT&T and to T-Mobile and to all these. So some of them are, you know, hey, we're going to partner with you, we're going to be friends so that we don't get, you know, steamrolled here. Right. That's, that's certainly what you're talking about there with the T-Mobile thing. Right. So AT&T, guys, is really a surfing— or sorry, a swimming company, because they are responsible for building all of the infrastructure. That makes it possible. If they stop building, their product does not get better. If someone else does something better, they have to try to keep up. And so what this means is these companies have to— if you're a swimming company, you have to spend a huge amount of money just to keep up, right? Let alone lead, right? If you want to lead, you gotta spend even more money because you gotta figure it out and you gotta be the the pioneer and the one who's figuring out how to do that. And that's even more expensive. All of you guys, including me, we're all entrepreneurs at different stages and whatever. We're all building things and wanting to start a new business in a weekend, right? It's part of entrepreneurship, right? One of the biggest risks to entrepreneurship is how much money you spend in building your product and producing your product, right? Especially before you've made any money, especially if you're really early on, right? If you're like Tyson, we've already got a big business going, that's great. You can spend money, you can invest it because You have money to spend. Amazing. Still, you, you have to be, you, you want to spend as little as possible if you can. And so, and so you have to, you have to try to figure out how to construct your business in a way where you can take advantage of these technologies instead of trying to compete with them. If you tried to build an LLM right now that's competing with ChatGPT, like if I was like, you know what, I like Jeff, I'm going to try to sell Jeff. Right now I'm going to make him more general, but I mean, he's going to, he's going to still have his little attitude or whatever, but he's really just going to be general. He'll, he'll answer things as well as ChatGPT, and then I could go and sell a $20 a month subscription. You can talk to Jeff instead of to ChatGPT, and it would accomplish all basically the same things. That's actually not a terrible idea because I'm not actually building the technology, but if I was trying to build the LLM that it's built on, fighting against Elon Musk's and Sam Altman's and everybody's hundreds of billions of dollars and all of their servers and everything you have to have to run it, there's no way I'm going to win, right? So I want to build on top of these other technologies in order to surf it. Okay, here's the final example I'll give. This is, uh, my old chief strategy officer, um, in my company. I don't know, 6, 7 years ago is the last time I worked with him, but, but he and I, he and I still remain good friends to this day. His name is Jeremy Whit, and he is writing a book called— actually, he's not writing a book, It's done. You can buy it now. It's called Surf, Don't Swim. And, you know, he, so he really came up with this concept, but he's a, he's a developer, but, but kind of an old school developer. He learned how to code in the '80s and kind of, he made a bunch of money during the dot-com bubble. He's probably about 20 years older than me, but he, after he sold some money in the dot-com bubble, he bought a nursing home. He's like, you know what? I think nursing, people are always going to get older. People are gonna have to take care of them. I think it's a pretty, recession-proof and pretty good business, gonna last a long time. So he bought a 200-bed nursing home in Kansas City. Okay? And so that's been just one of his sources of income for a long time. And AI comes out and he's like, this is interesting. One of the, one of the main problems, one of, at least a big problem that happens at nursing homes is diseases of the mind, right?

Alzheimer's, dementia.

They just kind of start to lose it. And why do they lose it? They're not talking enough. Okay. And so he said, okay, I know that insurance will actually pay for me to have people come visit and talk to my residents. Like they will literally do that. Insurance will literally pay to just have someone go talk to people, right? Just to keep them, keep their minds active. 'Cause it helps, right? They would rather put, it's preventative medicine, right? It's cheaper to prevent than to treat. And so he said, what if I just, hook up 11Labs and a couple other systems and build this kind of business that will automatically email nursing homes, sell it to them, email back and forth, do the scheduling, call up, you know, Betty and Sue and Bob and all these residents at any time automatically and have an agent just chat with them for an hour a day if they want, 4 hours a day if they want, call 'em at any time, call 'em at a scheduled time, whatever it is, right? And they'll just have a subscription, they'll just pay for it.. So he calls, he calls up the insurance company, says, hey, would you pay for this if I, if I made this? And they said, yeah, we'll pay you $90 per head per month as long as they're using it. And he's like, sweet. 90 times 200, $18,000. That's cool. All right, awesome. So he built the system out, took him like 2 days to divide, go this whole thing and put it all together. He is now making $3 million a year passively. He does nothing. He has one executive assistant that just manages the billing and answers some questions on email, but most of the emails are answered by agents. So he has an email agent, he has an organizer system, he has, he has all, he's connected a bunch of things, but he's surfing on a ton of technology. He didn't actually really build any of the technology. If ElevenLabs goes under, or 11 Labs. Let's say if 11 Labs gets better, his product gets better. If 11 Labs goes under, no big deal, he just swaps the API key out for another voice agent because there's a lot of different voice agents out there. If the email agent, someone else comes better, I just swap the key out. You can just hot swap all these things super easily and you can just get better and better and better. And he just sits there and it prints money for him. He is completely surfing on a bunch of other people's technology. He invested almost zero, like literally less than $1,000 invested. His API cost at the end of every month, he told me, are less than $2,000 out of all of that. $3 million of business, $2,000 of cost. That's his margin, 99.999, whatever that is, right? It's unbelievable. Those are the kinds of opportunities that are here today when you understand, okay, I'm gonna surf, I'm gonna, I'm gonna, I'm gonna figure out how I can just let other people take the risk, and I'm going to have minimal cost on top of it to then figure out some unique way to present this to the world. That's where, that's where the magic lies. By the way, is he talking about that a lot, that business a lot? No, he doesn't want to just give away his little secret sauce, right? It's, it's so, it's so easy, right? And so you guys are all sworn to secrecy on that. But, but the point is there are so many opportunities out there like that right now, and I hope you guys can see how that all fits together. I hope you think about your own business and say, okay, I could build right now into my— I could vibe code into my solution. You know, maybe I, maybe I need— let's start sharing screens here again for a second. Maybe I need in my solution a— let's see, where is it? A Calendly. Yeah. Yeah. Sorry. Yeah. So like, like this, I need, I need to be able to schedule a consultation with Jennie IPI, right? I could choose to, cause it wouldn't be very hard. Rebuild the functionality of Calendly here, right? I could easily do that. Any of you guys could rebuild Calendly. How? You go to Replit, you say, build me a clone of Calendly. I need to connect an API to my, my Google Calendar, and I need some sort of an interface like this, and we need to be able to send an email off for this time, you know, whatever. Okay. And it'd be like, great. And I bet you could rebuild Calendly easily. In an hour. Really, all of you guys could do it. And maybe 2 hours, whatever. Now you've built it and now you've saved that. Okay, but what happens when Google updates their API and then they change permissions on something? And then what if you want to be able to invite multiple people, multiple guests? Oh, I don't have that feature. I don't have this. I don't have that. I don't have whatever, right? And so you know what, it's actually just easier to pay calendly.com $20 a month or whatever it is I pay, right? View plans, right? $20, $12, whatever. It's, it's way cheaper for me to just buy this, surf on it, and then I get all these features. Look at all these cool features I get. I don't have to go build them. I don't have to maintain them. I don't have to do all that, right? And so don't, don't be afraid of spending money to make money. Okay? Sometimes it's just way easier to pay someone else to do it because they know it's better. And that could be, that could be, you know, you're hiring somebody to take care of something for you. Could AI do it? Could you build a system that does it? Sure. I'm sure you could. And sometimes that's the right decision. Here's a time when it might be the right decision. Maybe you want Salesforce. You need a CRM, right? They don't even put their pricing on here because it's ridiculous. Salesforce is so expensive. It can be like, $12,000 ahead per year. Like nuts. So if you have 10 users, but I mean, you're, you're talking, you're talking $120,000, you're talking a 6-figure contract to just take care of keeping track of things. Salesforce has lost a lot of value in their stock price over the last year, couple years really, because of AI and because people are realizing this. I want to build a CRM or I want to keep track of, because, because a lot of people don't use all of the features that Salesforce has. They need to keep track of their client names and the business name, and so their contacts, the company, and the, and the communication, or any deals that you have with them, right? So you just need to be able to track progress and whatever. But you can rebuild that thing in, in a couple hours, and then you have enough. You have the basics right now. If you need all the complex connections and things and whatever that Salesforce does, you really don't want to worry about it, and you're making tons of money, then great, go, go spend money, go spend money and spend $120,000 a year on Salesforce. But I don't do that. There's no way I'm doing that here at Jenny IPI because I don't need all the complex stuff. I just need basic stuff that keeps track of things. And so we've got our own stuff, right? We build our own things. And a ton of people are doing that. They're rebuilding their own things. They're vibe coding their own things, but there's most certainly, there's most certainly a buy versus build. Calculation and everything. And buy versus build isn't quite the same as surf versus surf don't swim. Surf means position your company in a way so that as technology gets better, you get better. Okay? And so Tyson, I invite you to do that. Brittany, I invite you to do that for your company. Think about this. How can we get better? I'll help you work through those, those thoughts, right? Eli, as you're considering, you know, going out and helping people with AI, one of the cool things I like about what you're doing and what I'm doing, right? We're helping people get better at AI, right? And Travis, same thing. Well, as AI continues to improve and more things come out, does that help our business? Totally. We have more to teach, more to learn, more to help, more people to serve, right? And so we are surfing even though we're not building core technology, even though, you know, I am over here at GenAI, we're building really cool stuff. But, but even if you're just an educator and you literally are building nothing, Right? That's actually kind of a surfing thing. It's pretty cool. I'm going to share the last example of surfing because I want this, I want to, I want to hammer this in. And this one, this example will only take me about 90 seconds. Okay. I'm going to introduce you to my mom, Carolyn Cheney. Okay. She loves to go travel and take pictures and, you know, she posts these little videos. Here's a, here's a lion, like 7, 7,400 views on that. That's pretty good. 216 likes. 36,000 views on the elephants. The elephants are really popular. 951,000 views. Look at that. How much do you think my mom invests in Facebook's infrastructure? Zero. She's not an owner. Even if she's a shareholder, even if she owns stock, because I'm sure she does, right? She does not invest a dime to make their streaming better, to make their algorithms better, to make their system better, to make their discoverability better, to make whatever it is, all these different things, right? She makes somewhere between $10,000 to $20,000 every single month. About 2 years ago, she started posting. It's just posting. Look how many followers she has now. 953,000 followers. Social media influencing is the ultimate surfing. Business. You have to invest basically zero. You have almost no risk. You do have to invest your time, right? You got to put some time and effort into it. But my mom is at the point now where literally she posts— this is 4 hours ago she posted this. This is February 9th. That's 2 days ago. February 8th. She basically does one a day. February 7th. So she did, and she's not even technical enough to schedule it.

She just like opens her phone, finds a video, and posts Right.

That's it. I noticed her description outdoors was misspelled as well. I love the outsource. Incredible. She's not even getting her— again, my mom is the nicest, most amazing woman ever. She is not technical. She is like, Jon, please help me. I can't remember my password. I don't even know where to type it in. Mom, it's written down on the paper right next to your computer. Oh yeah, thank you. Yeah, I'm like, she is not technical and she's pulling in like 6 figures from posting her elephant pictures and all of her travel is free now, right? Like by a long shot. And all of her camera equipment and everything that she buys, by the way, guess what she uses for the vast majority of her photography? iPhone. She's surfing on the iPhone too, right? As it gets better and the videos and the camera and the quality and the upload speed and all these things get better. Her business just gets better. Okay. All right. I hope I've driven that hammer in, right? Surf. Find technologies that you can surf on. Don't spend so much time investing. All right, let's see if Jeff works. Hello, how can I help you today? I'm thirsty and it's cold outside. I understand you're feeling thirsty and it's cold. Would you like some suggestions for warm beverages that could help you feel more comfortable? Yes, but why are you being so nice? I am designed to be a helpful and polite assistant. My purpose is to provide— Yeah, so obviously it again, and I'm guessing that a lot of times AI likes to default to kind of a more helpful, friendly, like, Persona. So anyway, I could, I could ask it. I'm not going to spend time right now because we only have an hour left. I'm going to spend time to debug this. But you know what I would do? I'd be like, it's being so nice. Show me the prompt. Did you change the prompt for this? Let's make sure we get this right, you know. So it should be— I'm surprised that it's not actually pulling the prompt from ElevenLabs that we, we set this. It should, it should be set right anyway. Okay, so, um, but it works. And actually, that was pretty smooth. I thought it was great voice quality. I thought it was very natural to tap and call It's saving some of this. It looks like the UI got messed up a little bit here. I don't have the ability to scroll up and down here, and I can't tap back over to the text option, right? And so I'd probably work on the UI a little bit, and I'd get the feel right, and that'd be good. But I added a voice agent now to my platform. Very easy. Okay. All right, guys, time to do some image stuff. This is gonna be fun. We're gonna work on a logo for Drinkable, and we're gonna work on a hero image for for Drinkable. Hero image is basically kind of the top image that you're gonna see on a website, right? If you go to gopro.com, right? They're the GoPro Hero, right? This is gonna be their hero spot. This is kind of the main thing that quickly tells you what they're all about. And anyway, so we're gonna try to design that for Drinkable and we're gonna use all my favorite tools to do so. So we're gonna use Gemini. Okay. Gemini.google.com. Why does that go so far down there? We're also going to use— come on. Oh, you know what? I know why. I know why it's not coming up. We need to use this browser. This is the one I need. And then we're going to use Midjourney and we're going to use ChatGPT and we'll use Groq just for fun, just so you guys can see how bad Groq is. But I like to do this, right? So first of all, we're going to start in ChatGPT over here. Nope, over here. And we're gonna say, um, give me, uh, let's see, I need a logo for Drinkable. Um, give me an idea. Give me 3 ideas. Be as concise as possible. I don't want it to just drone on. Bold wordmark. See, it's even like showing me the option. It's kind of cool. Circular badge with a little star, mark rating. I don't love that. I don't love this. I kind of like this drinkable thing. Um, clean, bold wordmark, drinkable lowercase, dot line becomes a small droplet. The infrastructure. Okay. Um, I wonder if I can click on this and then actually ask it to do something. I can't. I can't do that. That's okay. Okay. I like, I like this, the first option, but I love how the R and the I create a negative image of a drink. Give me a solid prompt I can use in other AI platforms. Give me an image. Give me a logo. All right. So, I'm going to use— I said, you know, normally I would be like, okay, I'm going to Midjourney. Give me a Midjourney-specific prompt. We're just going to go a little bit faster here. Okay? Which should be all lowercase. Drinkable. Is bold. Negative space. Color direction. Avoid literal glasses or straws, mascot script fonts, final result. Okay. Yeah, this is good. Okay. So we're going to go copy and we're going to go to Gemini and we're going to tell it to do it. And we're going to do this. And I'm actually going to say instead of design, I'm just going to click that. We're going to go here and then we're going to go to Grok and we're going to go here, create images, go. Okay. So the reason I do this, guys, is because I, I have— I subscribe to all of these different things.

And yeah, look how terrible. Crocs sucks, man.

Um, that's just bad. It's not impressive. That's pretty good. Yeah, Drunk Cable. Yeah, exactly. This is— yeah, Drunk Cable. Um, yeah, it does not look like drinkable. Drunkable is actually kind of funny. Um, and yeah, this is still not working. So this was— this is Gemini. I think it did okay. This looks like it says, "Drunkable." Okay. Drinkable. Okay. Here's one thing that Midjourney is terrible at. It's actually terrible at word marks. It's amazing at images. Look at this crazy image. It just had two young men playing the violin and the piano. Look how intricate this is. It's just amazing, the details. Midjourney is absolutely bar none my favorite, favorite thing. These are other logos and things I've had it work on for Hero, and I had it create this thing, and you can actually animate it in here. These are some treasure things I do. I have a fun little side treasure thing that I, that I do. I have a page called Treasure Map on Facebook, and I just post fake pictures of treasure maps and I get hundreds of likes on them and people talk about them and whatever. Sometimes I'll post real ones in there, but I just post stuff and it's just kind of for fun. It's kind of a social experiment. Anyway, look at this image. This was one I created where I wanted to get something that's like, took all your action gear and helped you organize it in a piece of software. So you kind of, you know, your kayaks are stored in the barn and your, your bike's down here and you keep your helmet in this closet and here's the serial numbers and here's the warranties or whatever. Anyway, so this was an image I was creating for that. This is a girl, you know, looking at a little baby elephant in her hand. It's just like beautiful, beautiful art. I love Midjourney. I could play with it all day long, but it cannot spell. It's terrible at spelling. Okay. ChatGPT. Tree. No. Yeah, I'm not impressed by any of these, by the way. None of these pass my test of anything. And so I might have to say, okay, I need to do something else. Okay. Try again and make it better. Okay. I want to go back to ChatGPT. This is too complex. Let's just do a symbol. Or a, a simple symbol of a cup or something. Cover the straw, but make it iconic. Give me the prompt and create the image. See how well it does that. Giving ChatGPT sometimes multiple things to do. Doesn't work well, but let's see if we can do it. Let's see if it can give me the prompt and do the image. But now it's thinking forever. I don't want it to think forever. We're going to go back and we're going to— okay, drinkable. All right, well, it went with this. Okay, that's something we've got. We've got a little thing. This is actually brandable. I could work with this if someone showed up with this. Here's what I'm looking for in a logo, guys. I need it to not be a paper cut. That's it. How many of you guys interact with a business because of their logo? Tell me. Give me an example, Brittany. What's something where the logo itself actually caused you to buy it? I know you might buy Nike because you know it, but let's assume you don't know it.

You see the logo and you're like, I like that logo so much, I'm going to buy their product. I have to be thoughtful of that. Um, actually, I'm looking at like this EOS logo on this lotion.

Like, I love the simplicity of the lotion. I really liked that, the look of that brand. You know what, I actually will agree with you on that one. I bought EOS stuff like chapstick and things like that because it looked clean. It's a good clean logo. It did. It looked very clean. I love it.

What about you, Travis? What was one that the logo actually drove you to buy something? It's actually kind of wild that we're talking about EOS because that is exactly what I was going to say. That's awesome. It's just, it's very simple, clean, inviting. And it honestly feels trustworthy. And their product just immediately complements that by like the soft matted feel of and texture of it. So it's just like, it feels luxurious.

It looks luxurious and you just immediately start start trusting it. Yeah, I love it. Yeah, logos can be so important, guys. And so I, I spent, like, when I, when I built GenAPI, I spent a good amount of time on the logo, right? I wanted it to feel trustworthy. I wanted to build professional. I also liked, you know, I went with— this is the logo I went with, right? I like the idea. My idea actually was I wanted a brain with, like, circuits, and I was like, my idea was how can I make AI and human work well together? That was kind of the original concept. I want to make that that, that relationship better. And so this actually did pretty, pretty good. Is it perfect? No, but it's pretty dang good. And the thing I like about it too is it's circle. And so it fits well right here. If you see, this is called the favicon and you can have your Replit and say, hey, install the favicon. So when logo shows up, right? And it'll, it'll do that, right? Or if you don't remember the word favicon, then just say, hey, you know that little symbol that shows up on browsers?

Can you install that and make that work? You know, Anyway, but you can get—

I think Poppy has a really good logo too. Poppy is a good— I agree, Poppy's a fantastic one. Yeah, if you want to see good logos, go walk around Costco. Costco somehow just picks good brands and they have amazing buyers and they put a lot of effort into bringing the best brands. But part of the best brands is that logo, right? And so I think that matters. And I think you definitely need to care about the logo enough to get it to where people are like, okay, I get it. Simple, awesome, whatever. At the very least, don't let it be a paper cut, right? Don't, don't use this one. Okay? Don't use the one that's going to throw them off or confuse them or whatever, right? So you got to get it good enough. But sometimes you can just do something as simple as if we ever get this image. I don't even know what is happening here. There's no way it is still creating this image. Refresh. Thank you. Okay. Stylized cup with teal liquid. Okay. I could use this. I could use this totally as my logo.

Okay.

Um, in fact, we're just going to go as simple as that.

Style.

Yeah. Let's see if I can just copy this. I don't know if I can. No, I can't, but that's okay. We're going to go stylized cup with teal liquid. Liquid. Okay. And then we're going to come back in here. I'm going to try it again. Tools, create images, paste, go Midjourney, go ChatGPT. We already did, so we don't do that again. And I'm going to say create an image of a stylized cup. Okay. So we're going to try this again. I'm going to try a couple of little variations. But again, I usually, when I'm trying to do something very quickly, I'm not going to spend forever on the brand. I'm not gonna spend forever on the logo. It's gonna be good enough. This is not a great option here. These are not great options for logos. And I didn't, I didn't really hold on. Let's try this. Size goes to liquid as a, uh, logo. I'm just gonna give it, I'm just gonna give it that. Um, it needs to be simpler and, and will be used as a logo. Um, make it a simple image. Okay, okay, these are terrible. Again, I told you guys X is bad at it. It's just not great. It's not very creative. Some things it's good at, it's not good at that. It's good at market research. Okay, this is awesome. Now we're starting to get into some, maybe some logotype things. They're a little bit complicated. Maybe you could maybe work with that. It's still just kind of crazy. I like to think about, would this look good on a hat? Right. And, you know, could I, could I brand this? I like the simplicity of this. I also like that this could very, very easily be thrown up in, you know, right here in the top left. It could be thrown up as a favicon. You could throw it on a hat. You could put drinkable underneath it. Maybe it's a combination of this with the, with the little drinkable, with this, right? You know, you could, you could do whatever, or, or even just literal, literal simple text like that. So that's enough on that, on that thing. But I just want to show you guys how I go through these different places here and do that. Now, let's say that I was going to go with this kind of more complex one here. Midjourney has the capability also of animating. And so we're just going to ask it to animate it and we'll just say we want a loop, high motion. Okay. And loops can be really, really useful. It can bring your website to life if you do it in the right way. It could be something cool. Tool to post places. Sometimes social platforms will let you do a GIF as your image, and so it's an animated thing or some sort of looped image. And so sometimes that's useful to have. It's incredibly easy to create this. Unfortunately, with Midjourney, you don't have the ability to really tell it what you need it to be. Like, hey, I want the image to do this, and I want it— I want, you know, this to come up and then circle around and do that. And you don't really have that kind of control. It's kind of like it's kind of like playing the lottery, right? Or you just, or it's, it's just kind of random. You're rolling the dice, see what happens. Okay. So we'll let that work for a minute while we go explore some other things here. Um, let's now work on a hero image, right? I need a hero image for the marketing website. Okay. Now I, okay, this is great. Um, now I need a hero image for the main, um, marketing page. Um, what could we do here? Okay, so we're going to ask ChatGPT for some ideas. Again, we're back to my project manager, understands everything about my entire business, all of my business plans. It knows everything, right? So that's why I'm having that conversation here. I don't have to explain myself. Okay. Okay, these are kind of cool. I think this is all right, except they're extremely boring. I need something that's a little bit more— it's just gonna be way more exciting than this, right? I gotta be like, anyway, so I don't, I don't love these things. I want it to be more exciting, but I do love the aisle explorer idea. This needs to be way more interesting than the images you gave in option 1. Okay, let's see if we can get to some sort of something that looks okay. Okay, it needs more cinematic energy. Here's a stronger version. That's kind of cool. This feels like a bar. Not exactly right. Anyway, person casually browsing drinks, hero image prompt. Okay, I like this. I like this actually. This is good. Okay, so we're going to copy this and we're going to go to Gemini. And we're going to do this. We're going to go to Midjourney. We're going to ask it for this. And we're going to try— we did ChatGPT already. And then just for fun, we'll do this. And then we're going to open one more. We're going to add Runway ML. Runway ML is one of my favorite platforms. It's more for video, but they do images as well. And so we're going to just try this. Let's go image. And here's my shot. Okay. And 16 by 9. Generate. Okay, cool. We're also gonna open Flow. Flow is by Google. Also known as Veo. Okay. By the way, Eli is an expert at video creation. I would say he's very, very good at this. He's taught me a bunch of cool things. And anyway, there's a lot of different tools that he uses, I'm sure, that we could use. There's also KLEEN, and there's so many tools, right? KLEEN.AI and Higgsfield.AI. There's so many different video and audio— or sorry, and the visual creations thing. I guess audio is correct as well.

KLEEN creates just unbelievable content. I, I'm shocked by what Klean does. Go ahead, Eli. Um, yeah, I'd love to, I can leave in the comment, I'll put all of the platforms I use. Awesome. Um, and then before I put it, just like one sentence summary is like Higgsfield is the best if you want to use different platforms. And so you just use credit, you have a certain amount of credits per month, but then if Jon, you go to the top left and click, like press the hover over the image or video, you'll see that all of these different video editors like Veo and Klean are accessible. To you. And then like there's just little things like Klean is the best at like fast action shots and like motion. Veo is my favorite just all around. And then with like image, like Nanobanana Pro is super good for cinematic shots, but then like there's like Kaspa AI if you want to do product photography specifically.

And I'll leave like some links to like videos and images made from those platforms too in the chat. You guys get bonus content. That's a huge amount of value he's about to give you guys. I promise you. Thank you for sending that, Eli. I think that's incredible. I'm excited to have you in the, in our Facebook group as well. So you can, you can teach everybody else some of these things. I think it's, I think it's awesome. All right, let's see how these, these different prompts are doing here. We've asked it to create a person. Yeah, see, this is, this is starting to feel cool right now. We're starting to get towards, this is kind of a neat hero image that I could actually use on my site. Right? This is cool. One of these could work. Right? They're not all perfect, whatever, but they're cool. I like them. We're starting to get towards the right thing. And this is, again, something where it's just, what do you think? Yep. That's pretty cool also. I actually don't mind this at all. This is cool. By the way, if I don't use it for my hero image, where else can I use this? Ads, social media. A header of a blog or, you know, whatever, right? There's so many different places you can put this thing, right? And so, and you can come in here to like Midjourney and say, you know what, I really like, I really like this, but I want a bunch of variations with this. And so let's go vary and we'll do strong. Let's do strong variations. And let's also, I actually really like this one. And so we'll vary strong. And so now I'm just creating 8 more images, right? And it's creating all these different variations of these things. Starting from this is the kind of core concept and idea, but varying it for whatever reason. It's going to be different types of people or maybe different colors. A lot of it's pretty similar. This is all pretty similar stuff, but they all are unique images. But I can just kind of come through here and, and I could even say, hey, you know what? I want to zoom. I want to zoom out. How do I zoom out? I know I can do that here. Zoom. Hold on. Let's do that. Okay. All right. So yeah, now I have different options here. I've got, this guy's got a drink in his hand. He's just came out of the, out of the aisle. Now he's got a drink in his hand. That's cool. I like that. Okay. Just little variations, right? Just the things they're wearing and stuff like that. Whatever. Super fun. Super fun to come in and play with this. Let's look at our animations here. See if any of 'em are good. I think the only one that's semi-usable is the fourth one here, right? It actually has a good thing. But what if that was your logo just sitting up in the top left, right? So, so watch this. I can then take this and I can, whoops. All right. How do I download it? More options. I don't want creation options here. I want to just download. There we go. Download as GIF. You can download as a video or for social media or whatever you want to do there, right? Then I can take this over here into Replit. And I can say install this GIF only in one place, right where I've selected. I'll provide different logos for other use cases. But this is— but I want this always visible, top left. Yeah, you know, I don't even need to say that. Well, we're just gonna say this. Go. Okay. So in a minute here, we'll come back and that will be installed. Okay. And that's as easy as it is. Just drag it in, tell it what you want to do and let it, let it figure out how to, how to format that and make that happen. Okay. All right, let's go into— well, why is that? Why is that doing that? Okay. Let's go into Runway here and see if we got something good. All right. So we got somebody, let's pick, I don't like the drinks in that one. Feels very Asian. This one's all right. This one's all right. We can start with this, but I can then use this. I can take this image, use it, and then say, all right, now describe my shot. She quickly scans the drinks, picks one, and turns around and walks down the aisle. Now it's possible she's gonna do some weird stuff because you have to have— you actually can do some things that make this a lot better. I'm gonna make this 8 seconds long and we're gonna do— yeah, we'll just generate. Okay. All right. But this is kind of the workflow. I'm just showing workflows and ideas so you guys can then take that and apply it in your own way, right? This will take a minute to go ahead and create. I'm going to come in here to Nano Banana. And what we'll do actually is just for fun, we will take the same image and we're going to download it. And then we're going to go into Flow. And we're going to click new project. And we're going to give it ingredient. I'll go frames to video, and we'll say this is frame 1. Uh, crop and save. Okay, wait, what the heck? Okay, good. Um, and then I'm going to give it literally the same prompt here just so that we— I just like testing. Okay. The best way to make a really, really good prompt, guys, is to do it with ChatGPT. And you can ask it to do very, very specific things. So like, we're making an 8-second video here. You could say, all right, from 0 seconds to 1 second, this is what I wanted to do. From 1 second to 2 seconds, this is what happens. From 2 seconds to 3 seconds, this is what happens. So every second is very specifically choreographed and planned. Lighting, audio, whatever it is you want it to do, it can do all of these different different things here. Okay. All right. We're at 27% there. 25% there. Let's see what else we added. Create here. Oh, let's go back to, let's go back to this. There we go. So now I got a little GIF. It's cool. So now my website is kind of just feeling a little bit more live and it looks good. I like it. This kind of brings that, brings that piece to life. Now I could put that. In, in other places too. Um, I could maybe give Jeff something. I could give Jeff like a little animation of having a, you know, he's a person, he's kind of just like moving kind of like a little loop or something like that. You know, you can just bring things to life however you want. Super easy to jump in and make that happen. We still haven't built community. We won't, I won't build that right now. But, um, any questions on any of these platforms, guys? I know I'm blasting through these quickly, but Midjourney for images. Is just my favorite. Um, Runway and Flow are gonna be some of the best for video. Um, that I, I use a lot, I guess. I don't know if they're the best. Let's watch the videos. That's amazing. That is crazy. Let's see if the variation, if that's any good. Made a weird noise on that one. There's also really, really loud walking with her tennis shoes, right? With her sneakers. And you can replace the audio, obviously. Yeah, this one, it didn't show her shoes. And so anyway, the point is, that's amazing. I'm super impressed by how well it did that. Lots of times when you try videos the first time, it does some weird stuff, but that motion was pretty dang good. That was pretty clean. I'm excited to see what Veo does— or sorry, what Runway does here. One of the best models right now that people are loving is Kling 3.0. Kling 3.0 is building unbelievable stuff. And the best way to learn about this, guys, the way that I learn about it is, again, you come in here, you're like, holy crap, how am I supposed to know about all all of these things, right? I literally just be like, okay, Quing 3.0. I'm usually going to go towards like the most recent things that are happening, right? Beo 3.1, Sora 2, Quing 3.0, whatever it is. And I'm just going to go to YouTube and I'm just going to see what other people are doing. This is how I learn, right? And this is what I'm trying to do is help you guys kind of, you know, branch out now, now that once my class is over, you can keep learning and keep growing. But I'll just go Wing 3.0 examples, you know, and I'll just watch it. See what's going on. How AI videos always look. You can use it completely to 6 scene cuts in a single generation. It's gotten a massive upgrade in how it understands camera movements or an over-the-shoulder shot. Wow. The scene cut had perfect timing with the explosion. And it really understood the camera briefly loses orientation part, which made the shot that much more realistic. Most AI models usually give you terrible textures when generating text. That's amazing, right? So you can learn, you can learn like what it's doing and even how to make better videos, right? YouTube's incredible. It's obviously free, so you don't have to pay for another course. Marta, I'm happy to, you know, do private training with you if want, but, uh, this is cheaper. This is way cheaper. Um, jump on YouTube and just, you know, find the few platforms. Obviously super grateful that Eli's putting this together here and helping us with all of this. Um, yeah, I need to make sure I copy and paste all this stuff before we finish up here. But, um, but anyway, um, very cool. So, um, yeah, Cling 3.0. So cool. So cool. Um, Klingon 2.6, Grok Imagine, Higgs Field. They have their own model even. Okay. Now I want to, I want to shift a little bit, guys. Obviously we could spend a lot of time playing around in here and seeing, you know, what kind of things we create. Let's watch at least this creation here and see what we got from. Yeah, that didn't work very well. I didn't like how natural that was.— but the rest of the walking was actually pretty good. So, as long as you cut this out, you could start it kind of right here. Right? She picks it, walks down the aisle, boom. I've got somebody, you know, picking this and this. I can use this obviously in my thing and I could cut it. Now, you're probably going to want to know how to use just basic video editing stuff, but there's free apps on your phone you can get. If you know how to use Premiere Pro and things like that, you can do that with Adobe. It's all so easy these days to be able to know how to do that. Um, I don't know this, Eli, does, does, uh, Higgsfield make it easy to edit video inside of here? Like, is this Cinema Studio? I actually haven't tried.

I think I've heard that they have a, you know, the ability to like splice things together. Yeah, that's recently updated a lot. And that's something with Klink 3.0 as well. It's something that no other video AI has ever been able to do is called the multi-shot feature. So you can get the different angles of the same setting.

Um, but I just use CapCut usually. It's just easier for me, at least. Now here's a cool thing. Watch this. Um, we're gonna switch to— let's see here. Um, I want to use There we go. Yeah, we're going to edit the video. And we're going to say, um, um, the girl is— yeah, replace. Girl with soldier, the male soldier. Generate. Okay. It's amazing what you can do with— and we're gonna do this. I'm gonna say— I'm gonna keep the same video. I'm gonna say there is no ceiling. It has been ripped off by a tornado. Things are flying everywhere. Things are flying everywhere, and you can see the sky, but she just kind of ignores it, and her motion stays the same. She just grabs her drink and walks away. So this is using something called Aleph, Gen 4 Aleph, right? Where you can take a video and you can just completely change it. You can say it's sunny outside, you can make it dark, and, or it's, it's, you know, rainy or whatever. Or like maybe you're taking a video of your family. Family, right? You're taking a video of your family in Times Square, and Times Square is packed with people, and you say, remove all the other people, right? Or it's just you. You just want a shot of you walking alone in Times Square. Everything else, all the billboards, all the videos, everything else will stay the same. Take your phone video, upload it, tell it to Aleph, and it will just change it. There's some really, really cool things if you go on YouTube and you look at Runway Aleph examples. Yeah, here we go. Different angles. This is some of what Eli was talking about. So you can just completely change what's possible. So fun. I wish we could sit here and just watch all these videos because they're so fun. But we need to, we need to just kind of finish up here in the right way. Yeah, really, really, really, really amazing stuff here, guys. Yeah. So now we've got a soldier grabbing his drink. So pretty much everything else is the same. Grabs it. Yeah, there was, there was some weird stuff in the back, so it had a little bit of a problem there. So I wasn't super impressed by that, but you could, you could clean it up, right? I'm excited to see if this, this one actually works where it's, you know, she's, things are flying around everywhere. But guys, now how do we use this content to, to sell more? Let's jump over and talk about sales and marketing a little bit. Let's talk about some principles at first and I'll show you a few platforms that I think are cool and worth using. Do you guys want to know what my number one platform is for sales? What do you guys think it is? LinkedIn. Bingo. LinkedIn. And I know that sounds like, well, that's not really helpful. Right? But it's true. Wait, she went to grab the drink, but okay, they didn't do what we wanted there. But, you know, whatever. Maybe she just pulled it out of the tornado or something like that. That's fine. So anyway, AI is still getting better and lots of room to improve, but this is as bad as it's ever gonna be. It's only gonna get better and better and better and better and better. And, and I, yeah, there's so many cool things in here, guys, that we can play with all day long and there's just not time. We could do an entire course just, just in Higgs field. Right? Eli, let's do that course. That'd be fun.

But yeah, I think a ton of people would love to just learn about this and spend—

could I, could I show one quick thing before we move on from it?

We've got, we've got bonus from Eli. Yeah, this is, this is something. Yeah, I'm just super, super passionate about it because I've been able to make more money in like a week than I made my whole entire life just from like this skillset. And like one, one, you just can get so creative with like working in your company or working as an entrepreneur of how to do like video ads or images. And like one thing I did was, uh, I don't know if you've heard of a vision board, but it's where you make images of like what you want to become. And for me, it never really resonated so much because it was just like Google internet pictures. But I don't know if you can really see, but these are all images of me specifically in the future. Like one of them is me on my marriage day and one is with like a million dollar check. One is at my— me in my outfit I wore on Thursday looking at my house. And that took me like 5 minutes to generate all those images., and my friends were like, oh, we'll pay you $30 for just like 5 of those images.

And there's just like so many ways that you could like monetize that as well. Amazing. And that's what I'm wondering, how do you guys monetize? It's, it's all amazing, but like what's the shortcut for that?

How Eli says, I've never made this much money. I'm like, what? How? You know, right now both Eli and I are working in, and Travis are working in, in this space of AI enablement AI training, helping companies. But the reality is, now that you guys are done with this course, you've taken this course, you're at the tippy top of the people just in terms of knowing what's possible. Are you at the tippy top of having the practical hands-on experience like I do? No, not yet. That's okay. That comes with just now you just gotta go do it and keep following up with it. But, but you're now in a position where you could say, okay, I think I could walk into most companies and tell them things they can do that they're not doing right now. So now you have knowledge, you have a knowledge gap that you can then go in and you can monetize, right? And I'm doing it at a super elite level, right, where I'm working with tons of companies, right, dozens of companies, um, simultaneously, team members and all these different things, and building out and, and, and building systems to make it all happen. But that's just the natural evolution of of that. And so if that's interesting to you, if you're like, hey, you know, I think I could teach people this, great, go do that. That's one way to make money, right? But again, I don't like courses to teach people how to make courses. You know, I want you to take what you love. You're only going to stick with a business if you really, really love it, right? Marta, you were showing, you know, all this makeup stuff and things that you were doing, right? Whatever, whatever your business is, right? I'm assuming that's because it's something you're really interested in, right? And so now you got to figure out how to make money with it. How do you make money? I don't know. Is it subscriptions? Are you selling products? Is it affiliate links? What is it? Right? And so at that point, the real thing is you come up with the idea, the solution, and then you just need to get people to know about it. Right? And one of the people that I really, really respect in terms of, um, how he does his marketing— I'm going to grab this book. Normally it's sitting behind me, but somehow it moved over here. And you guys know this guy and I already brought him up. Gary Vee. I get paid nothing for having her asking you to buy this book, but I think you should buy it. Day Trading Intention. I don't care what you are. I don't care if you are a brand new entrepreneur or the CEO of a huge company. This book will cost you $20 and it will teach you and give very practical examples, by the way, right? And they have it on Audible too. I listen to it on Audible, right? And I have the book too because there's, you know, actual examples. Here's what you should do Here's why it worked. Here's why this, you know, how this generates money. Here's how much money this generated. Here's what it did for this business. Here's how this happened. Here's what doesn't work, right? It's so amazing. But the basic premise of this is if you are not using the most powerful free distribution system in the world, you're an idiot, right? Instagram, Facebook, X, LinkedIn, Threads. YouTube, TikTok, Snapchat, wherever your audience is. You don't need to be on all of them, but identify where your audience is and post like crazy, right? If you're in staffing, talk about staffing every day. Just say something. Hey, we had this experience. We had a client that had this problem. We fixed it for them, right? Hey, we couldn't fix this problem. Any ideas on how we could do this better, right? Whatever, like Good, bad, the ugly, the nice, the whatever. Okay? That's one piece of it. Just talk about it. The fastest way, the best way for you to get no customers is to not talk about it. That's it, right? And I know that sounds crazy and simple, but it's true. The way that you get people to know about something is you talk about it. Now, you can use AI to automate some of this. I feel less worried about automating what a business is saying. So if I have GenAI PI automatically posting for me and I have AI helping me grab stuff and filter in things and do all that, I have no problem with that. That's less important. If I, Jon Cheney, am posting, I'm going to care a little bit more because it's It's kind of my identity that people see, right? And so I'm also— and here's another big thing that I do. How many of you guys have posted on your personal Facebook page, whatever social media page, for me that would be a Facebook or an Instagram, your personal page about what you do for work? Marta, I'm not surprised now that I'm getting to know you more. Brittany, yes, a little bit. Yeah, but most of you are keeping your hands down, right? Um, why not? Why—

for those of you that are no, why, why have you not posted about what you do for work on your personal Facebook page? I keep them separated from professional to personal. Um, anything professional goes to LinkedIn, and then also vice versa. I don't put anything personal on LinkedIn. Um, it just doesn't— it doesn't feel natural to me.

And maybe that's just because of the perspective blocker that I have with it. Okay, good. Anybody else? Anybody else have a different reason than that, or is that kind of in line with why you're just like, yeah, it's just kind of like buckets, places I need to put those things? If you go and look at Gary Vee's page— how many of you guys follow Gary Vee?

Yeah, Travis, who's Gary Vee's favorite football team? The Jets. How do you know that?

Because he's always talking about them. And even though they're not great, he is a diehard fan.

And what does he do? What does he do for the world?

Like, what's his job? Like, what does he do for the Jets?

Or what does he do? What's just his job? What does GaryVee do? How does he make money? He is the best, in my personal opinion, he is the best content creator of all time.

Yeah, he is just constantly putting out content, like multiple times a day, every day. Yeah, and he's just talking about various amount of things. Exactly. He's talking about his personal life and his professional life and everything in between and whatever comes to his mind.

And he is omnipresent. He's omnipresent.

He posts 60 times per day. Dang, I didn't know it was that much. That's crazy. And that's across all platforms, right? And so, but, and it might be formatted a little bit. It might be the same content, but maybe instead of this way, it's this way, or it's cropped, or it's whatever. It's just a clip of it. Instead of 40 seconds, it's only 12 seconds, you know, depending on the platform. And he has a team doing that for him, right? He does have a team. He's got the VaynerMedia social team doing that for him. He probably has 20 people helping him do that. Cool. But you can do that with AI. You don't need to do 60. You don't. But it is very possible as an individual to post to 3 or 4 platforms a day and to post a couple times on some of those platforms. If you— do you know how many times— so I post on LinkedIn, I'll talk about religious things, I'll talk about kayaking, I'll talk about piano, I'll talk about AI, I'll talk about entrepreneurship, I'll talk about fundraising, I'll talk about current events, I'll talk about whatever, right?

Do you know how many people I have had reach out to me and say, dude, I saw your kayaking post, also you do AI stuff, we should chat? Okay, so you're just—

you're getting that omnipresent approach of like, just like anybody that can relate to something that has nothing to do with what your professional life is doing actually creates a professional opportunity. 100%. And vice versa. I'll post professional stuff on Facebook. Now, I will say this: I post more personal stuff on Facebook and more professional stuff on LinkedIn. So it's not just I'm taking all the same stuff and putting it everywhere.. It's just, okay, I'm maybe 30/70 over here and I'm 70/30 over here, right? Just based on— so I make judgment calls and I might take the same picture and adjust the caption a little bit, right? I might say, like, let's say I'm going to say something about God on Facebook. I'll say something about it and it might be very personal and different.

And then on LinkedIn, I'll say it and I'll say, and this is why it matters for me in the workplace. This is how I have noticed that you, that you've done that.

Yeah. Yeah. Like, what does it taught me about sales, right? Yeah, totally. Yeah. Yeah.

Yeah.

I was kayaking and that led me to think about sales. No. Anyway, um, that's the kind of classic LinkedIn post, right? But, but, but the point is, guys, you can adjust it. But I really want to encourage you guys to post more. I promise you, none of you are posting enough. And two, to vary the things you post about. If you just always talk about staffing, you're pretty boring, you know what I mean? It's not that Tyson or Brittany are boring, this just happens to be their job. But if that's all you talk about, then that's what it is. But if you can add some depth to your character and, oh yeah, you like to go running and you like your dog and you like whatever, right? And, and, or you like dogs and you like this, and, and you can make— if you want to talk about your dog, you can bring it up in a business type setting. Like I said, you can say, hey, Hey, you know, this new thing came out, or hey, I just switched this food and it's amazing the difference that it makes, right? And, you know, I was thinking about that and how it relates to this. If you get just the right person in a staffing company, you know, it makes a huge difference, right?

You switch the person for another one. But anyway, so go ahead, Brittany. Did you want to say something there? Well, I was just going to say there are plenty of really funny content creators in HR and staffing and things like that that I follow.

So you can for sure get creative with it to draw attention. Yeah, you absolutely can. And I just want to encourage you guys to not be afraid to be the same person everywhere. It's so easy in life and on social media to, to just say, okay, I'm in this room, this is who I am. Now I'm in this room and this is who I am. You know how much easier it is to just be the same person everywhere? Like, I've— I'm speaking from experience of having been on both sides of it, right? Where I'm just like, this is who I am in this room, and this is who I am in this room. But when you really just say, okay, this is who I am to the world, and it's okay if not everybody likes me, right? The right people will stick around, and the people that want to do business will self-identify. The vast majority of my business at GenAIP is inbound. People just see that and they say, hey, I want to talk to that guy. And then we find a way to work together or not. And then we're just friends, right? I don't need to ever take another dollar from any of you guys, and I'll still be your friend, and I'll be in your network, and we can connect, and we can share funny memes. Um, I just genuinely want to help people, and if I can and we can identify that opportunity, then great, we're going to keep working together, right? And so, um, I know that's not really an AI section on sales and marketing, you can use AI in all the ways I've shown you over the last 2 days to amplify and do these things, to create content faster, to comment on things, to consider your ideas, to bounce your idea off something like, hey, how do you think this would hit here? Is there something I should tweak here? Yeah, if you change this word to this, it might hit a little bit better. Hey, thanks for that suggestion. I like that. Let's throw that in. I'll do that. I'll do that with AI. I'll ask it. I'll say, I don't want you to rewrite it. I just want your feedback, and then I'll rewrite it. Right? I love doing that kind of thing. You can use AI, you can use all of these amazing video tools and image tools, and you can create a website. You know, I, I'm, I'm a BYU fan. I want to post about BYU. How do I relate that to AI? I have Manus go and create a prediction, and then I share it and I teach people about this agentic tool. Now I'm talking about BYU, super authentic to me, football, AI. I'm providing value. Right? I don't just provide pictures of my dog, right, or my horse. I try to share it in a way that somehow whoever sees that actually will get something out of it. And so that comes in the text that I have to accompany with. Um, and I promise you, this book, this book, Day Trading Attention, is amazing. But ultimately, guys, the people that make the most money— Marta, you're like, what's the secret sauce? How do you make the most money? Get attention. Get attention. Right. You and I can talk more later for sure. But you've seen that firsthand. You've seen what getting lots of attention can do, right? Good and bad. But, you know, that is how you do it. You can make money if people know what you're all about and what's going on. Okay. So, guys, I'm really grateful to have you in this course over the last few days or a couple of days, I guess. I already uploaded yesterday's session. I sent it over to Eli because he missed it. He had to step away for a minute. But I'll send it to everybody. They'll send an email, just reply all to the email that, that we, uh, actually no, I'll do a new, I'll do a new email. I have kind of a good form that, that I send out, but it will have a couple things in it. It'll have the replays, it will have the link to that Google Drive. I will put the link for the Facebook group again. And, uh, so if you didn't join it, you can. And again, no, don't feel any pressure to join it. It's just if you want it, it's there. It's a free resource. I'll also throw my cell phone in there. Some of you already have that, and that's great. But, but I'd love to connect and help you guys with what you're doing, right? If you do have a business that you want me to work on a little bit more, I can, I can help with that. It wouldn't be in the format, form of like another course or something like that potentially, but, but I'd be happy to, I'd be happy to work with you. You know, you could, you could become a client if you want. Again, that's not the purpose of this, but, but I certainly won't you know, turn you away. I'd love to, I'd love to work with you. I love working with the people. I've had many people that have come through these courses become clients, and, um, and we're doing, we're doing really fun, fun stuff with a lot of people from, from startups, like, I've got an idea but I need somebody to just help me do it, and all the way to 5,000 employees. I think it's the biggest one, big public company actually. Um, and so yeah, their CEO joined and they're like, I need you to work with my company, we got to So anyway, really, really, really, really good to have you guys. I really liked a lot of the insight I got from, from all of you. You all shared wonderful things. It was great to have some of those discussions. Remember to surf and not swim whenever you can. Keep following the progress, keep watching the cutting edge, the forward edge of all these things. I said I'd share some of the places I like to learn. One place you can go, I'll share, I'll just share my screen again for just a just a second and just go into Instagram. Instagram is actually a place I see a lot of things, and obviously X— I'll click on X in a minute as well. But Instagram, if you just go to GenAIPI— don't follow us. I mean, you can follow us. Follow us. I have— we have 233 followers, whatever. We're— this is not a huge place where we get clients, okay? But it's kind of there. But the important thing is, who are we following, right? Who are we following? And this, the, these are the ones that I follow. Luna.ai, Meta.ai, Perplexity, Insider Tech, Google Gemini. There's one called Metaverse that I just freaking love wherever it is. Meta, Metaverse, Metaverse with a 3 right there. This is one of my favorites, right? They, and he does like all kinds of, you know, tech, new tech, right? So you'll get, you'll get Metaverse stuff in there. You'll get 3D stuff. You'll get AI stuff. You get crypto stuff, you get all kinds of things, but it's very not salesy. Like if he's talking about crypto, he's talking about how is this changing the world and why is this going to make a difference? Not like, hey, go buy this token. You know, it's— he's really, really good about actually giving you good stuff. Business Insider, things like that. But this is how I learned. These are the actual channels I follow in order to keep up with things. And so just go to GenAIPI Institute. GenAIPI wasn't available. I don't know why, but someone had taken it for something. So Jenny IP Institute is on there. X.com, you can follow me on there as well, or follow us on there, I guess. Follow me. If you really want to learn, just keep following me on LinkedIn and other places because I post about all the cool stuff that I see that I like. But if you look at my X here, right, like, boom, AI agents for Coinbase and being able to transact here. AI article on this. This is what's going on. And here's AI this. I gave an AI $50 and told it pay yourself or you die. 48 hours later, it turned $50 into $3,000 and it's still alive. This is definitely fake. I'm just going to tell you that. I know it immediately. I've seen this a lot. This is actually people on PolyMarket trying to get people to sign up for PolyMarket. So it is actually kind of fun. And I actually legitimately want to try this and actually see if I can get it to do it. That's pretty fun. I gave an AI $50, told it to pay herself. Are you dying? I'm going to save this just so I remember the idea anyway. But this is what I do. I just, I just start following people. And the cool thing about social media these days, and you'll learn a lot more about that in Day Trading Attention, the Gary Vee book, is social media is now much, much more. And this is why, Travis, you shouldn't worry as much about what you're posting. People are shown things they're interested in. Less than— so instead of it being a social graph where it's, oh, I follow Travis, so I see what he says, it is following more of an interest graph. So it's following the topics. If Travis posts about AI, amazing. If he posts about, you know, I don't know, what's something that I'm not interested in?

Rugs.

I don't know. Then I'm— I, I might not see it. I probably won't see it. It's not something I'm actively following. But somebody else that doesn't even follow you that is interested in rugs would be like, Who's this Travis dude talking about rugs?

And I like rugs, right? Anyway, so you see that a lot. Like, I, uh, love to follow like a lot of book talk, uh, or yeah, yeah, reviews. Book talk. Yeah, so yeah, book talk. And some of my favorite videos are the AI-generated book trailers. Like, those are really cool. I love that people have invested time in them.

I love the creation in imagination around some of it. They're really cool. I like them. I love that you said something really cool that I really love that you said is I love the imagination and creativity that they put into these AI videos. And a lot of people were worried as videos started to come out, like, we're going to lose that ability to create, right? But, but no, we just have more ability to create now. We can do, we can do anything we can imagine and we're not restrained buy resources, right? To create many of the videos that we created today. I mean, just, just these simple whatever, right? To create this, I know it's not great, but to create this with other, you know, effects would, would take a long time and, and some real tools. And this one, I'd have to, you know, get a soldier and get somebody that has this, these clothing and find the, the vest and find a store and go down there with a camera, right?

And I don't have to do that anymore, right? I can— I was paying attention.

I think about on the Super Bowl, probably at least 70% of the commercials that I saw were completely AI generated. Completely. You can kind of tell, but yeah. Totally, totally agree. It's only going to become more and more. Does that displace a lot of work? Yeah, it does. It moves it, it shifts the talent, it shifts what's happening. Those shifts have happened from the beginning of time, from when the wheel was invented to when the internet and everything in between. And, and now AI, and, and it's going to continue to move if you guys continue. I hope, I hope this kicks off more of an active, um, you know, Tyson, you said at the beginning, you're like, you know, I just, I know I need to lead by knowing, right? And I hope this unlocks some of that because now if somebody's— you ask somebody for a report or they're taking forever to analyze something, you might, dude, go use Manus and get this done in the next 20 minutes. Like, I've been asking you for this for 3 days, right? Or someone gives you a report or something and say, you know what, use Manus to build a website. I need an interactive website so I can click through, or I need a calculator that allows me to actively mess with this stuff. You can create that stuff in Replit in 5 minutes now, right? You can do so many amazing things. I invite you guys to try things like, hey, I've got a PowerPoint presentation I need to deliver, and you've got the PowerPoint, drag it into Replit and say, hey, make this this into an interactive website.

See, instead of just delivering a boring PowerPoint, you deliver that presentation.

That's what I was literally doing this morning. Cool. Amazing. So cool. And it's fun, huh? It's, it's amazing what it does. And you're just like, this is so much better.

If I'm sitting in front of a client or a coworker or whoever, and I'm trying to present something, now I'm at the tippy top of like, wow, Brittany knows what's going on. It for sure takes practice. I can tell I need practice.

It's—

that's okay.

Nothing to write home about quite yet, but I'm going to keep, keep working on it for sure because there's value in it. BYU is one of my clients and they are sitting down and working specifically with the athletic department there. Right. They have 600-something athletes and, and really just trying to help them learn AI. My job is to help the athletes learn AI so that they can be better prepared for the world. Right. Because not all of them are going to go pro. Right. Most of them won't. And so the, the main guy, he's kind of like their principal, basically a guy named Travis Daugherty. One of the things he has to do is he has to review their grades and, um, and then, you know, work with, work with students. He has to provide reports, look at the teams and how's everybody doing. And he, and he puts together all these like graphs and things like that. And all he does is he gets dozens of Excel spreadsheets of reports from teachers and grades and all these things. And then once a month he has to spend about 3 or 4 days compiling all the data and putting it into a presentation that he can then show the higher-ups, the athletic director and other people that he reports back to. And I said, do you have like a folder of like last month's stuff or whatever? He's like, yeah. And he just, I said, okay, let's drag that into Replit and ask it to create an interactive page that summarizes all of this for you automatically. He's like, there's no way it can do that. We drug it all in and it's just like 10 minutes later, probably. He's like, you just saved me 4 days a month. Month of time. And now I'm presenting something that's interactive and we can dig into the data in a way that I've never been able to even do before. AI is enabling things that you just have to keep trying and pushing. And I wonder if— but don't just sit there in the shower and think, I wonder if. Get out of the shower and go try it. That's where most entrepreneurial ideas die, is in the shower, because you get out and then you got to dry off and you put your clothes on and then your kids talk to you or your phone rings or you open your email or you get in the car to go do something and you've just forgotten and your idea is gone, right? But think of the idea, make it happen, follow up, and you can just do incredible things. So guys, thank you so much. This was incredible. I'm gonna drop my cell phone number actually right here. Those of you that want to, I know some of you already have it, please send me a text. I just dropped it in the chat. Just send me a text and just say who your name is and I'll be happy to be connected to you and we can, we can continue this relationship.

Thank you all so much and have a wonderful day.

Thank you so much, Jon.

This was great.

Thank you so much, Jon. This is super valuable. Appreciate your time. Okay, see you guys. Thank you. Learned a lot. Thank you.