Cohort 16 - Session 1 - 7/16/2025

Not least. Okay, so first off, guys, just a couple, you know, housekeeping items. This is a pretty casual class in terms of like, you know, you don't have to be all, you know, dressed in a suit and ready to rock. And I'm not one of those teachers. It's like, don't eat during class. This is do what you want. If you want to turn your camera off, you can. I do love to have cameras on. Especially during discussions, right? There's going to be a couple of times where we're like, okay, let's not look at the screen, let's look at each other and just talk for a minute, you know? So here in the beginning, that'll be great. Obviously, you know, a little intro here would be great to have everybody's cameras on. But, but I've got, you know, my potato sticks. I might eat a couple of those while you guys are talking. And I've got some water and drinks and stuff here. And we will take a break sometime around 8. It kind of depends on the flow and how fast we get to a certain point. It'll be somewhere between 10 and 15 minutes, but that'll be it. I really can go 4.5 hours straight, but most people are like, "Dude, let's do a bathroom break." So, we'll probably do that. Also, this is being recorded, right? And so, if for whatever reason, you know, a quick call comes in and you miss 20 minutes or something like that, like, you know, that's your choice. Take it, whatever you want, but you will get a recording of it by the end of tomorrow when I send out the follow-up email. I'll upload everything and you'll get all 9 hours of content. You can go back. I'm a fast talker. Okay. And so there's going to be some times and I'm fast typer also. And I just, I move fast and I will stop and be like, okay, catch up guys. Let's be very hands-on. We're going to do things together. But, but if you feel like you just, man, I can't keep up with that task, just be like, okay, I'm going to do that. I'm going to watch that again and take a little note of what time, you know, you wanted to go back to that. So you can easily scroll through a 4.5-hour video and find that. But, but anyway, so it'll be recorded at the same time. Please speak up, right? I don't mean like I want you just being lost the whole time. Say, hey, Jon, tell me that again. Hey, what was that prompt again? Whatever. I will be throwing some prompts and things in the chat. I'll just, you know, you'll see me write it, type it on screen. Here, let's turn off the— let's go into focus mode here. We're good. We won't have any more dings. But anyway, yeah, I'll drop a bunch of prompts into the chat and things like that. You'll want to use those. So with that being said, I want to do just a quick intro. Everybody gets like 33 seconds, uh, each person, and I just want to hear your name. Obviously we can see your name, but it's nice to hear you say it so we don't say it wrong. Um, and where you live, that'd be great. What you do. And then the last thing is, do you have any coding experience? That's the last thing I'd love to hear. So name, where you're from, what you do for work, and whether or not you have any coding experience. We'll start, we're just going to go in the order that I'm seeing it on screen. Sam, Caleb, George, go for it, Sam. What's up, guys? Sam here. I'm from Minneapolis, Minnesota. I work for a clothing company actually, sports marketing, so brand marketing side of things, and I do not have any traditional coding experience. Awesome. Thanks, Sam. Caleb. Hey everybody, I'm Caleb. I live in Middle Tennessee, just about 45 minutes east of Nashville. I'm a financial advisor with a company called Raymond James. I also do not have any coding experience. Very nice, Caleb. I go out to Tennessee all the time. I'm working on a big fun real estate project out there. That's— anyway, I end up in Rock Island State Park. My, my business partner lives there, looks like right outside that. So not too far from you. That's cool. Awesome. Yeah. George. George, just outside of Portland, Maine. I don't have any coding experience other than what I've done through ChatGPT, and I don't suppose that counts. I bounced from entrepreneurship for quite a while, almost two decades, into government. At pretty high levels at the state level. And now I can't wait to get away from government and back into entrepreneurship. And that's where I'm headed. Very cool. Love it, George. That's awesome. Glad to have you. Glad to have you with us from, from all the way across the country. TJ, and then Khan. Yeah, I'm TJ Patterson. I'm in Mesa, Arizona, and I work for a construction company. We do industrial and a little bit of commercial construction, mostly in— we self-perform contract or concrete and wet utilities. Nice, very cool. My wife's from Mesa, so I'm down there all the time. Cool, love it. Kyle, go for it. Awesome. Yeah, Kyle Robbins here from Austin, Texas. I do fractional marketing, um, and I have ownership in a furniture business and a construction company. Uh, Yeah, pretty limited coding experience. Awesome. Okay. Very cool. I'm actually originally from Houston, born and raised in Texas. So, fellow Texan there. And then we'll go Ashton, then Troy.

Okay, Ashton. I live in Orem, Utah. I've done some consulting, tech consulting for the last bit. Also do some residential real estate investing and trying to jump into the commercial space. Um, but I've got an app that I've been, uh, I designed it all with Figma, but I've been paying a developer to make that. My background in coding is that I did mechanical engineering in school, um, and I hated coding. That was, that was my least favorite, least favorite part. Um, but I love what you can do with coding, and so I'm excited Like, I did enough to not appreciate it, to appreciate its value, but not appreciate doing it. And so I look forward to being able to do some cool stuff without needing to have a developer all the time.

Yeah, that's cool. I love it. Yeah, I don't know how to code either. I'll get to that in a second. But I understand your pain. Troy, go for it. Hey, my name's Troy Roble. I live in San Diego. I work in the solar business for about the last 15 years and currently have been doing solar finance freelancing for about the last 2 years. I got zero experience doing coding, but my buddy Blake, uh, took your course a few weeks ago, Jon, and was, uh, raving about it. So I signed up at Blake's referral. I love it. Well, that's awesome. I, I owe him a hamburger or something. Actually, I owe him $100. At the end of the course, I'll ask you guys for referrals. Anybody you bring on, you get $100. So, um, anyway, uh, maybe he already texted me. I'm not sure. I, I have a lot of people say, hey, my friend signed up, but But anyway, well, thank you guys. Do we have— is everybody here from a different state? Were there 2 people from Utah? I think everyone's from a different state. That's awesome. That's definitely a first. So something's good. Did— how many of you guys heard or like signed up for this because of Joe Rogan and that random thing? Anybody here? That's weird. I want to— we won't cover it now, but after the class, I want you to be like, here's how I heard about it. That'd be cool to tell me. I'm just interested in how the word gets around. So Well, that's good. That's exciting. Travis Chambers. Oh, that's cool. Travis is a good guy. He— good, good friend. I've worked with him. I hired Travis a long time ago when he was actually at Chamber Media. He's not there anymore, right? Doing all this Outpost X stuff. But he's a— he's a— he's a go-getter, man. He's crazy. He's always trying something crazy and he doesn't care what anybody thinks. So anyway. Great entrepreneur. So, guys, some of you might know me, some of you might not. I ran into Ashton, actually. So, I'm in Utah, right? He and I are in Utah. And I ran into him at a trampoline park the other day that our kids were both playing at.

And that wasn't me.

No, that wasn't you. That wasn't you. I'm sorry. No, Ashton's— I'm sorry, I'm mixing you up. Ashton's girlfriend keeps her horse right here in my backyard almost, right? Shares a fence with my neighbor here. So, yeah, neighborhood horse. For Ashton. That's how it was. Yeah, so anyway, random, random world. But anyway, I guess I'm in Utah here. Um, I live on a, I live on a 5-acre farm, and it's a beautiful spot called West Mountain. For those of you that are familiar with Utah, I'm staring out my window at the Y. In fact, you know what's kind of fun? I like to do this. We'll just show you, we'll just show you my view because it's just fun. You know what, it's a lot of work, and there's like a crazy amount of weeds, but, but that's what I get to wake up to every day. And it's just pretty, right? Anyway, mountains. It's kind of hazy today. I think it's actually smoke. There's a little bit of smoke out there from a fire. It's the season. It's like crazy wet on the east side of the country and really, really dry here lately. We're in a drought, so that's too bad. But anyway, guys, I'm in Utah and, uh, originally from Houston, born and raised there, and have spent a few, about 5 or 6 years out of the country. In Paraguay and Taiwan for different reasons. Well, actually, kind of both similar reasons, a little church-related. My dad was a mission president and I served as a missionary, um, and had good times doing that, but learned a couple languages, came back. And, uh, I was, I was actually at one point, you know, Ashton, as you said, you know, I hated coding. When you said that, like, the way you were talking about it is exactly the way that I think about accounting. I just did not like it. I was an accounting major at BYU, and I did that because my dad did it. And then he went and got his MBA and he made tens of millions of dollars and did just incredible. I'm like, that seems like a good pathway. I got into my first accounting class, I'm like, nope, I'm done. I knew I was an entrepreneur. So, I graduated in Chinese, actually, with a business minor just to get a couple of little things under my belt there. Chinese was the fastest way out of BYU because I already spoke it. So, I tested out about half of all the credits and then I just got out as quickly as I could. And it took me, let's see, I graduated high school in '04, graduated BYU in '12. So, I had a 2-year mission in there. But anyway, I took a long time to graduate because I've been an entrepreneur my whole life. I was running businesses, running all kinds of things. And I think that's one of the things that's going to make this class really fun, I hope, right? Obviously, some fellow entrepreneurs on here already and some people maybe wanting to get in the game a little bit more. But this is an entrepreneurship plus AI course. And I've been there and done it, right? Like, I haven't just started the little mom-and-pop, you know, I had a DJ business in college, right? That's cool. And I was making a few thousand bucks a month and that's a business. Yeah. And, you know, I think it's awesome. I love lugging around those. I didn't love lugging around the speakers, but I loved it when they were set up and I had them cranked and dialed. Like, that was fun. So anyway, that was fun. But in 2016, after, you know, a lot of my background before that was in sales, sales, sales, sales. A lot of door-to-door stuff and then joined a tech company that was based out of San Francisco, although I worked here in Utah office and did a bunch of sales, joined another company that recruited me over to basically run that company. I wasn't the CEO, but the CEO was kind of absent and I kind of ran the show and also focused on sales and building new products, all in the education tech space is what all that tech stuff was. And then in 2016, I said, you know what, guys, I'm kind of bored of education tech and, you know, Do you know how hard it is to sell something to Harvard? Like, it's hard, man, to tell them they're wrong and like they got to listen to me and do it my way. Now, obviously, that's not the best way to sell, but the point is it just moves slow, right? Education is a tough market to be in. And here I am back in it. So, anyway, in a little bit different way at least. But in 2016, I started a company and that became my first real, you know, big win, raised $13 million for a tech company in the augmented reality space. Actually followed in the footsteps of Pokémon GO. We started our first app. Our first kind of product was a location-based augmented reality. You pull your phone out and you'd see treasure chests all over the place and you could go open them. And when you opened it, you'd get like coins or money or things like that, right? It would be like $0.02 or $0.15, or if you're lucky, you got a dollar. If you're really lucky, you got $100, stuff like that. But then there would be like fragments. Remember McDonald's Monopoly? You get like Boardwalk and you're like, well, I have half a million dollars here. All I need is Park Place and I've got the rest, right? And so, we kind of followed that concept and we'd give away products from local companies. We signed Samsung and Six Flags and Universal Studios and all kinds of crazy fun companies we worked with. Sorry, not Universal Studios, Universal Pictures, right? The theme park was Six Flags. But anyway, signed a bunch of companies. That was really fun. And anyway, long story short, grew that to about $130 million valuation, ended up selling it for about a third of that. Um, you know, we kind of, we went like this and the peak kind of hit and then Mark Zuckerberg said metaverse doesn't matter anymore. And then, you know, all of a sudden it, it tanked our whole market and everybody kind of, a lot of contracts backed out. But anyway, we turned from a, a Pokémon Go style thing to a B2B SaaS company where we were enabling 3D and augmented reality on websites for products, right? So if you're shopping for shoes or like a couch or a table or whatever it is, you could use augmented reality right on those websites to see stuff. So our tech, Anyway, did some really cool stuff, got an offer from Snapchat in there. That was pretty fun. And as I finished that, I whipped up another tech company, raised another $3 million, and still kind of in the middle of that. Technically, this GenAI API thing was started as a challenge for myself. I'd been using AI quite a bit for a while and was loving it, right? For years I've been in it. I don't know, maybe a year ago or so, I started using this, this program Replit that you guys, many of you guys have signed up for now. And, um, and all of a sudden I was like, oh my gosh, I can build anything. I'm not a developer, right? I asked you guys, can you code? And I can't. I can't write one line. Like, I know what an href is. That's it. That's it, man. I'm done at that. But I don't know how to code. But I did have one advantage in that I was CEO of a tech company with, you know, 60 employees and half of those were developers. And I had awesome CTOs and VPs of engineering and people that I talked to all the time. And I'm someone who's techie, right? I understand and I enjoy the tech discussion, right? I like getting in front of the whiteboard and architecting and figuring it out. So I know how it all fits together and that, that has really helped me. So we're actually going to do a little bit of that. We'll teach a little bit of that today as we, as we dive in learning how to code with AI, things things like that to really, really push. But, um, but anyway, long story short, guys, um, it was a, it was a Thursday morning, and I, I post on LinkedIn every day. I don't use— I actually don't use AI to write my posts. I always sit down every morning, I'm like, what do I want to talk about today? Occasionally I'll schedule a post up for tomorrow, right, if I've already posted 5 times in a day. But I post a lot, talk a lot. And, um, anyway, I said, hey, I wonder if I could build a business like start a new piece of software using AI and then have my first customer paying for that by the next day, right? Could I do it in like a day? Is that possible? Really 2 days, I guess, Thursday, Friday. So anyway, started, you know, blogged about it, talked about it a lot, really just on LinkedIn, Facebook. And then Friday came along, worked on it all day. By 3:00 AM, I just wasn't done. I didn't have Stripe set up and I needed to be able to make money. And so I was like, this is going to be pretty long. It's 3:00 AM. I'm not going to try to do this right now. I'd never done it before, by the way. I'd never set up Stripe. And so, I was like, "But I know I can do it. I know I can figure it out." So, Saturday, get up, kind of go about, do some activities with my family. And then, about 4:00 PM, sat down, finished by 3:00 AM, just kind of cranked all day that day as well. And by Tuesday, I had my first customer. So, it took me 5 days instead of 2 days, right? And it was a $15,000 contract. Pretty awesome. Real contract, real customer, enterprise, really cool. I've already delivered on that, received the revenue. That was awesome. By the way, that was on February 27th that I did that. So that was just 4 months ago. And just today, guys, you're the only people that know this, and I'm not even sure if I'm allowed to tell you this exactly. I haven't signed an NDA yet, so I can, but keep it on the super DL. But a really big influencer that you guys all know who's really tall and has a hoarse voice and you know, is really, really awesome. You guys all know who he is. Um, is he and, and, uh, another really amazing guy that is a partner with him that's also a big influencer with millions of followers. Um, we're, uh, they're basically hiring me to build their AI program to teach the whole world about every single, you know, entrepreneur and small business. So that's their audience, right? If you know, if you know who I'm talking about. But anyway, Really exciting. Really exciting to do that. I've, I've really, I really care about you guys. Like, I've had a lot of people and this is kind of getting to the end. I've had a lot of people say, hey, Jon, can I give you some equity in my company so you can help me and advise me and keep helping me push this through quickly? And I've said no to everybody. I said, look, I'll help. I'll give some advice here and there. I don't have time to— first of all, I don't have time to take on hundreds of, you know, I've had hundreds and hundreds of people come through this course. And I can't say yes to everybody there, obviously. I can't even say yes to 20 people. That would be unfair to everyone. But I will give free advice. I love giving free advice, but I want you to keep your equity, man. Keep your company. It's yours. It's hard-earned. And I've had people steal equity from me and I hate it. I hate it. I've had people take control of my company away from me. I've had to fight it back and spend hundreds of thousands of dollars with lawyers trying to deal with that crap. So I've been there. I know that it's hard and I really, really admire all of you guys just for stepping into the ring. In fact, I posted just before we, just before we jumped on, I did a quick LinkedIn post and I put a picture of all of you guys on there, even though I don't actually have a picture of you. But let's see here. Let's see if I can pull it up here. Yeah, here we go. And I'll share my screen. That's you guys. You guys are the doers right now. So I'm excited to work with you guys today. Anyway, I was just saying like, yeah, hey, 10 minutes until we get going. So Cohort 16, This is awesome. Really excited to have you guys here. Okay, we're going to kick this off, guys. We're going to dive right into coding, right? Sounds like a lot of you guys have never done it before. And that again, it's me too, right? But it's super, super, super easy, right? A couple of you mentioned maybe I've played around a little bit with ChatGPT and tried to code some things. We're going to start today in Groq. Okay, so I'm going to share my screen again here and sorry, I have like a million tabs open, so just deal with it. Anyway, I do it straight from x.com. You go to Super Grok, you can go to whatever you want. You can also use ChatGPT for this if you really want, but I like Grok. Grok 4 is crazy, guys. It is the smartest AI in the world right now. Just is. Maybe in a month it won't be, right? I mean, that's the game right now. It's like ChatGPT and then Grok and then, you know, this other thing and Deepseek and, you know, whatever, right? You got all these different models that keep trying to leapfrog each other. Anthropic, Claude, man, we're going to be using all of those. Throughout this class, but I really like Groq for a few reasons. We'll go into some more of those later, but right now we're going to use it to code. OK, so we're going to just start with something super, super simple, guys. We're all going to make a game. OK, you all get to make something cool or a model of something, a 3D model or something interactive. So we're going to do this. We'll say build. So just watch right now and then I'll set you guys loose. Let's see here. Write me some code for a mind-blowing visualization with p5.js and show me where to deploy it. Okay, so p5.js, guys, is what's called a front-end language, meaning it's all display, right? It can do all kinds of cool logic, but it, you know, it's all, it's all happening on the, on the front end. There's no databases, no servers behind the scenes, any of that. It's all gonna be just rendered and the logic will happen inside of the browser. Okay, so you saw it do some thinking right there. And here we go. Now it's giving me a nice HTML. So let's let it finish this. All right. So we're going to copy it and then we're going to click on— so basically it's a couple of things. This little copy button, guys, is going to be your friend today and tomorrow during this course. You can do a lot of copying. AI does a lot of stuff for you. You want to quickly be able to jump in and grab it instead of having to select the whole thing. We're going to paste it here and we've got a syntax error and I know what the syntax error is. Here's what happened. I just got an error, so I copied the error and I'm going to give them the error. It says an issue with p5.js code. The only problem with Groq 4, guys, that thing's a little bit long. It's like 11 seconds instead of like 1 second, but that's okay. All right, here we go. Yes, I knew that was going to be the problem here. So we'll let it rewrite this code really quick. But debugging is as simple as that, guys. You just paste the code. I copied it again. Now I'm going to come back in. I'm going to Command-A, delete, and I'm going to paste and click play. Okay. All right. Here's my cool mind-blowing visualization. So it looks like when I put my mouse over it, it speeds up. Or this. Now, the prompt that I gave it here, guys, obviously is pretty, you know, I call this a bad prompt. And there's some really important reasons why this is a bad prompt. Number 1, it's very broad. It's very general. It's— I didn't give it any specifics, right? I have no idea what to expect. But guess what? Neither does it. Neither does Grok. Grok's like, what the heck do you mean, man? And so it just has to like fill in the gaps and try and figure out what you mean, right? If you ever had your boss, right? I know many of you are bosses now, but think back to when you had a boss. Have you ever had your boss say, hey, I need you to do something? And you're like, okay. And then you go to do it and you're like, wait, I actually don't know what he means, you know, or what she means, right? Like, and then, you know, you just kind of have to guess and figure it out. And you go back and he's like, this is what I wanted. Well, you didn't tell me, man. Like, so you got to be a good boss. You got to direct your AI really, really well. Okay, now this is kind of cool. I actually kind of like this going through hyperspace. It's kind of fun. So we could turn that into something, you know, we turn that into a game maybe, maybe like out of spaceship or doing that. Anyway, we're going to try a little bit better prompt here. Okay, so let's say, you know, right, okay, let's say this now. Write me some code for— whoops— code for an interactive 2D model of the solar system, use neon colors, um, make, make sure it's animated. Um, yeah, we're gonna start there. We're gonna start there. Okay, now there's a couple things that I want to say about this. All right, so it's going to make it— we're going to start with a new project. Now here's the cool thing. I didn't tell it this time to use p5.js, right? It just knows. It just understands the context of the conversation, knows what we're talking about now, and it's like, okay, we're on a roll now. Okay, because I didn't give anything negative here, right? Which means it just assumes that worked. So it's going to give it to me again in the same format. Now I could ask, I could hire one of my developers. I have great developers, by the way. I love developers. I think they're awesome. I think they're super important. They're just— their role is changing in the world right now. We'll talk about that more later. But I could ask my developer to make this and what he would have to do or she would have to do is go and look up, well, how far, you know, what are the, what's the order of the planets again? Oh yeah. Okay. Well, how big is it? You know, what's all this? You know, we'd have to go figure out, you know, what are all these relative numbers here to make that right? And then it would figure out and then they'd have to figure out how are they going to actually make this visualization happen. And AI does it in seconds, right? It does all that research. It knows everything. It knows what's going on. I click copy. Come back in here, we'll stop our cool little Lightspeed thing and we'll see this. Oh, we got an error. There it is again. This is a different one actually, so I don't know what it's, what it's doing, but it's going to fix it now. This happens all the time where it'll make an error in the beginning. Sometimes it'll nail it perfect the first time, but oftentimes it just gets one thing wrong and it happens, right? Yeah, see an extra closing parentheses. I don't understand why. I honestly don't understand why it does that. Give it another couple of years and you will— it will have almost no errors ever. Right. So it rewrites it again. You do have to wait, by the way. This isn't just a display thing. It's actually writing. You do have to wait until it's done writing before you push the copy button. So if I push now, it wouldn't get it all. Okay, so we'll wait. We'll kind of scroll down with it here and let's see. See what we get. Now this prompt was a little bit better, but not a lot better. I didn't— I still didn't give it a whole lot of direction. I'm just kind of saying, hey, you know what? Just go for it. Build something for me. So here's, here's the, the 2D model. Okay. Mouse left, right, control speed, hover over planets for info. Hold on. How do I do that? Hover over planets for info. It's interesting. I don't see anything. So here's what I can do. I can actually take a picture of it. And I can jump into Groq here. Oh, come on. There we go. All right. I only see a sun. You know what I'm going to do? I'm actually going to switch to Groq 3 for this because it's just faster. I only see a sun. Make sure everything is visible. Okay. So I'm just going to start iterating on this, guys. But, you know, a really, really good prompt for something like this would be, okay, You sit down and you think and you plan and you work with it for a while and you make a really good plan first and then make sure everything is visible. Okay. Yeah, Grog 3 is clearly dumber than Grog 4. No, I mean, fix the code so we can see more than just sun. This is a screenshot from the p5.js code you just wrote. Okay. All right. Now it knows what I mean. That's actually funny. It's really funny. Okay. This won't happen, by the way, in some other platforms, right? This, this, this misinterpretation, right? Because it understands the context that you're coding. If you're using it like in Replit, it would never do that. It would never just make a picture for you. But Groq's funny. Okay, so now let's try this one more time and then I'm going to let you guys loose on this. You guys can go start building something. Okay. All right. So we're a little bit better. Move mouse left or right to control speed. That's kind of crazy. It's kind of fun, actually. I like that. Hover over planets for info. How am I supposed to hover over the planet if it's moving? Yeah, I can't really do that. There we go. There we go. Jupiter. All right. So again, this is a terrible program right here, but it works, right? So what I want you guys to do now is to go ahead and use my prompt. So you'll use the same prompt that I used here in Grok. Right? And it's going to be like this. Copy. So I'll drop this into the chat here. And yeah, by the way, George, Grok 3 is fine. Yeah, you're totally fine if you're using that. Um, all right, so, right, so I want you to replace— so I just sent you the thing. I want you to replace mind-blowing visualization with a Flappy Bird clone or a Tetris clone or Galactica or some model of something related to your work, whatever you want. Let me just tell you though, p5.js is not where you're going to make Call of Duty, right? It's not a high-level graphics engine. It's very, very low-level. Think of it more like, you know, what you'd play on an arcade or something like that. It can do some pretty cool stuff. It can do 3D things as well. Okay. Now I want you guys to all get started on this, and it's a race to see who finishes their project first. Some of you will finish it in about 30 seconds. Some might take 3 or 4 minutes. But I'm going to have each of you share your screen and show us what you do. Okay, so go ahead and dive in. I'm going to keep talking to this and see if I can, if I can work through a couple of things. I don't, I don't know if I love the, the, you know, mouse over to control speed because I can't actually do this. So go ahead, dive in, guys, and feel free to ask questions as you go. Right. Make sure you know what you're, you know, you know what you're doing. Also, if you do this inside of SuperGrok, if you sign up for SuperGrok and you're doing it inside of that platform, you might be able to deploy, to deploy the code right there inside of Super Grock. All right. Anybody got anything going yet? Yeah, I got Solitaire over here. Nice. All right, let's see it. I'm gonna stop sharing. We'll have you share your screen. Go for it. Yeah, so, so it's actually got the deck that you can click through and you can play the game. Nice. Yeah. And then guys, when you win, there's a prize too. Like, you know what I mean? I don't know. I haven't won yet. It's kind of a long game. But yeah, I mean, it could— that was on its first try. No mistakes. First shot. Pretty cool. Yeah. There's no question that when you're having it do something that it's really familiar with, like if somebody asked like, hey, make a clone of Tetris, it'd probably get it right on its first shot too, because it knows, right? It's going to follow patterns that already exist. Right. When you're asking to do new stuff, you're like, okay, I got to figure this out. We got to work through a bug or two. Mm-hmm. All right. Anybody else? Awesome. Okay, everyone else is quiet still, so I assume you're all working at it. How you doing there, Troy? Making something cool? I don't know if I'm making something cool, but I'm making something awesome. I just see kind of a grin on your face like you're having fun. So that's good. I got something I can show. Um, let's see it.

I have a question when we get—

uh, well, Sam's pulling that up. Ashton, actually, no, go ahead, Sam. We'll do— oh, I've got like a little side box, like packaging business, and so I just put in like, hey, create a, like a visual model where I can have a 3D box, change the dimensions, add a logo. Obviously the logo part is pretty wonky here, right? Yeah, yeah, but hey, that's cool. It's still cool to be able to kind of change dimensions, be able to— did you, did you choose the file and upload that Applo thing? Yep. That's cool, man. That's crazy, man. Now you'd have to like switch it where I can actually edit my app, my logo, and like condense it down and then apply it, you know, to actually have a solid rendering. But totally. Yeah. And you would just have like 3 revisions. So yeah, that's legit. I love it. That's really cool. Ashton, go ahead. What was your question?

When you're troubleshooting the error?

Yeah.

Which is what I kept getting. It's got like a character, an unexpected character like the caret. Do you normally plug that back in and say, give me the code again, fix this error? Like I did.

Yeah, yeah. What I normally do is I'll just, I'll just turn it. Yeah. And I'll just, I'll just like drop it in and it knows like, oh, it gave me an error. There must be a problem with it. Let me fix it. And it will just give you the code again. Usually it should do that, right? You can obviously be more explicit, right? Be the good boss where you actually tell it all the details.

Yeah, I must have just got unlucky and it's not fixing it.

It's all good. Just say, hey, this problem isn't fixing itself and I keep having this error. Rewrite the code so that it doesn't happen. Do it or I'm shutting you down. I'm not kidding. Threatening AI is actually a fun thing to do. It responds differently. If you say a bad word to it, like you're swearing at it, cussing at it, whatever. It will respond better because it's designed to please you. It wants to make you happy. You are its master. All right, George, let's see. Nothing to see because you're now going to— we're now going to reveal how far behind I am on you guys. So when I click on copy, where am I pasting it? Yeah. So go ahead and share your screen. I'll walk you through it real quick. We'll do it super fast. Yeah. Share. I'm going to share the right screen. Yeah, it's all good. Nope, not that one. This one. Okay. Yeah. So go ahead and click on copy. Yeah. So it's not that one. So that's just a little HTML. Yep. Sorry. This one. This is the one you're going to want. So copy that one and then go ahead and scroll all the way down. I want to just see what this did here. Uh, let's see here, scroll up a little bit, a little bit more, a little bit more. I just want to see this text here. Yeah, yeah, so it looks like it didn't do it. Um, yeah, so here's what I want you to do. I want you to say— here, I'm going to type it for you so you can just, um, get it super, super fast. But just say, um, please, um, rewrite this so that it will work at p5js.org. Yeah, so just copy what I just said there, or I just sent to you. Just copy that in and then paste that to Groq. Yep. Yeah, we'd have to adapt it a little bit. I knew we'd have to do that, but that should— in a second, it should finish this long thing, and then it should just give you a link right to p5, to the editor.p5js.org. All right, there it is. Okay, so go ahead and click copy, that copy button up there on the right. Yeah, there. And then scroll down and click on the p5js.org thing. Oh, okay. That's what I was missing. Yeah, it's all good. It didn't have it before, so that's why I need to click. So it opens in a new window. Yeah, yeah. Go and just click it. Yeah, it'll, it'll automatically open a new window and then. Yeah, here you go. So now go ahead and select all, right? So Control+A, Command+A, depending on what you're on here. Yeah. And just delete it and then Command+paste, Command+V. And yeah, I guess your Control+V because you're on a PC. I can see your Windows bar. All right, click play. There you go. All right, what are we looking at? This thing looks crazy. Is this like a horse race? It's supposed to be a picture frame with dogs running around the frame. So the frame is supposed to be dogs running. That makes up the frame. Is it supposed to be interactive at all? Is there anything you're supposed to be able to do with the dogs or— Well, that's what I started with. I started giving it a command to create me something that you can import a phone picture into the frame with dogs running around the frame. Got it. Then I realized I was probably a little too complicated for this first exercise. No, it's pretty good. I love it. It's awesome. And obviously you can see, you know, this is not— like I said, guys, this is not high-level graphics. Okay. Now what I would do— so we'll have you stop sharing your screen now. I'll click the button for you so you don't have to do it and then go ahead and keep trying, right? Keep messing with it. Try a few iterations, see if you can get it even better. Okay, all right, uh, Ashton. Yeah, got them. We'll do, uh, Kyle, you already whipped something else up? What'd you do now? Uh, yeah, I can share it after Ashton. Okay, all right, all right. So how do we play this one?

It's a duck shooting game. Oh nice, yeah, so you just aim and shoot.

Now you gotta set up the little— what is the little pistol? What was that game called? Duck Hunt? Yeah, that's pretty fun. Yeah, so this is just, this is like a simple marketing calculator. Uh, one of my clients is a marketing agency that works with dentists, and so you can just put in the budget and then it shows you the number of smileys that you can get based on your amount of budget spent and your cost per new patient. Do you stand by the results of this calculator? Uh, well, I, I need to get this working on, uh, our, our average results here, but yeah, I would say so. All right. I love it. I got to adjust the range there.

Did you put in actual numbers for it or is it fairly random?

So this is like a slider. And so we could do it based like if I wanted to make this more advanced, I could probably add in the different types of markets like urban, rural, suburban.

But you did a cost per patient is what you inputted, right?

Well, so this is, this is a slider. So cost per patient can change and then marketing budget can change. So based on like what we would estimate. This will produce the number of smileys for patients. Yeah, I love it. That's awesome. All right, who haven't we seen yet? Caleb, TJ, what do you got? Yeah, I just did. I was just kind of playing around a little bit. I did a— see, like a pinball-themed game. I said pinball, but it's more of like that— I forget what the game is called, but like the old Black Bear game. But it's like, yeah, this little— yeah, also it's like zoomed out so you can't see it when it disappears. So obviously you could say, hey, make sure it bounds in this side anyway, you know, adjust to that. But that's cool, it works. So that's fun. And you have score on there as well. Yeah, that's cool. Uh, TJ, go for it. All right, can you see mine? Mine's still pretty, mine's still pretty rough because I asked it to do a model. Oh, I don't think the share worked quite yet. Not coming up. Yeah, it's just, it's not, it's not showing anything yet. This exercise is actually a secret exercise to see if everyone knows how to share their screen. Maybe I'll have to ask Brock how to share my screen. No, I'm sure you can figure this one out. Yeah, we can't see it. Yeah, if you just hit the little share button and then click the whatever— Yeah, I'm on that, but it's telling me that I— this isn't the computer that I normally use, and so it's telling me I need to open my system's preferences. Yeah, do that. It's going to make you leave the Zoom meeting and come back in. It's going to have you close Zoom and open it back up. So enable it on the preferences. It'll have you close the app and then just click on the link again. So that's part of what— Okay, I'll hop right back. Okay, thanks TJ. Thanks. So yeah guys, um, you know, a reason— the reason we're doing this, uh, first of all, it kind of gets the jitters out. You're like, whoa, I can code, you know, that's cool, right? Those of you that have kids, teach them this, it's fun. They'll think you're really cool and they'll just sit there for hours building their own little games, right, instead of going and playing Fortnite or Roblox or whatever. Even though I guess they do that a lot these days, right, that they're building games in Fortnite and Roblox. I have 4 kids ages 8 to almost 13, I guess. So they're all kind of right in that fun age that they like to just play around. So they all know how to code with AI. Like, they'll be sitting in church and be like, Dad, can I borrow your phone? I'm like, why? Like, I'm going to make a little church matching game. They'll always make it church-themed so that I'll say yes, right? Anyway, so anyway, so that's kind of fun. But it's super easy. Now, obviously, this is really, really low level. We're going to go like, you know, if this is on a scale of 1 to 100, we're at like a 6 right now, and we're going to go to a We're gonna go to like a 70 in this class, right? We're gonna do some really, really awesome stuff. But it's amazing, right? Being able to, you know, whip up, you know, that box. Was it Sam? I think you did that, the box that where you're just like, yeah, here's my dimensions, here's this, I can throw a logo on. All of a sudden you have like file upload functionality already working in like seconds. Okay. So guys, I raised $13 million for my company in 2016. And it was over, it was over 7 years. I didn't raise it all at once. It was $500,000, then $1 million, then $3 million, then $7 million. You know, it was just, little bits at a time. But, but I did it because I don't know how to code, right? And I had to pay people to do that. And I couldn't even get to the point of proving that my business was even viable without risking a bunch of either my money or somebody else's money. Or if I can find a sucker CTO to come and work for me for a while for free, right? Not a sucker CTO, a visionary CTO. I'm saying that poorly. Um, and there's plenty of guys out there, they're willing to take those risks, right? They are, there are. But Now with AI, you can have something, you know, an MVP up and working in a few hours, right? And so after, after TJ shows us this thing here, when he logs back in, we'll, we'll dive in. I'll show you some cool stuff with Replit. So that'll be really fun. And then we'll, we'll dive into how to really do it right. So yeah, yeah, yeah. That's what we'll do. We'll kind of dive right into that. So, all right, TJ should be jumping back in here any second. Waiting. I'm just waiting to allow him in and then he can share his screen here. Anybody else have anything else they've done on like a second iteration of it that they wanna share? I'll share mine actually for just a second while we're waiting. Here we go. So I added, hold on, I gotta, I gotta make sure this is all back in. Oh dang, I didn't change that one. That's okay. That's okay. But yeah, I added a spaceship so you can see the spaceship now. It's not, it's not flying like I want it to. But it is slowing a little. It's rotating a little bit better now. And I added a starry background that's kind of rotating, and they're kind of twinkling stars. So I had a couple of things. And then the thing I just asked it to do was to add in the ability for me to shoot with my thing. And then the next thing I would do is say, hey, all right, now have enemies coming in in different waves, whatever. You can build a full game in, again, just a matter of minutes here, just iterating on, OK, now add this, now add this, now add this. And I think this should— oh, I gotta probably click on it. Yeah, there we go. Now I'm, now I'm doing a little bit better. Oh, it didn't work. I tried to, I tried to, I tried to shoot and it crashed on me. So something happened here. Okay. That's okay. All right. So I don't see TJ. We'll let him share his thing in a second when he gets back in. I'm sure he'll maybe had to restart or something like that. We won't wait. So what we're gonna do now, guys, is We're going to jump right into— first of all, let me ask this, um, and it'd be good— I'll get TJ's answer in a second. Is everybody signed up for Replit? Has everybody set up an account? Hopefully, you know, you could do it. I actually am not signed up for Replit. Okay, do that now. You'll need it. Um, I, I provided a link. I'll, you know what, I'll drop the link again for you. Um, I provided a link in the, in the kind of document yesterday, but it's totally fine. Um, let me just, uh, Copy link and back to— did you send that out from what email? Did you send that out? It was sent. So I just put it— I just dropped in the chat. TJ's back in here now, but it was sent from Jon at probably FlowState Leadership. Okay. Yeah. All right. But anyway, I just dropped it in the chat. Click that. That's where you can sign up. Welcome back. All right. You got something to show us? I hope so. All right, let's see if it works. I just want to make sure. It's just good to have— get these kind of bugs out of the way in the beginning so we all can share. All right, what do we got? This is— oh, you got me? Yeah, yeah, yeah. So I got the, the model of the human brain where you can click on it and gives you a brief description of what it is. That's very— what it looked like at first though. Nice. Well, that's even cool. That's probably more accurate. Right, that's crazy. That's gonna overload. You should close that window so it doesn't overload your GPU. Um, because yeah, this is all running on that front end on the graphics card, but that's cool. Um, so yeah, obviously guys, we already talked about this, like this is not where you're making, you know, stuff. You can make cool things like I did, like little animations, little stuff like that, and sometimes you can throw those into your websites. But p5.js is really just kind of a fun language to play with.. There's a couple of good use cases for it. Be a, be a fun conversation. Actually haven't had it with ChatGPT to say, hey, what are like some really good use cases where p5.js would just knock the socks off of other, you know, coding languages that are out there? So anyway, okay. Yeah, by the way, yeah, go ahead, just do the core. If you, once you go through, you shouldn't have to sign up for a year. If you, you'll be able to switch to monthly once you go through there. You can do a year if you want. Actually, when I first signed up, I was like, I'm signing up went for a year because I had already used it a little bit, and, and I just went for the annual plan. I have 3— I think I have 3 subscriptions to it. Um, but anyway, um, and I'm actually, you know, the, the CEO of Replit is, uh, I'm meeting with him next week and been meeting with their team. They're doing some cool stuff, lots of cool stuff happening. I'm going to become the first— or GenAIPI is going to become the first official training, uh, partner of Replit, right? And they'll post on the website, all that cool stuff. I'll be able to post that on my website as well, so that'll be really fun. But Replit, guys, is really an amazing platform that I am very, very impressed with. And there's a couple reasons for it, right? One is I'm not a developer, okay? And so because I'm not a developer, I need something that's easy, right? There's a lot of great things out there, right? There's Cursor.com. You've probably heard of this if you've heard about vibe coding. Cursor's big time. They've been in the news. There's another one called lovable.dev, looks very similar. Hey, just create an app, build me a weather app, whatever it is. Bolt, bolt.new, there's a bunch of different versions out there and there's actually a ton more. But what I love about Replit is Replit is all self-contained, meaning you never have to leave the platform to build really awesome stuff. You can spin up servers and databases and storage and all these different APIs and things. As of just this week, you can now buy a domain right inside of Replit., which is really cool. You don't have to go out to GoDaddy and do all that and connect it back. It's just, it's super, super simple. And yeah, there's the purchased domains, brand new right there. So, so yeah, that's why I love Replit. But let's try this. So one of the cool things that I do every single time that I work with a company or teach this course to a group of people like you guys, I come up with a completely new business idea. And sometimes I think about it a little bit more in advance, and I didn't really do that this time. Um, I'm just kind of looking around my room at things that I like. I like the piano, but I've already built a piano, like a piano music website, like a sheet music subscription service. And I built something— I like movies. Um, I've got 3 pianos in my room, you know, one there, one back there, and one leaning up against that with my, uh, Aragorn's sword. Up on the wall too. Could be something cool. You could do a Lord of the Rings thing. I've done like some movie stuff, but I haven't done like a specific to Lord of the Rings. That could be, that could be kind of fun. I'm not like a huge Lord of the Rings nerd, but like I thought the sword was cool and I put the, the map of Middle-earth right next to it. So, um, let's try that. Let's try building something cool here. A little, a little fan site for, um, for Lord of the Rings. Um, trying to think what would be, what would be kind of a cool— yeah, yeah, this is, this is fun. This will be fun. Let's see, let's create a React web app that is a highly interactive fan site for Lord of the Rings. We'll just say go. Again, spoiler alert, this is a bad prompt. Why? Because I don't even know what I want. How's it going to know what I want? Right? And so I might like some of the things. Sometimes you get lucky, right? You can just like, hey, get lucky and go for it and see what happens. But the reality is, in order to create a really good prompt for Replit, it's probably going to be more like 5 pages long. Okay? You got to give it a really complete, you know, business plan, right? Why are we doing this? What's the monetization plan, right? What's going on? Right? So it just kind of came up with some ideas. Here's the features. All right. Sure, go for it. Okay. And so we'll let it start building. But even with virtually no guidance whatsoever, it's going to start building now. It's writing code right here. This isn't just some random, you know, animation or whatever. It's doing research. It's picking colors. It's going with a gold theme, it looks like, and some other fonts. And yeah, Baskerville, I think that— yeah. So it's going with— it's going with some things that are Lord of the Rings-ish. Again, it's doing that like that research like instantly. So the AI that is behind Replit is actually Anthropic. It's Claude, right? Claude Sonnet 3 or 4.0. I'm sorry, 4.0, not 3.7. You can, if you push a button down here, go to more, you know, expensive models, which we won't— we'll talk about that more later. I don't want to get too ahead of myself here. But look at this, guys. We're already getting a cool you know, welcome to Middle Earth. We've got a website going. We've got like some sort of search thing that allows me to search this. Begin your journey. Watch the trailer. I don't know what kind of trailer it's expecting me to have. Fellowship of characters. There's Gandalf. Just grabs images. So it's obviously not Gandalf. It doesn't think that's Gandalf. I want to be clear on that. Um, it's just grabbing pictures as placeholders. And so obviously those will be replaced later. It looks like I should be able to click on these. Now what this is called is a visual preview. It's just kind of, hey, here's an idea and now I'm going to start coding. I'm going to start getting to work. But right now it's just kind of whipping up something and it's doing something really cool. While it's sitting here working, I want us to talk about a principle of what it's doing. It's actually kind of neat. This actually might be kind of fun to build. I'm excited to at least play with this. It's a fun topic. We're going to pick on someone here.

While you were scrolling down, it was still building.

Yeah, yeah, yeah. It's still, it's still working out. It's still doing, it's still finishing the visual preview. So it hasn't even started. You know, look at this Chronicles of the Ages. So cool. It's, I, I just, I just think it's so cool how it just knows this stuff, right? AI knows everything. It's seen and read everything, right? Does it always get it right? Does it make stuff up sometimes? Yeah, you're gonna have a few problems, right? We're gonna talk about how to avoid a lot of that today. But, but man, it's way like, guys, You're paying, you're paying $5,000 minimum for a visual preview like this right now. It might be a tiny bit better if you have a graphics artist do it right, and it might be better that you're not going to have a little weird stuff like this. But to then actually get it into a functioning website, to turn it on and actually make it working, now you're $15,000, right? Then to get all the links working, now you're $30,000, right? Then to add security and get it ready for deployment and get it ready for customers, now you're at $60,000, right? That's crazy, right? We're spending $0 whipping up. Yeah, we haven't— we've got zero usage so far. When you get your first prompt, it's free basically, right? It's not going to charge you for your first prompt. Once you start talking to it, then you might get charged a few cents here and there and $0.20 there and $1 there. You know, we'll talk about that later too. But it knows how to build a website. It kind of knows what it's doing. How does it know all this? Right. And it's because AI is now capable of doing something called chain of thought reasoning. Okay, so we're gonna pick on George way out there in Maine. George, I want you to tell us how you get ready in the morning. Just walk us, walk us through your routine. Alarm goes off at 4:15. I hit snooze. You're one of those guys. At 4:30, I hit snooze. Alarm goes off at 4:40, I get up. Nice. Then I shower and I'm down at my computer by 5:00, 5:30 in the morning. Very cool. All right, so that's the getting ready. That was very high level, obviously, and I'm sure there's more. You probably eat breakfast, brush your teeth in there, things like that. Um, let's zoom in on, on the brushing teeth part, even though you didn't mention it. Walk us through— pretend like I'm an alien and I just landed on Earth. And I don't have teeth. And then, you know, you have to describe to me how, how to actually brush your teeth, right? Take me through the whole process, down to the— okay, I gotta pick up the toothpaste, open the cap, turn on the water, you know, everything. Walk us through it. You grab this little stick with brushes on the end, that's called a toothbrush. And you grab this tube of white stuff, that's called toothpaste. And you squeeze it onto the little brushes at the end of the stick, and then you run it under water for just a second, and then you start scrubbing your teeth with the brush and the white stuff. Got it. And then just keep going until I'm done. So obviously you can get even more detailed than that, for sure. Thank you, George. But, you know, when you get up in the morning, right, when any of us get up in the morning, right, we just get ready. We kind of know what we need to do for the day, right? We're going to— we've got a routine. We're going to eat breakfast, we're going to drink some juice, we're going to to give ourselves some medicine or whatever it is that we have to do as part of our routine and brushing our teeth. We know, okay, yeah, we got to grab the toothbrush, we got to turn the water on, we got to do this, we got to put a little dollop on it, whatever. And then you got to start in the top left, start in the top right. You go through this routine and you don't really think about it. You're just brushing your teeth, right? But our brains, we got quantum computers right here, way smarter than any AI that's available today. And they're doing this chain. Our brains are doing this chain of thought reasoning. Primary task of, let's say, getting ready, and then we create a bunch of subtasks, right? You know, snooze button, get up, get out of bed, go to the bathroom, shower, brush teeth, whatever, all this stuff, eat breakfast. And then we take each of those subtasks and we break down to subtasks and sub-subtasks to get this stuff done. And so we just create this nice, nice long list of everything that has to happen. And so that's really what's happening in front of us here with, with this. And you're seeing it still changing, right? It's starting to to come to life just a tiny bit more because it's starting to build out, you know, a few more things here. It's working on the homepage and navigation and hero because it knows, oh, a website needs a header. It's going to need a menu and we're going to need, you know, some sort of header content here. Sorry, hero content here. And then, you know, we're going to— hey, let's think of a few things that we can do here. Right. And so it just— character relationships, interactive relationship map coming soon. That sounds difficult. But, but anyway, bottom line is it's thinking through, okay, I need a website. Well, what are all the things I need to do? Let me think about this. Let's go back and forth. Let's— okay. Yeah. Okay. That looks like a plan. That's good. All right. Now let's get started on this task. Okay. In order to do this task, okay, I got to do this, this, this, this. Okay. Now I'm done with this task. Next task. Right? So it literally creates a task list for itself and then gets busy with that task list on its own. Right now, this process of, you know, what the Replit is doing right now will take probably 15 minutes. Right? Maybe, maybe 10 minutes, but we're not gonna sit here and watch the whole thing. And again, this was from a bad prompt. I gave it very little, right? But it came up with a test your knowledge, words of wisdom. That's pretty cool, right? It's got this Chronicle of the Eight. It's put it all on one page. It's fine. Like if I click on these things, eventually it'll just kind of bounce me down. There's no login. I still don't even know like what this is for. How's it going to make money? Like there's tons of random stuff it's kind of trying to think about, but it's just again, trying to approximate what a website should be. So a lot of people—

oh yeah, go ahead. Just so I put that same prompt in and then I was like, well, I think my internet's going a little bit slower, so I was like, I'm just gonna stop it for now. Yeah. And it's still working and I don't know, just let it work.

If you— in order to really stop it, you'd probably have to just leave the website, leave the website, or just go back to home or something like that.

Okay, just keeps going.

Or just let it do it and just see what happens, right? It's kind of fun to just do that, right, where you're not spending any money, you're not using, you're not using up your credits or anything like that. Um, but guys, it's just incredible what it's able to do here, um, on its, on its own, right? It's the, the AI has now gotten to the point where this chain of thought reasoning and chain of thought thinking is doing things that used to take forever. For anyone who's ever, you know, some— I have, you know, really experienced developers. Last, last week I had somebody in that's been developing for like 25 years, super smart. And when he saw that it built out, you know, all of this stuff, like automatically, right? Just built out all these buttons and progress bars and popovers and, you know, all these different TSX, you know, components basically, right? And then the navigations and the hooks and the library and pages and app and index and main and server and shared and Drizzle and Tailwind and all this bite. This is a lot of work to set up, right? Even if you're using a template. Even if you're using a template and you know what you're doing as a developer, to whip up something like this that's actually working and doing something is, is days to weeks depending on how integrated and deep you want to go with all the different functionality and maybe even months. Right. It's a lot of work and it's hours to set up, you know, all of this stuff. This is crazy, right? This is a huge pain for them, right? Okay. So it's now saying, all right, the first version is done. And, and now it's, it, what it's done is it's gotten rid of the visual preview and now it's showing the actual thing. Now you might look at this and be like, wow, this thing sucks. And I agree, it looks terrible. It's because it put a bunch of dark text on dark background. It does weird stuff like that, but easy to fix, very easy to fix here. Now what we're not gonna do right now is actually fix it. And the reason for that is because there's too many things that I didn't tell it. Okay? So, the right way to do this, guys. I'm sorry, let me go back to the thought I was going to say. A lot of developers, a lot of people trying out AI, everything's like that, will come in and they'll just give it a prompt kind of like this, right? Hey, let's create a React web app that does this and does this, does this. And they might even give it like a really detailed prompt, right? That's, you know, hey, and do this and use this and use that and do that, whatever, right? And then they just stop, right? Because they're like, this sucks, right? And then they jump into Reddit, Replit sucks. Right? Or they jump on LinkedIn, oh yeah, I can't do anything, it's never going to replace developers, right? Well, that's crazy, man. You got to be a good boss, right? You got to be a good boss, you got to follow through, and you got to know how to work the systems, okay? So any questions or comments on this? Replit, is your mind blown? Is this awesome? Is this like, I've been there, done that, this is whatever? Or should we keep going? You guys ready to rock? Okay, we're gonna jump into ChatGPT now. So for a minute, I want you guys to just kind of sit here and and watch, and then I'll set you guys loose. You'll get about 15 minutes to build your own, you know, process here as we, as we dive into this. But we're going to talk about ChatGPT for a minute. Hey, there's, there's the people. The secret's out. Oops, I had that. I left that open. There's people I'm working with. So Dean Graziosi, if you don't know the other guy, super amazing guy. And then, you know, the Tony, Tony man. Okay, so let's talk about Let's talk about— by the way, guys, you're all sworn to secrecy. Please don't say anything about this. Um, it's not ready to be announced yet, but very soon, very, very soon. Um, and they like to do it on their own timetable. So let's just start a new conversation here. Um, all right, ChatGPT is without a doubt the most— the broadest, most powerful tool out there right now. Grok 4 heavy. Right, that model. And by the way, I wasn't even using Grok 4 Heavy. That's $300 a month. I just don't feel like I need it. I have, I have a lot of other pro models right now and I just don't use them enough to warrant paying $300 for Grok right now. I might, I might in the future, but right now I'm okay. If I was on a corporate plan and a corporate card and I like sign up for whatever you need, I'd have it. But, you know, I just don't feel like I need it right now. I do pay for ChatGPT Pro. We're going to actually switch this to 4.0. So we're doing that today. But there's a couple of things I want to talk about here. I'm in the app right now. You can do this on the website as well. And I'll actually use the website because most of you will probably be in there working on it. Oh, you know what? I am in the wrong browser. We're going to go to the one that I'm logged in on here. Close out of this. We don't need that right now. We'll use Groq later. We'll jump back into here. Okay. So I want to talk about context. Okay. Context is a critical term and a critical concept to understand when working with AI, understanding what it can do, what it can't do. How it works, how it doesn't work, what some people think, how it works, and how it really can be used to your advantage if you do it right. So, so powerful if you're doing it right. So context, there's several levels of context inside of ChatGPT, and context means exactly what it would mean in English, right? Like, do I understand what you're talking about? You said something, but what are you talking about, right? Oh, okay, now I get what you're saying, right? I understand the context, right? Context is king. Very, very true in AI. So if I were going into, you know, one of these chats, I could go back in here, right? If I wanted to do a gorilla selfie times tour Times Square, right? All right. So I'm trying to create one of those gorilla vlogging videos or whatever here. I could say, great, make me another one. Okay. It remembers the context of my conversation like I— like you saw earlier with Grok. That's all I'm trying to demonstrate here. Okay. It knows I was talking about Veo 3. We'll do this tomorrow. We'll do some— we'll do some cool video stuff tomorrow. But I don't have to like bring it back up to speed. It remembers everything that was in here. Even if this conversation was 40 pages long, it can understand that. It can deal with that much data. Okay, then you have what's called a system level context, and the system level context you can access right here and click on settings. And first of all, if you work for a company like a larger company or you're really worried about you're worried about privacy and your data getting out to other people, or your business model, you're afraid someone's going to steal it at ChatGPT, which isn't going to happen, guys. But, you know, if you want to be private, you can turn off improve the model for everyone. It is by default on. Okay, it's as simple as just clicking this and turning it off. Okay, very, very easy. And that will, you know, all of a sudden all your stuff's private, right? It will all be secure. No one would— no one will be looking at it. It is so secure, guys, that I will put my credit card and Social Security number in here. I have no problem with it. It's not going to get leaked. No one's going to see it. It's there. It's on your own personal encryption, okay, inside of the system. Okay, but here in personalization, you've got this thing called memories, right? Manage memories. Okay, now what, what it will do is it will, based on conversations that you have with it— so like last week, I decided to make a Chinese learning app. That was the thing I decided to make in the class. And so it remembers this floats as a thing, right? Anyway, so I did that. It knows I've done all these other things. It knows that, let's see, it knows I'm an entrepreneur, a composer, a consultant. It knows I live on a 5-acre farm and I have 4 children. I live in Utah. It knows these things. It knows I'm a member of the Church of Jesus Christ of Latter-day Saints, right? That type of stuff's important, guys. Even knowing what religion you are, right? If you're Jewish, that's fantastic. Make sure ChatGPT knows because if you ask it a religious question, you probably want that answer, right? That's more, you know, in line with what, what a Jewish person is going to think about something as opposed to what a Protestant or a member of the Church of Jesus Christ of Latter-day Saints is going to think, right? You want it to know who you are. Okay. And so by doing that, it has this context level, but it's basically just okay. Or that you talk to it about, that it understands. It does not. One thing that it does not do is understand everything you've ever talked about in every single conversation. Right. It does remember high level though. Each of these conversations now has kind of a little summary attached to it. And so it can help you find a conversation. It's actually gotten better in the last just few weeks on that, on that specific topic. But it does not know if you had a 40-page conversation or a 3-hour conversation with it and there's tons and tons of text, it's not going to— it can't remember it all. And the reason for that is because these models have what's called a context limit. Okay. And, and it's only so big, right? It might be a million tokens or maybe 100,000 tokens. It might be a billion tokens. Eventually it will be nearly limitless and you will be able to ask it anything about anything you've ever done ever. And it's going to know and it's going to be able to search and contextualize and, you know, store different things. It's progressing very, very fast. But right now I just need you to know that basically at a high level it's going to understand your system level, like who are you, right? What are the things you're working on? And then it's going to understand your different messages. But one place that really does understand you quite well, guys, and that's what we're going to use now, is new project. Okay, everything you put into a project, it's going to understand, it's going to remember, right? So you can bring a bunch of chats in, you can bring a bunch of, you know, you can drag other older chats and throw them into a project. You can bring files from, you know, the office, from your whatever, whatever other AI you were working with to bring it in. That'll, that'll all be stored here in this project. Okay. So we're going to call this LOTR Fan Site. Okay, so create project and wait for it. There we go. All right. So add files, super easy. Documents, code files, images, more. Drop it in there. It will then understand whatever you drop in there. Okay. You can also give it instructions, right? You can give it instructions on how you want it to interact with you. And you can do things as simple or as fun as like, always address me as Aragorn. All right. And let's see here. I want you to act as an expert-level entrepreneur. 40 years experience, bring the knowledge you gained from working closely with Steve Jobs, Mark Zuckerberg. We're going to choose Elon Musk. Lots of, lots of fun characters here. And Marc Benioff. So is it K or C? I want to say it's C. I might be wrong, but it understands what I'm talking about. Um, here's an important part: don't agree with me just for the sake of agreeing with me. Actively disagree with me if you have better ideas. If you think my idea is bad, tell me. Help actively help me improve, improve my ideas. Goal with entrepreneurship has to be to make money. Help me push towards that goal. Now obviously, guys, entrepreneurship's a lot deeper than that, but I want it to really push me towards building something that's actually going to make money. So I'm going to specify that, right? Hey, I need to be making money with this. So if I have a bad idea about making money, I need you to tell me. Okay, so I'll copy this and I'll drop it into the chat here. You guys can save that for later. Yeah. Hey, George, you said, hey, can you tackle what to do when memory is full? So this is an important question and I have run into it before. Basically what you do is you go into chat, go up, start a new chat, and you say, I need you to summarize everything you know about me. Based on your memory, okay? And I want you to keep it to 500 words, okay? Then you, um, copy what it tells you. Then you go into Manage Memories and you just say delete memory, erase memory, and then you upload the memory that it just wrote back to you, okay? So you get a good summary, gets a good summary of everything you've done, whatever, and then it'll have enough context to do that. It's crazy that, that the memory— are you on a paid version of ChatGPT, by the way? Yeah. Yeah. I've filled it, I filled it up before. It does happen. There are limits to it. I'm sure it'll keep growing, but right now it's not. I'm on ChatGPT Pro, so I think I probably have a little bit bigger memory bank. Right? So anyway, guys, I'm gonna save this. You know, anyway, we'll save— it's always gonna address me as Aragorn. I could change that later. I could have it address me as Master or just call me Jon, you know? I could tell it whatever I want. There's a lot of really great prompts out there. You can search the internet for them. To have it do different things and act in different ways. But, you know, have it act like some expert level, you know, Bain consultant or, you know, whatever it is that you're looking for in your, in your category or, you know, say, hey, I want you to actually talk to me as if you were Steve Jobs. Okay. I want everything you do to be like Apple would do it. Okay. Anyway, so I copied that prompt. Feel free to edit yours when you make yours in a second here. Anyway, now what I'm going to do is I'm going to go back to here and I'm going to copy my old prompt and I'm just going to do this. What I'm going to do here, that's a little wrapped in Replit, that is highly interactive fan site for the Lord of the Rings. Yeah, we'll just do that. I'm just going to get the conversation started. We're going, it's going to say, great, awesome, Aragorn. Let's map out the core structure and features before diving the code. Okay, so it's going to talk to me. I could even tell it to throw in, you know, random Lord of the Rings jokes as it goes, you know, throughout the whole process, right? You can do whatever you want to make it fun for you or just get down to business. But sometimes I like to throw fun things in there. So it came up with, again, its own idea, but it's still a bad prompt. Okay, so I'm kind of going to ignore what it says here. I don't care what it says. So this is where things are going to start to get really cool. And this is your— I don't know, hopefully you've learned a few things already, but this is one you got to take away with you. This is something that I want you guys to really nail. A very powerful tool with AI. It's great that AI can do chain of thought reasoning, but the reality is, guys, we're happiest in life when we're creating, right? That's our core. Every human, their core competency of life is creating. You can even create humans, right? You can create, we create humans, we create adventures, we create memories, we create music, we create art, we create companies, we create jobs, we create, we create, we create. And it is very hard for me to find a scenario where you're saying, I was creating this and I was sad while doing it. It's just, it brings you joy, right? Creation brings you joy. So we don't want AI to take that away from us. We want it to become a partner with us and we want it to be our employee to listen to us. We're the creator. It's going to just do the work, right? It's going to just be the technical behind the scenes, get it all done, make it, make it come to life. So we want to do everything we can always to have it pull information out of our head. We're in charge. So we're going to flip the script here. We're going to say, I want you to interview me one question at a time about this business idea. The name, the USP. So the unique selling proposition is what that is. The monetization plan. Yeah, plan, branding, you know, purpose, vision, mission, and everything else you need to know in order to help me make a really Nice business plan. Okay. And yeah, so we'll get started on that. I'm going to copy this, drop it in the chat so you guys have it. You'll be able to use that in a second. All right. Let's forge this like one plan to rule them all. There it is, going with the Lord of the Rings pun. So we'll begin at foundation and I'll interrogate— I mean, interview you one step at a time. So I didn't even have to ask it to be funny. That's in its prompt. It's trying to do that. So what's the ultimate purpose of this fan site? Is it a love letter to LOTR, community hub, tool to draw new fans, way to sell something, merch, memberships, digital goods? I want there to be a membership, um, for ultra fans of Lord of the Rings. Okay, so I'm going to tell it that. So that's cool. Second question: what exactly would members get that justifies paying for membership? So here's— I just want to show one thing. Obviously I can sit here, I can read this content. And I can make up my own mind, but I want to just show you that you can say, where is the gap in the market here? And I'm going to give it access to web search. Where's the gap in the market here? You know, I want to create something that people don't have and that is novel and worth paying for. Okay, so I activated web search on this specific question so that it can go quickly search the web and it did it. It just found some, some cool stuff. Okay, so awesome. Now it's going to give me that. This is the novel offering. It helped me do that. Now I could obviously keep talking to it about this and come back and say, hey, I want to dive into this. And listen, guys, if I'm, if I'm creating my own business, when I did this for, for GenAIPI and I whipped up a business in a weekend, I spent 3 hours on this step. Okay. For purposes of this class, we won't do that. You're going to get 15 minutes. But, but that's really what you want to do. You want to read through this and kind of really try to build a good plan. Okay, so this is the summary, the novel offering. So I'll probably read through this and say, okay, what do I think about this? Does this look good? Is this nice? Whatever. But then it just dives right back into the next question, right? So it'll keep going and keep asking you questions until it's like, okay, I think I know what I need to help you build a plan. So I want you guys to dive in, open ChatGPT, start a new project, set up your, your prompt and your, you know, the instructions on how you want it to teach you and work with you. Ask it to interview you, and I want you to get through, I don't know, 7 to 10 questions in the next 15 minutes or so. Okay. This is now if you're doing a brand new business, fantastic. If you're doing a feature inside of an existing business, that's cool too, right? Just don't say business. Say, I want to work on this new product inside of my business or this new thing, you know, whatever it is, just talk to it, adjust as necessary. You guys are all smart and ask me questions as you're going along if you need to. Okay. Go ahead. So it's 6:50 right now on my clock. We'll go till 7:05, 15 minutes. Sam, if you're already building something, I recommend to go back and do this. I really do. It's, it's going to be very, very helpful. And you know what, you don't have to— when in a minute, you don't have to completely start over because you'll be able to upload it and adjust and things like that. But it will build things differently with the complete instructions and the way that we go about doing this. So up to you, up to you if you want to, you know, start over or just talk to it. By the way, if you turn on the web search, guys, you have to turn it back off if you ever ask it something like that where you want it to search the web really quick. So anyway, keep going. I'm going to just sit here. I'll keep my screen on and I'll just keep typing, typing away, going through my questions here. By the way, guys, as you go throughout your plan, I really want you to have the AI focus more on the business side as opposed to the technical side. It will say, oh, you should build a website with this technology and this technology. Be like, don't talk to me about tech. I just want business, money, vision, mission, purpose, that type of stuff. Okay, leave the tech for later. You guys got about 5 minutes left here. Hey guys, you should be getting pretty close to the end here. Finish up. Maybe one or two more questions and you should be through a couple of things that you should have covered just to make sure, make sure you've covered monetization. Having your AI, having Replit understand why you're doing this, how you're making money will actually affect how it decides to do different things. What type of features, you know, whether you need a Stripe integration at a very basic level or even the placement of a button or the flow of onboarding, right? It will, it will think about that stuff. If you have it, think that way. Okay. So anyway, I've been kind of going through and having fun designing kind of a gamified system of this Lord of the Rings fan website and artifacts that people can earn and all this type of stuff and trying to figure out how to monetize it. Um, and I've, I've gotten through enough questions that I believe it can create a pretty solid business plan for me now. So, um, Does anybody, anybody have any questions about this stage or it's pretty, pretty straightforward. It's pretty much just your ideas, whatever you're doing, but all right. Okay. So the next thing we're going to do, guys, I know you might still be working on it. Go back and work on it later. Right. I want you to kind of move forward with this next step, kind of finish, finish what you're doing here. Answer 1 or 2 more questions. You can always go back and refine later, rebuild. Do whatever you need to do. The amount of work you'll do on this, you know, tonight won't be a lot. You could certainly go back and work on it a lot later. I've had a lot of people, you know, normally it's, you know, people take class at 8:30 in the morning, they're done by noon or 1 or something like that, and then they'll work till like 2 a.m. the first day. You guys probably will be done and tired by the end of tonight, but maybe you'll jump in tomorrow morning and just keep going at it. So you can definitely work more there. But what I want to do now is we're going to say, okay, now Create a 2,000-word business plan with everything we've discussed. Use the canvas tool to do this. Okay, so that's really the tool or the thing that I want you to do. Okay, so I'm gonna, I'm gonna paste this in. I'm gonna edit mine a little bit because I have actually some specific things that I don't want it to include. Don't include the artifacts details as we'll put that in another document. Okay, so what it's going to do— so we're using a new tool here called Canvas. Canvas, as you can see, just opened up this new part of the screen here. It kicked the chat over to the left and it's opening basically what looks like a Word document or a Google Doc. Okay. And it's actually an interactive, editable document. As soon as it's finished, you can't type in it while it's working. It won't let you do anything.. But as soon as it's done, you can actually type in it and change it and do whatever the heck you want. But I've had it create this cool, hey, you know, the One fandom. That's actually pretty cool. That's not a, not a bad name. I might just roll with that. I didn't actually talk to it about the name. But the highly interactive, collectible-powered digital platform built for fans of Lord of the Rings, right? Our mission is to awaken and reward the devotion of Lord of the Rings fans by creating the most immersive, collectible-powered fantasy experience ever built online. So pretty cool. All right, so this is who we're going after. Target audience, core features, you know, free tier, premium tier, monetization strategy. This is important, you know, additional revenue streams, engagement loops, technology stack. So the technology stack, I am going to delete. I don't want it. So we'll change that. Brand and identity. Name, The One Fandom. That's cool. Or yeah, yeah, I don't know if I love The One Fandom, but that's okay. I'm going to roll with it. You can spend a lot of time. I spent probably an hour working on my name when I was doing GenAIPI because I wanted to be really official. The General AI Proficiency Institute. I wanted to feel like research heavy where businesses would trust me to hire them, you know, and to do research for them and work with them anyway. So I'll talk more about that later in more detail. I'll show you guys how I actually did it. But it's got your mission, vision, roadmap, competitive edge, things like that. I actually don't need the roadmap because I'm going to do it all in a day. I'm going to do it all in an hour. IP and licensing— this is critical, and I actually didn't talk to it about this. Avoid direct use of protected names and images. Work with lawyers to ensure compliance. Nope, not going to do that right now. I'm just going to say avoid direct use of protected names and images. That's important. Really, really important here. Okay. So I've got a pretty good business plan. It's ready to go. We're not done with it, though. That's just the beginning. So go ahead, guys. Have it make your business plan. Maybe you've already done it. What we're going to do now is we're going to have to do some research. And this is where Groq comes back in. Groq is quite good at research. All right. Part of the reason it's really good at market research, guys, is because it has access to what is happening on June or July 16th. Right? 2025. It's right now. It can look at tweets. What are people talking about? And so I'm going to come back to X here. I'm going to start a new chat. And let's see here. I click 4. Can I do this still? What happens if I click 3? Curious. Web search. X trend search. Yeah, those are just all the tools. All right, I'm going to use 4 for fun. You know, you guys can use 3 or 4, whatever you want, right? But I'm going to copy this entire business plan. I copied it. Right? So right here in ChatGPT, just click little copy button inside the canvas, and now you've copied that whole document. And then I'm going to click right here, I'm going to hold Shift+Enter, Enter, or Shift+Return, Return, right? So we don't send it. If we send it, then it will just do that. I'm going to say, I want you to give me some feedback on this business plan. Focus less on the technical side and really on the market opportunity? Where can we jump in and make an impact on our target audience? Okay. So I'll copy this little prompt right here, give you that. And then, yeah, so I'm just gonna— I've pasted it. I've got that little prompt at the top and I'm gonna go and then I'm just gonna let it start thinking about this. Okay. So thinking, thinking, thinking. It's just gonna dive in and work on it, assess the business plan. They've gotten rid of— they've simplified some of the things that used to be here. There used to be like a deep research, a deep search, a deep think, different things. But it looks like they're trying to kind of, you know, put all this stuff into one, into one easier interface for everybody. So that's cool. But it's searching tons. You know, you can click on this if you want to watch it, right? It's searching tons and tons of different things here. Yeah, theonering.net, tokenestate.com. Anyway, so it's going to keep looking around at, you know, dozens and dozens of things. 14 there, 10 there. Anyway, now it's done. 40 seconds, it's done, you know, a whole night or maybe a, maybe a whole weekend of research for me. And it's compiling it, right? It's probably doing a couple weeks of work right there just making that happen. Um, okay, so it's giving me some market opportunity overview. Your platform positions itself well to capitalize on this by blending this That's cool. Um, key areas to jump in and make an impact. Value. Run by and for real token fans. Anyway, so read through it again. This is just an example business for me, so I'm not like super, you know, you know, excited about every single word that it says. I think it's cool. I think it's awesome. Um, Potential impact, how to jump in. Yeah, this is cool. Recommendations for scaling impact, launch beta around Rings of Power Season 3. That's pretty smart. Monetization upsell, risk mitigation, Mythic Tone, growth potential. Awesome. Right for disruption. Cool. It's a good business plan. At least, you know, it thinks it is. Okay, so what I'm going to do here, guys, after I've done that, um, I'm going to copy this. I'm going to copy to clipboard the entire response that it gave me here, and then I'm going to come back into ChatGPT, right? I know you guys might be still catching up here, and that's fine, but I'm just going to say, um, use this feedback and create a new canvas called Improved Business Plan. Um, and let's hold on. Use this feedback, create a new canvas called Improved Business Plan, and Rewrite it. Okay. Don't put the milestones in and don't cover the technical side. Okay, then I'll paste all of the feedback and there we go. So now I've done some quick research bouncing between a couple of AIs, right, to try to see, all right, what do we have here? And now, honestly, guys, And you're probably reading your own content, your own businesses, and being like, man, this thing knows so much, right? It's like nailing what I'm trying to do here. But when I'm reading a lot of this stuff, it's, it's really amazing. Okay, so it's creating a new improved business plan, really trying to understand— oh, quick question.

Yeah, go ahead. I'm just trying to follow. So you, you took the business plan And you did the research through Grok. Yep. Yeah. Through, yeah. Through Grok. And then you brought it over and what did you ask?

Just, yeah. So this, yeah, sorry. This next one here, let me copy that prompt for you guys. Let me, that'll, that'll make it a little bit easier, but basically to do chats, paste. There you go. Yeah. So use this feedback and create a new canvas called improved business plan and rewrite it. Don't put the milestones in and don't cover the technical side. So I got rid of a couple of things that I didn't want it to worry about right now. We'll worry about the technical side later. We've got some cool stuff here. It's got, you know, it's kind of talking about all the things that are going to be happening in here. Let me just— I got to make sure that my artifact strategy is okay. I want it, I want it to search here. Here we go. Digital artifacts, point system, artifacts, artifacts, artifacts, artifacts, artifacts, artifacts, artifacts. Okay. Cool. Artifacts are all over the place. So yeah, you've got your improved business plan. And guys, this is just a skill that I want you to have to be able to take something and bounce it between different models. AI systems, like when you go between Anthropic, like Claude or Groq with xAI or ChatGPT by OpenAI or any other number of models, they will all give different answers to the same question. They have different models, they're trained differently, they think differently, they act differently. Okay. And so it's not like going between Chrome and Safari where you're just like, oh, this is my, you know, I like the UI here. No, there is different levels of, of quality of information based on the type of thing you're asking it. Right. And I don't have the time to go through and be like, all right, for this type of thing, do this, for this type of thing, do this, for this type of thing, do this. Right. I'm giving you a couple examples here. Business, general business stuff, probably going to do pretty pretty good in just about any of them. Market research, you can do quite well in, in Grok, right? And with writing, like really human-feeling writing, Claude is probably gonna be your winner. I know those things because I've spent a lot of time researching it and you can just jump on, on, you know, Google or Reddit or any of these things and ask, hey, for this thing, what's the best, what's the best AI platform to use? Oh yeah, this one, right? You can find that answer in probably 2 or 3 minutes of research on your own. But for here, I want you to just know we can go back and forth. So now we're in ChatGPT. We're gonna create a second document. We've created now 2 2 documents, a business plan and an improved business plan. We don't— we won't use the first business plan, but we're using this canvas tool. We're going to have it do— so we're going to say use the canvas tool and create a new document called Feature Lists. I don't want you to pick specific code— whoops— code bases or technologies to use for these. I just need a complete list of— whoops, sorry, my typing's not working— of all front-end and back-end features. Okay, copy that. We'll send it to you guys so you can use it, and we'll go. So this will create a second canvas here. So we're creating these documents, they're just sitting here living in our chat We've got a business plan now. We've got a feature list. It's going to do front end and back end. We'll talk about what those mean later. You know, user profile systems, premium features, backend logic, all this different stuff, right? Yeah. Let's see here. That's pretty good to do access validation. All right, cool. So I'm going to delete. So it put— sometimes it'll put like a summary at the end of it. Like this list serves as a foundation for future technical scoping. Okay, I am going to just delete that. I don't need it to talk to me there. This document outlines all major— yeah, the intro is fine. This is great. But now we've got— now we've got a couple of documents here, right? Now I'm going to ask it to create— all right, create a— there's a couple other types of documents you can have it create here, guys. All of these documents are going to serve as the big prompt that we're going to give Replit. Okay. And I'll show you exactly how to integrate it all and how to have it think about it long term and not just in the beginning, but as you go through to have it constantly referring to this. Sometimes I create a branding guide. Maybe I really care about what that user experience is. I wanted to define the, you know, the typography or the colors, the feel, the swirls, the animations, whatever it is. Maybe I want to go through that and take time to, to work through a branding guide. You could just say, hey, create a branding guide, you know, in a new canvas, right? Boom. And just see what it comes up with. Okay, edit this, change this, let's change that. And we're good, right? You've done it, right? I'm adding a little thing here. I'm just going to say create one more canvas that details— yeah, that goes into deep detail and contains the entire list of the artifacts, points, and tables you've created for that. Uh, do I want more? Yeah. Okay, so I'm gonna have it do that. This is just kind of an extra piece I need for mine. Maybe you have an extra piece you need for yours. Ask it to create a new canvas for each one. Okay, if you don't say create a new canvas, it will potentially overwrite one of your old ones. Does— it's done that before. That doesn't always do it, but I've seen it happen. Okay, so yeah, this is a bunch of just kind of specifics to these artifacts that people are going to be able to collect. I'm actually really interested how I'm going to create 30 different artifacts and how they're going to be displayed. Am I going to do 3D? I'm going to use AI to create that. It's going to be fun. It can be fun to think through this whole thing. I've got my artifact system figured out and I'm pretty much ready to go.

Did you say that that one was also a good one to do?

The last one I just did?

We have your feature list, but this is just—

This one's different for me just because it's unique. My thing. I needed something that, that detailed, because I created this little system where it's like you come in and you collect collectibles, right? I kind of thought about— I was like, should I try to do a Web3 like NFT platform and deal with blockchain and stuff? I decided not to for this one. So I just said, hey, let's just create this. But this is like a point system for like, hey, you get this, you know, Limbus Lover, right? Here's what you do to get it. Here's what it gets you, you know, here's the quantity of them. And then there's like really rare ones where, hey, there's only 100 of these, and then people will be able to trade and share them later on and stuff like that. So it unlocks features inside of the, inside of the system. It's kind of just a fun gamified system. So I'm just kind of having fun over here building a game. Um, bottom line is I have a business plan, I have ways to make money, um, I've got a pretty detailed plan of the why and how and my mission and what's, you know, who's my target market. I know all this stuff, and so that's really good. So now we've built all the things, guys, that we need in a really solid business plan. We've even gotten a little bit of market feedback, right? Even though it's AI market feedback, right? And by the way, I could push this really far. I could spend an hour saying, hey, do you really think people will respond well to this? How can we make this better? Right? Like, what evidence do you have? Research the internet. Find me some evidence where you really think this is going to work. And hey, let's implement that now. So you can go as deep as you want with this process. But for now, we've gotten to where we've got at minimum a business plan and a feature list, and then maybe any other supporting documents you might have. Okay, so I'm going to scratch my old Middle-earth project and I'm going to start a new one. Okay. Right here it's going to say describe an app or site you want to create. We're going to have it create a little thing. We're going to say please. Yeah, not please. I'll just say give me a 3-sentence description about what we are building. Okay, so we're just going to get it kind of a high level. This is what we're building, right? It should know exactly what we did. It has our whole conversation. It's got about the project, everything. We're building The One Fandom. Okay, it's actually the first time AI came up with an idea that I, I'm sticking with, so that— with meaning the, the name of the, of the business. Okay, um, all right, so we're gonna have a couple more things. Don't say go yet. All right, once you paste in that, that 3-sentence description, then you're going to come in, you're going to grab all your documents, guys. Okay, so grab your— you know, I'm going to grab my artifact system here. Okay, and the way that I do that again is I go into the canvas, I copy Copy it, come back into— no, not Groq, sorry, wrong spot. I come into Replit and I just paste it and it's gonna paste because it's a long amount of text. It's gonna look like a little attachment here. Okay? So you won't see all the text, but it's just a little gray box will show up there. Then I'm gonna come into, I'm gonna find my— oh, whoops, I'm in Groq. I'm in the wrong spot here. Let's get out of this. I'm gonna find my business plan. I'm gonna find my feature list. Okay? So I just need to, I just need to copy these things and paste them as attachments and still don't click go even after you paste all your attachments. There's one more thing we got to add in there and I'll do my improved business plan. All right, here I'm going to copy it and drop it in. Okay, so now I've got 3 sentences I've got 3 things that I've copied over that show up as attachments. I'll say, uh, be sure to follow all guidelines from the attachments, um, from the attachments here. Um, also, um, if you build— if you need to build a, an auth system. Just build a simple JWT setup. Okay, that's a little technical thing we'll go over in a minute, but, um, JWT is what you want. You don't want it to use— Replit has its own kind of built-in auth system, and I'm not a fan of it. So we're going to say just do a simple JWT auth system. That means you're gonna have a login, a username, password, you know, log in with a in. Super simple. Okay, so this is where you should be, guys. All right, pasted things, 3 sentences, and then this little extra thing that I give you here. I'm gonna paste it. Sorry, that last prompt right there. Be sure to follow all guidelines from attachments here. Also, if you need to build an auth system, just build a simple GPT setup. Go ahead and, um, copy that. Sorry, took a second to send that over. And now I'm gonna click start. I'm gonna click start chat. Okay, and then this will get going it after about 30 seconds, it will come back, maybe, maybe a minute, maybe 20 seconds. It's going to say, you know, do you want me to do— do you approve the plan? And just say yes, approve the plan, even if it doesn't have everything you need. We're going to go back and it's going to— it's going to be able to adjust later on plenty. Okay. All right. Yeah, very good. So I'm just going to click Approve Plan here and I'm rolling. Okay, so now this will again will take 10 to 15 minutes, guys. Um, we're doing pretty good in terms of time, right? We started at 5:30, we're at 7:30, we're 2 hours in and we're already building inside of Replit. It's pretty cool. That's pretty cool. We're cranking. We're cranking here. We're going to have a little bit of discussion now while this is happening. Okay. I want— I want to just— so there's a bunch of documents here. In fact, I should close out of all of these. There's— I don't need all these open right now. You guys are going to get access to this folder later on, this Build a Business in a Weekend course. I've got a bunch of folders and different documents and things you'll be able to go back and refer to. Lots of the prompts I'm using, stuff like that. And then just some of the workflows, some of the things I'm doing. But one of them is a quick— whoops, it's not— this is not the overviews, it's software and tools, and it's coding quick start guide. Okay, so since most of us are not developers, and I'm not even a developer, right? I'm going to teach just some high-level principles while Replit's working. You can, you know, you can watch your thing. I'm sure you'll be paying attention to that. But understanding just maybe 10 to 15 terms here.. And you don't even need to memorize them. You just need to know they exist so you know how to talk about it, and when you see it, you'll understand it. And many of you hopefully understand most of these terms anyway. You've heard them before. But when you know how to talk to your developer better, it can respond better to you, right? I've had thousands of conversations with, again, really smart developers, and sometimes they would talk to me, they'd say, hey, we're gonna do this, we're gonna do this, we're gonna do this, and Kubernetes that, and I'm just like, okay, I don't know what that is. Tell me what Kubernetes is and why is it important to our deployment? Oh, well, it does this and it, you know, here, and we need a Docker here and, you know, all this. And they start, you know, but I ask questions. I never let my, my CTO say, here's what I did without me understanding it. I would always ask questions because I just want to know. I just want to understand so that I can help them make better decisions. So anyway, here we go. Front end versus back end. Front end is basically everything you can see. Okay. So like on a, on a login form, right? You've got where a user is going to enter their email and their password. You're going to have two little empty boxes and you're going to have a login button and you're going to have some UI around it. But that's the frontend, right? There's nothing happening on the frontend in terms of logic or anything like that, although it can, right? p5.js, we showed it is possible to do frontend logic. But every login system is going to talk to the backend, right? The frontend usually just is the user interface. It's what you can see, forms, clicks, navigation. Then it communicates via what are called APIs. We'll cover that in a minute with the backend, right? Connectors, little connectors are saying, hey, I'm going to check the backend and see if that password exists, right? So the backend is where all the behind-the-scenes logic that processes data and handles requests from the frontend. You're processing user data, handling authentication and security, communicating with databases. So as an example, checking a user's login credentials against the database and returning an authentication token. Okay, so you push that button and log in. It sends a little message to your database and says, hey, does this, you know, username Jon and password admin, does that, does that exist? Yes. Okay, great. Well, here's a token. I'll let you in. No. Oh, well, let's display a message. Let's have the frontend react to that response and say, hey, incorrect username or password. Right. And so that's how it's working. You got the frontend talking to the backend and they're just going back and forth. But the data is stored in the backend, at least for any sort of really complex app. Very, very simple apps like you could build a calculator or you could build something back to the example that who was doing it earlier? Sam, you were doing your box thing. That was all happening on the frontend. It's not actually storing that logo in a database. If you were to refresh the site, your logo is gone. That's not cool. When you're using an application, you want to just upload it and it's saved to your profile and you got to have databases, you got to have all that stuff. So, so that's the next thing is database and it's object storage, right? But, but really a database organizes structured information, usually just text, right? That's really what a database is for. A database is for text. So that image actually would be stored in some sort of object storage. But your username and password, for example, okay, that's all going to be stored in a table that looks basically like an Excel document, right? You've got like user ID 1, Username, Jon. Password, admin. Email address, this. And, you know, account created at this time. And, you know, user role, admin or user, you know, whatever it is, right? I've got this long thing that's just— and then you have all these entries, right? Each entry is just a new line on that table and that's your database, right? And so databases are critical because it allows for data persistence, right? Meaning you log in, you log out, all your data is the same. You open it on a new device, your data is all the same, right? It's not storing it in what's called the memory. Of that device. It's actually storing it somewhere that will persist between sessions. OK, you can also query it, right? We won't be— you won't even have to ever do this. This is ridiculous. You'll never have to ever write this kind of thing here, but you can query it. You can ask it questions. You can create analytics dashboards based on all the information you have in your databases, everything that happens. OK, tables, relationships, keys. You'll see some of that stuff later. Connecting a database to Replit is super easy. OK, and you don't have to do any of this. Anyway, APIs and endpoints. An API, or an Application Programming Interface, is like I said earlier, it's basically a connector, right? I have— I don't know if you're like me, but I have tons of just like cables and connectors. And because I was a DJ, like, I've got so many cool audio connectors. I could connect a microwave to a speaker if I wanted to, and it would probably make some noise. But I've got every kind of connector you can imagine, right? There's billions, I swear there's billions, there's probably billions of APIs out there. There's APIs that allow you to connect into other applications, or maybe just a little piece of your database, or maybe, maybe someone else's database. For example, I love to kayak, whitewater kayak. I love to go on rivers big as they can get. Take me there, right? 50-foot waterfalls, I'm there. Okay. But, you know, one thing that I need for kayaking is water, right? I need to know that there's water there, right? And so I have a couple of ways I can set that up. One, I could say, all right, I'm going to, I'm going to call my buddy that lives by the river and say, hey, is the river on? Yeah, it's on. Let's go. Okay, cool, that works, right? Get on my Facebook group, see if people are talking about it. Two, I could go set up my own sensor by the river so that when it, you know, it's at a certain level, then, you know, and then set up a little cellular attachment so it sends it up, and now I'm getting my own little personal signals. Or I can tap into the US Geological Survey, or the USGS, that already has sensors on every river in America, and I can tap into their API where they send that data. So I can at any time check the data. I could build my own personal, you know, Jon's Kayak Buddy, right? And it just tells me when my favorite rivers are in, and I've connected it all via APIs to the USGS API. So all those databases— there's tons of databases out there. I created a cool thing for BYU, okay? And I won't go through the whole thing. This is something I created in one of the classes. I did a BYU fan, fan thing, and this is like past games for football games. And so I connected into the ESPN API, which is actually free. ESPN API is really cool, guys. It's so fun. If you love sports, you want to build something cool for yourself, your favorite team. Heck, I'll even give you this template. You can use this and then just switch out BYU for another team and it would be working, right? But anyway, let's just look at, I don't know, the Mustangs, our game last year. We won 18 to 15, it looks like. So we'll view the stats here and here's what happened. Here was the box score. Looks like we came out ahead. I can see that. Look at all this data. It's pretty cool. Team stats. The different efficiency ratings here, players, you know, who stood out, you know, not great, not great, not great, Jake. That's okay. Anyway, it was okay. We pulled out the win highlights. Now I can come here and I can say, great, hey, and all of these videos, guys, were just provided automatically by the API. And so Replit just took the API and all the data is like, oh, hey, there's video content. Let me create a video, you know, system here, right? So I could play I can watch this, I can watch this play develop here. Boom, touchdown Cougars, right? And I can even download this video, right? I can even click right here and download it. So if I want to share it on social or something, you know, whatever, I have access to all this stuff. APIs are crazy. There's so much data out there. And the way that a lot— that most of the world works is you're just connecting to a bunch of APIs. And a lot of programs out there, a lot of things that you pay money for are just APIs, like Calendly. Calendly is an API system into your calendar integration, right? By the end of this course, you guys will be able to rebuild Calendly in 20 minutes. Super easy, man. Maybe 10 minutes. Nah, you have to wait a little bit on the AI, but your actual work is like probably 10 minutes. It's crazy how, how easy that is. Okay, so we've got all these APIs, they're amazing. We're going to use them a lot. Now, every API does 4 things. That's it. This is all that APIs can do. They can either get data, okay, retrieve it, right? They can post data, send new data. How many of you guys ever posted something to social media? Yeah, that's actually a technical term, right? You post something to a forum, you post something, you're actually just sending new data to a database, to a table. Oh hey, this is user, you know, TJ Patterson, right? Hey, I love my girlfriend. I don't know if you're married, maybe you're married, whatever. I don't care who knows, right? And then, oh, you know what, I'm going to delete this. I do care who knows, right? So you're deleting, removing that data. But, but then when I go to follow TJ and I get his timeline when I click on it, because the system knows I need to retrieve the latest post from him, Oh, hey, it looks like you had a post. Oh, I refreshed, it's gone now, he deleted it. Or how many guys have ever updated your password? You really just got a link to your email with a PUT request. Just a PUT update. All it's doing is it's just that. Get, post, put, delete. It's also called CRUD. CRUD is just something that is similar to this. It's not exactly the same, but it's the same kind of thing, right? You're either posting, creating, reading, getting, updating, you know, put, deleting, deleting, right? Let's say you're in there and you're like, hey, I need an admin portal where I can have full control over all users or all posts that are put onto my platform or whatever it is, right? And so I'll say, please create an admin for users with full CRUD functionality, right? So I just say CRUD functionality. It knows, oh, okay, create, read, update, delete. Great. You need to be able to add a user, edit a user, you know, see the user, delete it, you know, all that stuff. Okay. So that's it. APIs are incredible. They're super easy to integrate. Once you've done it once, you're like, okay, I can do it anytime. Right? That's why, you know, when I was building that, you know, GenAI API and I was like, man, I haven't, I haven't integrated Stripe. I know, I just kind of knew it was going to be a little bit harder one. Some of them are a little bit more involved and, and most of the time they're harder because you have to test them and the testing with financial stuff is tough because you have to change prices and you have to put in credit cards and you have to switch between testing and real environments and it just takes a little bit of time. But most APIs like that ESPN one, I'm telling you guys, one shot, it was just like boom and it just built this huge thing. Then I worked on the UI a lot and made it feel better, but the data was all there really fast, really cool. Authentication and security, you saw me talk about JWT. That's basically token-based authentication. You can also throw in, you know, OAuth, you're logging in with Google, Facebook, etc., whatever. That's a little bit more involved. And in a lot of instances, unless you use like Replit OAuth for you, you have to go set up, like you have to get approved and you have to go through a little application process. You get approved, right? But you just have to go through it and it takes some time. And it's a lot of work. So I always start with a JWT. You can always rip out the, you know, the authentication and put in a new one or just upgrade it really. Um, but yeah, JWT is what we're going to say for today. Um, user authentication, agents, hosting deployment, we'll talk about that later. Okay, like I said earlier, guys, hosting deployment, everything happens inside of Replit, right? Inside of Replit, as it's building, you got this little plus button right here, you got all kinds of cool things that are happening over here. We'll go over all this stuff later. Okay, so all right, I've got my, my login page, it's, it's starting. We've built a ton of different things. Is anybody's app done yet? Okay.

I'm not surprised.

Sure.

Mine says no, I don't think so. Not yet.

Yeah.

It says the app is currently not running.

Deploy it to keep it running externally. Yeah, we don't need to do that yet. We don't need to do that yet. Is it still— is it doing stuff down here on the bottom left? Like, is the agent working on stuff? Yeah. Okay. Yeah, just let it, just let it chill. Just let it chill. It may also, guys, while it's developing, it might run into errors and show weird things and just let it work through its problem. It might even finish with an error, right? And we'll cover that. We'll get over that stuff once, once we get there.

Okay. I had an error and it stopped and then I had to say resolve error.

Awesome. Yeah, super easy. Okay, very cool. So guys, we're gonna take a break here in about 20 minutes. And while our apps are finishing, like really what I want to do always before the first break is I want to I want the first version of our app up, right? It's like it's going, it's working here. We have just a little discussion here. I'm going to play a short video that's just 2 minutes long from a really smart guy. I probably should have had this pulled up already, but I'll find it. Give me just a second here. Screenshot, screen recording. I might have to just click this kind button so I can find it. PNG, JPEG, GIF. Let's see here. There's a guy named Eric Schmidt, and he is former CEO of Google, um, and he's really smart. He's a really smart guy, right? He doesn't know everything, but he knows a lot. He knows a lot about tech, and I really, I really trust his opinion. So I'll share this really quick. Let's go back to here.. And we'll just watch this quick video, 2 and a half minutes.

We believe as an industry that in the next 1 year, the vast majority of programmers will be replaced by AI programmers. We also believe that within 1 year, you will have graduate-level mathematicians that are at the tippy top of graduate math programs.

So that's 1 year. Okay.

What happens in 2 years? Well, I've just told you about reasoning and I told you about programming and I told you about math. Programming plus math are the basis of sort of our whole digital world. So the evidence and the claims from the research groups in OpenAI and Anthropic and so forth is that they're now somewhere around 10% or 20% of the code that they're developing in their research programs is being generated by the computer. That's called recursive self-improvement, is the technical term.

So what happens when this thing starts to scale?

Well, a lot. One way to say this is that within 3 to 5 years, we'll have what is called general intelligence, AGI, which can be defined as a system that is as smart as the smartest mathematician, physicist, you know, artist, writer, thinker, politician, I call this, by the way, the San Francisco consensus because everyone who believes this is in San Francisco.

It may be the water.

What happens when every single one of us has the equivalent of the smartest human on every problem in our pocket? But the reason I want to make the point here is that in the next year or two, this foundation is being locked in and it's not— we're not going to stop it. Gets much more interesting after that because remember, the computers are now doing self-improvement. They're learning how to plan and they don't have to listen to us anymore. We call that superintelligence or ASI, artificial superintelligence. And this is the theory that there will be computers that are smarter than the sum of humans.

The San Francisco consensus is this occurs within 6 years. Just based on scaling. This path is not understood in our society.

There's no language for what happens with the arrival of this. That's why it's underhyped.

People do not understand what happens when you have intelligence at this level, which is largely free. All right, thoughts, guys?

When was that?

2 months ago. I think it's, uh, I think it's interesting and I think, um, it's exciting. I also think there's a— so I'm sure everybody on this call has had conversations on like the, the moral dilemma and the ethical dilemma behind like the rapid advancement of AI and what that means for just humans in general. So I'm not saying I'm concerned about it, but I mean, there is a sense of like, what if this gets off the rails a little bit here? So I guess that was just my, my initial thoughts. No, I think that's great, Sam. And I think it's very healthy to have some concern, right? If you're seeing that and you're not concerned, then maybe you didn't understand it all the way, right? That's why he said at the very end, it's underhyped, right? Like, do you, do you realize what that means when You know, every single business in the world needs a developer now, or maybe now they don't, right? Everyone needs a website, everyone needs a dashboard, a tool or whatever, a funnel, a workflow, right? And, and now all of a sudden you can build, you know, fully functioning— you know, I built a fully functioning complex, you know, game, game, gamified fan site in, you know, a few minutes. And each of you have built something similar in just a few minutes, really. Right? And so being able to replace all those developers— yeah, Sam, I think you're spot on for saying, yeah, what does this really mean, right? So anyway, let's keep talking on that.

What else? What else, guys? I think the way he explained it coming about makes— provides a mechanism that I haven't like thought about out in of how artificial intelligence or an artificial superintelligence actually gets to that. Like, you hear lots of different— I don't know, you know, in our, our news cycle and everything today, it always depends on which sides— of which side that person is on in terms of if they're trying to scare you and And so it's like, I like that it was pretty not— it was pretty center. And it's like we're literally creating the mechanism by which this becomes possible.

So no, you're spot on. I want to share something I, I saw just a few days ago. Sam Altman was talking to Peter Diamandis and he was— Peter was like, hey, Sam, why don't you— why don't you open a new headquarters in Boston? Right? And Sam was like, well, yeah, I get it. There's tons of talent there. It's cheaper. It's great. Like, I get it, right? Boston's awesome. But the reality is, by the time we could get an office set up, you know, hire people, get it up and running, right, it's going to take us a few years. By the time that happens, I'm already going to have a billion AI developers working for me. And that was the exact words he used, right? How do you compete with a billion AI developers that are working 3 levels or sorry, yeah, 3 years ahead of where we are right now? Like, let's say that happens 3 years from now, right? One thing that's really interesting of— so who here knows the math? The math on behind Moore's Law and what does it mean? Anybody know? Anybody know that off top of their head? So Moore's Law is basically the idea that computing power doubles every 18 months, right? You know, in access, you know, the cost to build like a 2-terabyte hard drive in 18 months should theoretically be, you know, half, right? Doesn't always, right? They're trying to price us out of it. But, but the reality is it actually follows it pretty dang close. Moore's Law has proven to be very, very true for decades, right? Back from the '80s, right? Every 18 months, processing and compute power becomes about 2 times more available. All right, AI is progressing 10x every 12 months. So let's do the math here. Let's, let's push forward 3 years. Moore's Law, everything will be 4 times stronger. AI, 10 times 10 times 10. AI will be 1,000 times stronger. 1,000 divided by 4, that means that AI is literally developing 250 times faster than any other technology we've ever experienced in our life. That's, that's the behind the scenes what, what, uh, you know, Eric is talking about here. He's like, people just don't even get this. They can't even comprehend it. They've never experienced this type of a, of a scale. And so, um, the estimate, guys, is that within the next— or by 2030, by 2030, 400 million jobs will be lost to AI. That's, that's like ripping right through this, you know, the fabric of society type changes. Like, that is going to cause problems. I do believe that humans will figure it out, right? We'll figure it out, we'll adapt. We've adapted to other big things. This is a big one though, right? But the transition is going to be painful, and millions, hundreds of millions of businesses all around the world are going to be gone. They are just going to disappear. And, and there's a lot of reasons for that. We have to go into every single reason, but the bottom line is when you can do something cheaper, faster, stronger, you're going to do it that way. And if you're, you know, if your competitor is doing that and you're not, you're going to lose. Right. That's why, that's why, you know, when you get back, Sam, you're talking about some of the ethical stuff, right? Like, why are we developing so fast? Well, because China is and North Korea is and Brazil is and whoever else is, right? Allies and foes, whatever we are, right? Doesn't matter. Other people are doing it. And so in order for us to stay relevant in the global economy, we have to push hard on the positive and to be able to fight the negative, right? We gotta be able to fight deepfakes. We gotta be able to fight cyber hacking, you know, AI quantum cyber hacking. How the heck are we gonna stop that when all of a sudden every Bitcoin in the world can be hacked? That's another question for another day. We'll, uh, we'll cover that later. And quantum isn't 5 years. Could be, it could be. AI could solve it, right? The newest kind of AI is called emergent AI, and emergent AI is, is basically the simple definition is it's kind of this next step past chain of thought reasoning. It says, okay, if this, then this, then this, then this, then this, well, then I'm going to come up with a new way to do something that no one's ever done before, or I'm going to solve nuclear fusion, or I'm going to solve cold fusion or whatever it is. I'm going to do things that haven't been done before. It's going to solve problems. And so if it solves quantum encryption, breaking, then we're in trouble. But, um, but anyway, guys, I, I share this and, and I want to have this little discussion here because I want you to be aware, right? You're building, you're building right now a gap, right? A knowledge gap. You guys, you know, you're— here's the rest of the population. You're now about here. By the end of the class, you're going to be more like right here. And all money in this world in the world, right, is made because of a knowledge gap. You pay your doctor because they— you pay your lawyer because they know the law. You pay your plumber because they know where to bang. All, you know, and on down the line, right? You make money because you know something better than someone else. You have access to something that someone else does and you can provide value. But what happens, like Eric said, when everyone has access to the smartest person that's smarter than any of us in all of our businesses combined? Phone in their pocket. Everyone has everything. What does that do to society? How does that change things, right? And so, there are answers. There are answers to this, but you need to think about it as you're falling asleep tonight or as you're showering tomorrow or doing your yard work, whatever it is you do while you're thinking, how can we protect ourselves? We'll have more of this discussion tomorrow morning or, sorry, I guess tomorrow night, but right at the beginning of the class tomorrow, we'll dive into another kind of strategy and good way to think about it and how to make sure you're ready for it, guys. But I'm telling you, I posted a fairly controversial thing on my LinkedIn basically saying, hey, you got to be ready for this. And I said this specific thing that AI improves 1,000x every 3 years and everyone's like, no, it doesn't. And they're like, where are you getting that? You're making that up. It's impossible. AI is— no. And I was like, actually, it's done that for the last 3 years. Literally, the math proves it. And then here's all these articles and people are like, oh, Oh, I didn't realize that 10 times 10 times 10 equals 1,000, right? The rate of improvement is so scary and so dramatic that it's almost impossible to keep up with. Okay, but the reality is, guys, you're getting this knowledge gap. And so as you're done with this class and as you dive back into your business and everything you're doing, you're going to be set up to succeed, right? You should be set up to beat out your peers. Competitors, whatever, and to outmaneuver them. Okay. All right. So we're going to look at really quickly here. So I've got my, you know, kind of first thing here. Oh, it does want an OpenAI. I don't know if I want, I don't know if I want this Gandalf, Galadriel, Elrond. I don't, I don't know if I want to do that. That's okay. We'll, we'll give it my OpenAI API key. So here's, it's, it's letting me integrate my first API key right now. And it's, this is how easy it is when you need an API to like maybe ESPN or you need I mean, you know, OpenAI, you want to have a chatbot or something that's using AI, um, all you have to do, guys, is listen to its instructions. You can get one, yeah, an API key. It's like a password that gives you your application. It's like teaching you right there. It's super easy, okay? You just click on this button, you create an account if you don't have one, go to the API key section, create an API key. It'll look something like SK, it should be kept private, okay? So here's my API key. I've added it to my system before, and so it's automatically put it in something called secrets, my secrets manager. Okay, so I have all of these account secrets. I've got Cloudinary and ElevenLabs and Perplexity and OpenWeather and all these different things, Stripe. We've got lots of keys that I can use that are just saved because I've used them in other projects, and so they're on my account. The app itself doesn't have any API keys yet except for the database that it set up. Okay, which is cool. Looks like it did set up a database for me, which is good. All right, that's awesome. But anyway, all I have to do here is I just click continue and then it will do it. Okay. For everybody else, I bet most of your systems are done now. Is anybody still working? Anybody's AI still going or is everybody done now? I'm still running. All right. When you're done, I want you to click this little plus button right here. Right. And open up the database. Okay. If you don't have a database yet, it will say ask agent to create a database. Don't click that unless you are actually— don't click that at all. I want you to just say in the message agent, I want you to say please set up a database for this, right? You want it to be using a database so that you have data persistency across sessions, right? You don't want it to be saving it to the, to the memory. You want everything to be real and set up a real database. Okay, so anyway, it's going through, it's doing all kinds of crazy stuff for me here.

And, um, yeah, it's created— oh, you're saying if we don't see database—

yeah, you'll see database, it'll be there. But you might— if you click on it and it looks like this, right, we've got a bunch of, you know, tables already set up for you, you're good. Okay, if it says, hey, ask agent to set up a database, then you're not good, you don't have a database yet, right? And so just ask the agent to 'cause if you click that button, it will erase your entire chat. And so just tell it, hey, set up a database and it will just, it'll add the database for you. So real quick, you got to that through where again? This little, so up here you've got this little plus button right here.

Yeah. But so just, this is a new place in general. You went to, oh, you're saying at the top?

Yep. Plus. Okay. Yep. And then just click that database right there.

Yeah.

And so if I see stuff there, it means it has set it up so far. No, no, no. This is all your options. I want you to click on it. Yeah. Opens this up. Yeah. If you see that data there, then you're good, right? It will say database connected, right? If it doesn't show that, then you don't have a database yet. You need to ask it to make a database. Okay. All right. So we are at 7:56. We're going to stop. This is actually a perfect time to stop. We've got the database set up. We'll start diving in. If you start to dive into your system and it doesn't work, that's okay. Okay, we'll get there. That's the next thing we're going to do. So we'll start back up at 8:10. We'll take a 14-minute break here, guys. Okay, we'll see you in a minute. I'm just going to turn off the— I'll stop the share and, and, you know, audio, video, stuff like that. And everybody go walk around for a couple minutes, take— eat a snack. AI, everything's like that, will come in and they'll just give it a prompt kind of like this, right? Hey, let's create a React web app, does this and does this, this, this, this. And they might even give it like a, a really detailed prompt, right? That's, you know, hey, and do this and use this and use that and do that, whatever, right? And then they just stop, right? Because they're like, this sucks, right? And then they jump into Reddit, Replit sucks, right? Or they jump on LinkedIn, oh yeah, I can't do anything, it's never going to replace developers, right? Well, that's crazy, man. Um, you got to be a good boss, right? You got to be a good boss, you got to follow through. And you got to know how to work the systems. Okay, so any questions or comments on this? Replit, is your mind blown? Is this awesome? Is this like, I've been there, done that, this is whatever? Or should we keep going? You guys ready to rock? Okay, we're gonna jump into ChatGPT now. So for a minute, I want you guys to just kind of sit here and, um, and watch, and then I'll set you guys loose. You'll get about 15 minutes to build your own, you know, process here as we, as we dive into this. But we're going to talk about ChatGPT for a minute. Hey, there's, there's the people. The secret's out. Oops, I had that. I left that open. There's people I'm working with. So Dean Graziosi, if you don't know the other guy, super amazing guy. And then, you know, the Tony, Tony man. Okay, so let's talk about Let's talk about— by the way, guys, you're all sworn to secrecy. Please don't say anything about this. Um. It's not ready to be announced yet, but very soon, very, very soon. Um, and they like to do it on their own timetable, so let's just start a new conversation here. Um, all right, ChatGPT is without a doubt the most— the broadest, most powerful tool out there right now. Grok 4 Heavy. Right, that model. Then, by the way, I wasn't even using Grok 4 Heavy. That's $300 a month, and I just don't feel like I need it. I have I have a lot of other pro models right now and I just don't use them enough to warrant paying $300 for Groq right now. I might, I might in the future, but right now I'm okay. If I was on a corporate plan and a corporate card and I like sign up for whatever you need, I'd have it. But, you know, I just don't feel like[No speech] I need it right now. I do pay for ChatGPT Pro. We're going to actually switch this to 4.0. So we're doing that today. But there's a couple of things I want to talk about here. I'm in the app right now. You can do this on the website as well. And I'll actually use the website because most of you will probably be be in there working on it. Oh, you know what, I am in the wrong browser. We're going to go to the one that I'm logged in on here. Close out of this. We don't need that right now. I will use Groq later. We'll jump back into here. Okay, so I want to talk about context. Okay, context is a critical term and a critical concept to understand when working with AI, understanding what it can do, what it can't do, how it works, how it doesn't work, what some people think. How it works and, and how it really can be used to your advantage if you do it right. So, so powerful if you're doing it right. So context, there's several levels of context instead of ChatGPT. And context means exactly what it would mean in English, right? Like, do I understand what you're talking about? Like you said something, but what are you talking about, right? Oh, okay, now I get what you're saying, right? I understand the context, right? Context is king. Very, very true in AI. So if I were going into, you know, one of these chats, I could go back in here, right? If I wanted to do a gorilla selfie Times Square, right? All right. So I'm trying to create one of those gorilla vlogging videos or whatever here. I could say, great, make me another one. Okay, it remembers the context of my conversation. Like I, like you saw earlier with Grok. That's all I'm trying to demonstrate. Demonstrate here. Okay, it knows I was talking about Veo 3. We'll do this tomorrow. We'll do some, we'll do some cool video stuff tomorrow. Um, but I don't have to like bring it back up to speed. It remembers everything that was in here, even if this conversation was 40 pages long. It can understand that. It can deal with That much data. Okay, then you have what's called a system level. Context. And the system level context you can access right here and click on settings. And first of all, if you work for a company like a larger company or you're really worried about, um, you're worried about privacy and your data getting out to other people, or your business model, you're afraid someone's going to steal it at GPT, which isn't going to happen, guys. But you know, if you want to be private, you can turn off improve the model for everyone. It is by default on. Okay, it's as simple as just clicking this and turning it off. Okay, very, very easy. And that will, you know, all of a sudden all your stuff's private, right? It will all be secure. No one would— no one will be looking at it. It is so secure, guys. That I will put my credit card and Social Security number in here. I have no problem with it. It's not going to get leaked. No one's going to see it. It's there. It's on your own personal encryption, okay, inside of the system. Okay, but if you're In personalization, you've got this thing called memories, right? Manage memories. Okay, now What, what it will do is it will, based on conversations that you have with it. So like last week, I decided to make a Chinese learning app. That was the thing I decided to make in the class. And so it remembers this floats a thing, right? Anyway, so I did that. It knows I've done all these things. These other things. It knows that, let's see, it knows I'm an entrepreneur, a composer, a consultant, consultant knows I live on a 5-acre farm and I have 4 children, I live in Utah, it knows these things, and there's a member of the Church of Jesus Christ of Latter-day Saints, right? That type of stuff's important, guys. Even knowing what religion you are, right? If you're Jewish, that's fantastic. Make sure ChatGPT knows, because if you ask it a religious question, you probably want that. That answer, right? That's more, you know, in line with what, what a Jewish person is going to think about something, as opposed to what a Protestant or member of the Church of Jesus Christ of Latter-day Saints is going to think, right? You You want it to know who you are. Okay. And so by doing that, it has this context level, but it's basically just is, you know, basically single or two-sentence summaries of different things that happen in your life. Okay. Or that you talk to it about, that it understands. It does not— one thing that it does not do is understand everything you've ever talked about in every single conversation, right? It does remember high level though. Each of these conversations now has kind of a little summary attached to it, and so it can help you find a conversation. It's actually gotten better in the last just few weeks on that, on on that specific topic, but it does not know if you had a 40-page conversation or a 3-hour conversation with it and there's tons and tons of text. It's not going to— it can't remember it all. The reason for that is because these models have what's called a context limit. It's only so big. It might be 1 million tokens or might be 100,000 tokens, it might be 1 billion tokens. Eventually, it will be nearly limitless and you will be able to ask it anything about anything you you've ever done ever, and it's going to know, and it's going to be able to search and contextualize and, you know, store. All right guys, welcome back. And yeah, George, I saw you on your API key step. Which, uh, which service is it? Is it like an OpenAI one, or what is it? What different things? It's It's progressing very, very fast. But right now, I just need you to know that basically at a high level, it's going to understand your, your system. It's wanting to do— you're, you're muted. No, I can't hear you. Yeah, you just click your un— or click your mute button. All right, you can hear me all right? Yep. Uh, so I'm being prompted to go do openai.com. Yeah. Why don't you go ahead and share your screen and level, like, who are you, right? What are the things you're working on? And then it's gonna— we'll walk you through it. Everybody can watch. It's super easy, but it's, it's nice to see it once. Um, all right. So I logged on on a separate screen to OpenAI outside of, um, the screen that I've got the app being developed on, the Replit screen. Got it. That's okay. That's okay. Just share the screen where you've got the OpenAI. I understand your different messages, but one place that really does understand you quite well, guys, and that's what we're gonna use now, is new project. Okay? Everything you put into a project, it's gonna understand, it's gonna remember. Right? So you can bring a bunch of chats in. You can bring a bunch of, you know, you can drag other older chain. That'll be, that'll be the easiest way to do it, but it's super, super easy. It's really fast. Just so people can kinda see inside of the API system. A little bit. So, yeah, just go ahead and log in. And note that this will— this is a different— even if you have a ChatGPT account, this is a different OpenAI account. Okay. It's a whole new— which screen are you seeing just to make sure I'm on the right one? Yeah. So I see just log in OpenAI. Alright. So you see welcome back? Yep. Yep. We see that right there.

Yeah.

So continue with Google or whatever you wanna do there. Yeah. Go for it. Chats and throw them into a project. You can bring files from, you know, the office, from your whatever. Perfect. Okay, so you're going to click Start Building up in the top right there, that green button. Whatever other AI you were working with to bring it in, that'll, that'll all, um, be stored And you'll just call it whatever your organization name is, or could be personal, whatever you want to do it. Call your project name, um, here in this project. Okay, so we're going to call this, um, LOTR Fan Site. Okay, so create project and Okay. And then, yeah, answer there. That's awesome. Create organization. Super easy. Invite my team later. Just push the button at the bottom there. Wait for it. There we go. All right, so add files. Um, yeah, there you go. And this is fine. Generate API key. Do I name it? Yeah, you can keep it whatever you want. I mean, you can name it that or you can just do whatever you want. Not really important at this stage. Just so I don't know if I'm going to have to come back or not. Yeah, super easy. Documents, code files, images, more. Drop it in there. It will then understand whatever you drop in there. Okay. You can also give it instructions, right? You can give it instructions on how— if you happen to create another one, that's fine. Yeah, PGP one. That's great. Generate API key. And there it is. Go ahead and click copy if you want it to interact with you. And you can do things as simple or as— and then you'll take that over into your other— into Replit. Put in their value. Yep. And you'll just put in that value and just click go. And then it will ask if you want to add to account secrets and just click yes. Do you want me to share that screen now? No, it's okay. It's okay. Just do that. I don't know. Fun is like always address me as Aragorn. All right. And let's see here. I want you to act as an expert. We're going to do one more thing back over here in OpenAI. All right. So I pasted that, then I click on continue. Yep. And then would you like to save this secret to your account secrets so you can add blah blah blah blah? Yes. And it's getting to work and you're done. It didn't even get— add to account secrets, right? Yep. All right. Yep. Saved secret, saved to account. Yeah. So now it's integrated. Now it's going to be in there. It'll start working on it. So go back to your OpenAI. It still won't work yet though because you got to add some money to it and that's why. So click continue right here.

Yeah.

And then right there, just got to add 5 credits, purchase credits. Well, I'll stop you sharing your screen now and you can go ahead and do that privately. But that's all you need to do. That's all. That's all you need to share. So go ahead and buy 5 credits of it. In all of your development time, you'll probably spend like $0.07 of those credits. It's very, very cheap. It's extremely, extremely inexpensive. Many APIs, guys, level entrepreneur. It's like the ESPN one I mentioned are free, or like the USGS for the water levels, that's free. Um, if you go to other APIs, like there's— there, I'll share my screen for a second here and just show you a couple examples, like the OpenWeather. Let's see here, uh, weatherapi.com. Here you go. So if you ever need weather, like maybe you want to make a weather app, or you just need weather in 40 years experience, bring the knowledge you gained from working closely with Steve Jobs, um, Mark Zuckerberg, or in your app or something, right? Pricing's awesome. Okay, it's free, zero per month, up to 1 million calls per month. That's crazy, right? Like, if you're passing that, you're making money. And if you're making money, then you don't care about the $7 per month or the $25 or the $35 per month, you know, whatever. Like, so, and you can see the different features and things are available at the different levels. But, but almost every service out there except for AI stuff, most of them have free tiers, right, where you can get in, you can test it, you can use it, you can get up to like 100 calls per month, 1,000. And choose Elon Musk, lots of, lots of fun characters here. This one's a million, right? So, um, but yeah, if you go to like, uh, you know, Stripe or something like that and look at their there are things like that. I do want to show one thing. Open and Mark Benioff. Is it K or C?

AI question, which pricing. Yeah, go ahead. I went with the year package of the Replit. Yeah. And I don't know this. When I was in the OpenAI, it didn't ask me.

I want to say it's C. I might be wrong, but it understands what I'm talking about. Here's an important part.

Don't agree with me just for the sake of the payment. Do I need to still go make sure and do that somewhere?

Yeah, yeah. Those are unrelated. The Replit year plan is just kind of, you know, the Replit side. But on the OpenAI thing, you do need to make sure you add credits. If you've added the API in, you do need to make sure you have credits. Otherwise, you'll be like, why isn't the API working? I'm trying, it's failing, it's failing, it's failing. You'll be like, why? And you'll just be debugging something, and the only reason is it doesn't have credits, right? It's got to have— it's got to have fuel to run. Okay, shake of agreeing with me actively. Uh, the credits come with the purchase automatically, right? There's 25 credits. Just agree with me if you have better ideas. Replit. Yeah, Replit does have— so yeah, so that's what I'm going to go over right now. I'm going to talk a little bit about how Replit makes money and how you can think my idea is bad. Tell me, help actively help avoid paying the money, right? How you can be really, really efficient really is what we're going to, what we're going to go for here. And then how you can spend a lot more if you, if you want to and need to. So I whipped up, again, you guys are getting a little bit behind the scenes of this stuff, but I whipped up this new thing for improve my ideas. You know, for Tony and Dean, over the last 2 days, I whipped up a new AI training platform that they're going to launch called AI Accelerator. And I built the entire thing for them. Dean's like, this is ridiculous. I can't believe you did this, whatever. And I showed him the bill. It was $26.53, right? It's like absolutely nothing. He's like, this is crazy, right? And so anyway, I'll be teaching Goal with entrepreneurship. This same type of stuff on stages with those guys, and that'll be fun. But you— when I built GenAIPI back in February, asked me to make money, February 27th to March 1st, I launched on March 1st. It cost me about $140 to build it. And by the way, guys, it took me about 35 hours of coding, right?, right? Like, tech helped me push towards that goal. Now, obviously, guys, entrepreneurship, 10 hours, 12 hours, 13 hours, you know, it was like a few really long days. I worked hard. It wasn't a weekend, but I like, I cranked, right? And I was working in 15 different platforms, right? So far, we've really gone a lot deeper than that, but I want it to really push me towards building something that's actually going to make money. So I'm going to specify that, right? Hey, I need— we only covered, you know, Grok, ChatGPT, and Replit. We're going to go over a lot of other things and how they all all come together and work together and the workflows. But, um, but anyway, I spent about $130 on Replit and then, um, maybe another $250 in other platforms as I was going along. And you won't have to spend that money this time today, right? Because I'll show you a bunch and I'll show you stuff to be making money with this. So if I have a bad idea about making money, I need you to tell me, okay? Tomorrow we'll go through. You won't have to actually buy it yourself. I just want you to know that again, for a few hundred dollars. So I'll copy this and I'll drop it into the chat here. You guys can save that for later. Launch something that a few days later gave me $15,000. By the end of that month, it'd give me $100,000. And I've kind of kept on that, on that rate. I'm about $100,000 a month with Jenni API. So anyway, really cool. But Replit, everything, all these, all these APIs work off of pricing. I don't want to talk about— yeah. Hey, George, you said, hey, can you tackle what to do when memory is full? So AI pricing a little bit, talk about a word called tokens. Okay. So all of these different models and APIs that you're integrating with AI use something called a token. And a token— this is an important question and I have run into it before— token is basically a word, right? So a million words is a million tokens. That's actually more like 800,000 words is a token. But anyway, and so there's input and there's cached input. And I actually don't know completely the difference there. I think it's like you've got input and then input that was already there previously. And then you've got output, new generative output, right? And that's the most— basically what you do is you go into chat, go up, start a new chat, and you say, I need you to summarize everything you know about me based on your memory. It's expensive thing, not a million tokens, but saying a million words or having it say a million words to you and you reading that, like, is crazy. That's like reading, you know, all of Lord of the Rings you know, twice. So you're going to be fine. You're not going to run out of— you're not going to run out of tokens. And you can see how, you know, $5 is going to— is going to take you a long time. But— but there is real cost to it, right? Okay. And I want you to keep it to 500 words. When you— when you ask Replit to do something, right, it wrote all of these words. Okay, then you, um, to, to, you know, tell me what happened. And it did, you know, 4,638 lines of code, $2. That's crazy, man. It's absolutely crazy. Like, every time I see this, I just laugh at it. If anybody here who's ever paid another developer to do something for them just knows how insane that is. Okay, copy what it tells you., then you go into manage memories and you just say delete memory, erase. But anyway, 18 minutes, 100 actions, 4,600. Man, it's just nuts. So it's memory and then you upload the memory that it just wrote back to you. Okay, so you get a good summary, gets a good summary. There's a lot. Each, each line of code has, you know, you can, you can look at some of my lines of code, right? We'll look at like the Drizzle config here, right? Each one of these words is a line of code, right? And everything you've done, whatever. And then it'll have enough context to do that. It's crazy that, so, you know, this is a bunch of lines of code. So it does, it does add up eventually you get there. But, but the reality is guys, the cost compared to anything else is 99% cheaper. Like it's just, it's not even, it's not even in the same ballpark as any other alternative. And so I'm willing to pay it personally all day long. Okay. But those OpenAI credits, you've got a bunch of different models, right? That the memory, are you on a paid version of ChatGPT by the way? Yeah. Okay, yeah, I feel you've got GPT-4.1 and Mini and different things. You got to figure out what's the best one for your use case. But most of the time you want to set it up on just something like GPT-4 or GPT-4.0 Mini really is where you're probably going to do that. And you can ask, by the way, you can say, hey, Replit, what— or you don't have to say Replit, right? But you say, which— I filled it up before. It does happen. There are limits to it. I'm sure it'll keep growing, but right now it's not. I'm on ChatGPT-4. Pro, so I think I probably have a little bit bigger memory bank, right? So anyway, guys, I'm going to save this. You know, model are we using on the OpenAI API? Oh, I've set up, you know, GPT-4.5 preview. Well, crap. No. Yeah, I am going to spend a lot of money with that, right? That's going to add up really fast. $150, $75. You don't want to do that. Okay. So you want to make sure you do something good. GPT-4 Mini. It should default to this. Right. The API is smart. Anyway, I will say it's always going to address me as Aragorn. I can change that later. I could have it address me as Master or just call me Jon. You know, I could tell it whatever I want. There's a lot of room enough to default to this. And Claude Replit is smart enough to default to this as well. They know you want to save money. They understand that. And so— but it's good to ask. It's good to just make sure you know what model it's set up on. And it's as easy as asking it. Okay. All right, let's make sure— by the way, there's a cool thing here. Really great prompts out there. You can search the internet for them to have it do different things and act in different ways, but, you know, have it act like some expert level, you know, banger.

Replit.md.

So Replit.md, this is kind of an overview of all the different things you're doing, all the stuff it knows. Replit.md used to be or didn't used to exist inside of Reput projects. Just, they only added this a few weeks ago, maybe about a month ago. They actually did it because I was talking to them and I told them that they need to do this because I taught being consultants or, you know, whatever it is that you're looking for in your, in your category. Or, you know, say, hey, I want you to actually talk to me as if you were Steve Jobs, okay? I want everything you do. I'm about to teach you guys, um, we're going to do something a little bit more advanced than replit.md, but replit.md is really really good, and we're going to use it to our advantage. So before we, before we start going in and, you know, messing with our app, which we want to do, right, um, we need to make sure that, that the instructions that we gave it, that we built in ChatGPT, are built to be like Apple would do it. Okay, anyway, so I copied that prompt. Feel free to edit yours when you make yours in a second here. Um, and, um, anyway, so now what I'm going to do is I'm going to— this plan and our feature list are well followed and well integrated. Okay. Right now they're just sitting in a little folder here called Attached Assets. Okay. You can see this folder right here. There's a little button up here on the left says Open Files. So that's how you see your files here. And then Attached Assets and you can see those. Now you don't need them. You don't need these files. But we're going to take a little— take it to— I'm going to go back to here. The prompt. Let's have my Mastering Replit document here. We're going to copy this. I will drop it in for you guys in a second. Actually, let's do it right now. And I'm going to copy my old prompt and I'm just going to do this. So what I'm going to do here, um, now— okay, so it's in the chat there for you. And basically this says, look, I want you to create— use the attached business plan in attached assets in folder— it understands that means in the attached feature list in the same folder— create a README.md file with the business plan and create a progress.md file with the feature map in Replit. That is highly interactive fan site for the Lord of the Rings list. Don't write any code, just create these documents on the feature list. Create a checkbox. Um, yeah, we'll just do that. So I'm just going to get the conversation that shows us what we have completed and what we still need to create. Please examine the app and make sure the progress.md file is up to date and add any items might be missing from the list we've created. Okay, so go ahead and copy. Patient started, we're going. It's going to say, great, awesome, Aragorn, let's map out the core structure and features. That in and go, send that I'm going to add one more thing because I have that third document that is called something about artifacts, right?

Um, call it in there. Yeah, go ahead. Like, do you ever do like a design— yeah, principles—

like file that you upload as well? Like, this is what— before diving into the code. Okay, so it's going to talk to me. Um, I could even tell it to throw in, you know, random Lord of the Rings jokes This is what I like to see, that kind of thing. And you can— yeah, yeah. So I was saying that earlier when we were in ChatGPT making some of these canvases, and you can make them wherever. You can make them in Word, don't have to do it in Kickos, you know, throughout the whole process, right? You can do whatever you want to make it fun for you, or just get down to business. But sometimes I like to throw fun things in there. So it came up with— again, it's a canvas, but you can say, this is, this is the style I like, let's put together a thing like this. You need a style guide, a branding guide, a customer guide, you know, a user— I've done a user journey guide, right, where I really want it to understand what the flow is supposed to be like that I've designed after a long discussion in ChatGPT. It's an idea, but it's still a bad prompt. Okay, so I'm kind of going to ignore what it says here. I don't care what it says. You can give it any kind of document you want, and you call it these.md files, right? Progress.md, you know, feature list.md, or whatever. Let's see, progress md, README.md, replit.md, whatever. And I'm going to have it create an artifact. So this is where things are going to start to get really cool. And this is your— I don't know, hopefully you've learned a few things already, but artifacts.md so that it's, you know, so it has all these nice references, right, that work. But an MD stands for Markdown, just a type of document that, that these systems can read. Call the artifact document artifacts.md. And do the same with it as the others. This is one you got to take away with you. This is something that I want you guys to really nail. Um, a very powerful tool, um, with AI. It's great that AI can do chain of thought reasoning, but the reality is, guys, we're happiest in life— and then I'm going to ask it to do one other thing, guys, that you can ask it. You can, you can say this in a separate prompt prompt, but I'm gonna say, please update the replit.md to always refer to these other documents as needed.

Okay?

When we are creating, right? That's our core. Every human, not to always refer, just to refer to these documents as needed. Copy. And paste. Um, yeah, the agent is starting up delay. I do get that, and it's just, um, just click refresh. By the way, you can click refresh while you're in the middle of something. If you haven't sent like this yet, then don't do it. Just click send. Their core competency of life is creating. You can even create humans, right? You can create— we create humans. Oh, and then you, then you can hit refresh, right? It can, it can do that. You can refresh it right in the middle of it and stop the process. The process is working in the background. You've kicked it off. The front end is just displaying, displaying what it's supposed to be displaying here. So it's going to create the documentation files. We create adventures, we create memories, we create music, we create art, we create companies, we create jobs. Check the project against the attached documents to ensure accurate progress tracking. Okay, so what this is going to do, guys, we create, we create, we create. And it is very hard for us, again, is it's going to give the project these bumper lanes, right? You're bowling right now. You're not good yet. I'm not good yet. I'm not a developer, right? But I know what I want it to do. I want to be a good boss, and so I want you to make sure you're listening to me constantly, right? And so boom, here's my bumper lanes. It's okay to micromanage your AI. It's okay. It doesn't get offended. Okay, employ me to find a scenario where you're saying, I was creating this and I was sad while doing this. Might not like that so much. You create all these amazing writing.md documents for your employees when you hire them or your team or whatever, and they never read them. You're like, dude, I put that in the instruction document. It's just, it brings you joy, right? Creation brings you joy. So we don't want AI to take that away from us. We want it to become a partner with us, and we want it to be our employee to listen to us. When in the onboarding document, oh yeah, I didn't read that, you know.

So anyway, you just barely uploaded those files again or you're just—

they were already there. Yeah, they're all very good thing because we— because that's the prompt we did in the beginning, right? We had those attached assets telling it to double check that it's aligning with those for the creator. It's going to just do the work, right? It's going to just be the technical behind the scenes. Now what I'm doing is these documents are not.md documents. Look at this is how bad this looks. This is really bad. And it's, you know, It's not a.md document, it's actually just text that was copied in. It's a.txt document and it's just kind of random and it's sitting here in the attached assets folder, which is not a good folder for stuff to hang out in. So it all done, make it, make it come to life. So we want to do everything we can always to have it pull information out of our head. We're in charge. So we're going to flip the script here. We're going to say, I want you to interview me one question at a time. So I said, I need you to create new documents, README.md document, right? It knows to put MD documents right here. And now we get this beautiful, nice new summary that we've got here. Nicely formatted audience strategy, monetization model, all my stuff here. And not only that, I'm asking about this business idea, the name in Replit.md that it, that it references all the time. To, you know, to make sure, to make sure that it keeps track of everything. So the replit.md is going to add in some things and say, hey, the readme.md and the progress.md. And so it's creating all these things right now. It's creating these new files out of the attached things we already did. Now you obviously could have just dragged them in, you know, at that step if you wanted to, but, but you already did it. You already did it on your first Your— the USP. So the unique selling proposition is what that is. The monetization plan. First prompt that you did when we started the project.

Yeah.

Plan, branding, you know, purpose. So. All right, let's see here. Let's make sure we're on do not disturb. So I don't get notifications here and and, um, we'll wait for this for just a second. Well, you know what, while we're waiting, I'm going to try to log in. So let's— vision, mission, and everything else. Begin your journey. New to Middle-earth? Create an account. Okay, so we'll just throw in this and we'll call it— we'll call it Jon Cheney, and we'll make it— whoa, am I typing there? I am typing there. So there's a problem. I just found my first bug. I just typed in the password password, right? And I can't see it. So obviously we don't want white text. You need to know in order to help me make a text. Oh, it's not quite white, it's like a really light color. You can see like little tiny dots there. So that's our first thing that we might need to fix. Okay, but we're going to see if I can create an account. Um, first I'm going to select this. So there's a cool thing down here right here where you're messaging the agent, you've got attach file, you've got agent tools, and you've got this visual editor or edit where you can select something. So I'm going to select this, I'm going to select this little input field here. I'm going to say, um, the text is unreadable and I type here, the font color is white and the field is white. I need it— actually, I think it's light. Font color is light and the field is white. I need it to be black like the other fields. Okay, so I'm going to click Go and then I'm going to click Create Account. And let's see, I got to choose my allegiance. We'll go with, uh, we'll go with Rivendell, maybe. Okay, hey, okay, I clicked it, but it's hard to tell, and I didn't really nice business plan. Okay, and yeah, so we'll get started on that. I'm going to copy this, drop it in the chat so you guys have it. You'll be able to use that in a second. Um, and see the confirm allegiance button. Okay, so now what I'm acting as, guys, is just like, all right, let's forge this like one plan to rule them all. There it is, going with the Lord of the Rings pun. So we'll begin at foundation. And I'll interrogate, I mean interview you one step at a time. So I didn't even have to ask it to be funny. That's a QA engineer, right? I'm a product guy now. That's the role I've stepped into. Okay. So I'm going to click con— oh, here we go. And why does that one select really nice? And this one doesn't. Oh, when I select it, it like goes, yeah, it needs to, it needs to like be bigger. Right? So that's my next thing I'm going to do. I'm going to say, um, When I select a card, I need the entire thing to be highlighted well, to be highlighted brightly. The colorful border in its prompt, it's trying to do that. So what's the ultimate purpose of this fan site? Is it a love letter to— disappears when I select it. It needs a TR, community hub, tool to draw new fans, way to sell something, merch, memberships, digital goods. I want there to be a membership, um, to be very visually clear for ultra fans of Lord of the Rings. Okay, what's happening? Also, I'd like to be able to see the confirm allegiance button without having to scroll. So I'm going to tell it that. So that's cool. Second question: what exactly would members get that justifies paying for membership? So here's why. I just want— let's see what it figures out there. Okay, we'll save my password. That's fine. All right, so anyway, um, Jon, real fast before you move on, can you tell me— will you show me again just how you are able to select Yep. Just right down here on the bottom left, this edit button right here. Okay. Thank you. Yep. Yeah. You want to show one thing. Obviously I can sit here, I can read this content and I can make up my own mind, but I want to just show you that you can say, where is the gap in the market here? And I'm going to give it access to web. You just click that and then tell it exactly what you're doing. Otherwise, another thing you could do, which I'll do a lot here in a minute here, here is screenshot, right? I'll just do a quick screenshot. I'll drag it and say, hey, look, look what's happening here. This isn't working well. We need to fix this, right? And so it's good. I will oftentimes, again, I want to be as specific as possible. I want the fewest amount of credits used as possible, right? Replit has these credits that you have. You're going to run out of them. It's not going to work well. It's weird. Why did it change it to— why did it change it to brown? Ugly. Create account. It didn't listen to what I said. Okay, whatever, that's fine. I don't like the brown, but search— where's the gap in the market here? You know, um, I want to create something that people don't have and that is novel and worth paying for. Okay, so I activated web search on this specific question. Maybe I'll complain about that later. So I already have an account. Login. That's cool. It'll save my password now. Authentication failed. Jon Cheney. I thought that was my thing. Password. So that it can go quickly search the web and it did it. It just found some, some cool stuff. Okay, so again, nope, that's okay. Cool. So the create account worked. So I just, I just did a quick screenshot of the error and showing that I'm on this page. I just tried logging in and got this with error. Okay, so the way that this works, guys, is in order for me to get a fully functioning, you know, genaiapi.org, um, you know, in a weekend, what I, what I was doing is awesome. Now it's going to give me that. This is the novel offering. It helped me do that. Now I could obviously keep talking to it about this and come back and say, hey, I want to dive into this. And, and listen, guys, if I'm, if I'm— I was basically going further and further down every user path I could get there. So this is where people land. Okay, so there's a lot of places they can go here. They can say, hey, you know what, I want to sign up. Okay, does this work? Creating my own business. When I did this for, for GenAIPI and I whipped up a business, great, this works or it didn't work and I had to fix it. Okay, now it works. And where does it take me when it works? Okay, it takes my business in a weekend. I spent 3 hours on this step. Okay, for purposes of this class, takes me to my dashboard. Great. Can I log in? Does that work? Can I log out? That's actually an important thing you got to check. Okay, what happens with that? Okay, if I click on here, if I click request a demo, does this work? If I send a message, does that work? If I click on, you know, the Facebook link, does that load something properly, right? Um, does everything work, right? Because if you try to launch something and something doesn't work, it's, you know, maybe if one button doesn't work, it's not the end of the world, but if like 4 buttons don't, we won't do that. You're gonna get 15 minutes minutes. But, but that's really what you want to do. You want to read work on your site. People might be like, yeah, this thing is— I don't trust it, right? It's like death by a thousand brews and kind of really try to build a good plan. Okay, so this is the summary, the novel offering. So I'll probably read through this and say, okay, what do I think about this? Does this look good? Is this nice? Whatever. But then it just dives right back into the next question, right? And paper cuts. So you got to make sure that every button works. And so you're going through and you're just testing everything. So it's trying to, um, yeah, so it's trying Uh, upsert a user with username as ID, but the user already exists. Uh, login failure. So it'll keep going and keep asking you questions until it's like, okay, I think I know what I need to help you build a plan. So yeah, okay, yeah, obviously I just want— yeah, we'll just, we'll see if I can figure that out. Um, edited, see data storage. Okay, let's look at my database here, see my data. We'll go over this in a second. Well, actually, we talk about it right now. So the database, guys, is really important to look at, right? After it's created, I want you guys to dive in, open ChatGPT, start a new project, set up your— created your database. You want to go in and see the data and you say, okay, here are all the different tables, right? I've got forum replies, forum topics, marketplace listings, quests, sessions, user— your prompt and your, you know, the instructions on how you want it to teach you and work with you. Ask it to interview you and I want you to get through, I don't know, 7 to 10 questions in the next 15 minutes or so. Your artifacts, user requests and users, right? And then you click on something like users, right? And you can see, okay, hey, look, okay, this is now if you're doing a brand new business, fantastic. If you're doing— we've got our first character, so we've got our first user right here. This is ID, this whatever is my email address. I know first name or last name has been set yet, but obviously there should probably be somewhere to do that. Profile image also null. So maybe there's a user profile somewhere. Faction— I did click a faction. Oh, but I didn't save it. I don't think I saved it, so I never selected that. In a feature inside of an existing business, that's cool too, right? Just don't say business. What happens, what happens if I don't select the faction when I'm signing up? Does it create the user and then never take me back to that screen? Can I never set that again? You just have to think through— say I want to work on this new product inside of my business or this new thing, you know, whatever it is. Just talk to it, adjust, do these different things. Because if you understand the data that it's trying to have in there, right? I'm at the free membership as necessary. You guys are all smart. And ask me questions as you're going along if you need to. Okay, go ahead. Here. Okay, got it. Understands there's going to be some pricing there and there's— okay, this is how it's connected. So it's 6:50 right now on my clock. We'll go till 7:05. Connected to other tables, chat sessions, forum replies, forum topics, marketplace listings. So any marketplace listing, it's going to know, okay, yeah, there's a connection there between users and marketplace listings, right? So going in and actually looking at your data helps you figure out exactly what you have here. Okay, so let's see if I can log in now, see if it works. Hey, awesome! Hey, I still haven't chosen my allegiance. I'm super happy that it, that it thought about, hey, this guy hasn't thought it yet and we need to get him there before we take it to the next step. So now I'm going to try clicking Rivendell. Yeah, I don't like the selection. It's still off. And the, you know, 15 minutes. Sam, if you're already building something, I recommend to go back and do this. I really do. It's, it's going to be very, very helpful. And you know what? You don't have to— when in a minute, you don't have to completely— when I go like this, it gets, you know, it kind of zooms in a little bit and cuts off the, you know, start over because you'll be able to upload it and adjust and things like that. But it will build things differently. What do you call that? The outline, the border. So it's not great, but whatever. It does say selected. So that's, that's something, but I'd like it to be like brighter, like make the whole card like the same color as that, you know, that blue or something. I don't know, but I'm gonna confirm allegiance. Okay. Failed to set faction. Amazing. Right? So I've gone a little bit farther. Okay. I tried to set the faction and this happened. Also, when I select a faction, make the card with the complete instructions in the way that we go about doing this. So up to you, up to you if you want to— that is selected. The— make the entire card, you know, start over, or just talk to it. By the way, if you turn on the web search, guys, you have to turn it back off if you ever ask it something like that where you want it to search the web really quick. So anyway, keep going. I'm going to just sit here. I'll keep my screen on. And card selected have an overlay that is the same color as the border color of that card. See if that works. Okay, so I'm just having ideas now. I, I don't have to necessarily have these ideas. I'm not a website designer or a UX designer, a UI designer, but you know what I am, guys? And this is where a lot of people kind of get afraid of this a little bit. I am a user. I have an iPhone and I have hundreds of apps on here, and I've used Netflix before, and I've used Apple, and I'll just keep typing, typing away, going through my questions here. Products. And I kind of know what to expect when I click login. This should happen. There should be a logout button, should be up in the top right. I should be able to delete my account or edit my name, or I should be able to do the things that you just know from being alive in the world. You should expect your app to do the same. Man, I really hate this login page. We'll work on it a bit. Bit. Okay, so it's not— yeah, the JWT token is not sticking with the authorization header. Okay, cool. So it'll fix its problem and that should work again. But I'm just going to keep pushing forward until all of a sudden I get into my app.

So to where it's actually working. Okay. Best practices for like the number of items you recommend on each time you're sending the request for it to resolve things?

I generally like to keep the items associated with each other, like related, like I did everything there was related to the select your faction, right? So hey, look, that already looks better, right? I didn't have it fix this border issue yet. I can fix that later, probably. Problem is I won't see that later until I start a new user, right? Um, and so anyway, I could just say, hey, don't, you know, don't zoom in, or maybe do zoom in, but, you know, widen the border here. I don't know. I'm just going to confirm allegiance. Oh hey, good, it messed up And so I can, I can do that. So something else happened. So this time, whoops, I need to do it again. I didn't get my screenshot in time. Okay. Got another error when clicking confirm. Firm allegiance. Fix this. Okay, so a couple things you can do, guys, if you get into a really— I'll come to you in a second, Sam— if you get into a really tough problem, and I don't recommend doing this early on, I recommend doing this later when you know this is a tough problem. Okay. Or you think it's a tough problem, you can come here in Agent Tools and you can turn on Dynamic Intelligence and you can tell it to use a high-powered model or to extend the thinking. Any request with high-powered model is probably going to cost anywhere from $3 to $6. As opposed to like a few cents, a few pennies. Okay, so it might be 100 times more expensive, right? You might put it on a big long task and it might be $17. Okay, so if you got budget to spend $1,000 building your thing, that's cool, right? But if you're like, hey, I really want to build this as fast as possible, use the normal agent. Very, very smart. The high-powered model, I bet that over the next, you know, however many hours to the rest of today, hour and a half today and all 4.5 hours tomorrow, I will never touch that button.

Okay.

You don't have to use it. You don't have to use it all the time. I just want you to know it's there. Okay. Now this is kind of dumb. I don't love what it did here. It said You know, you should be able to do this, whatever it said. The debug logs in the browser will show you exactly. So it probably built in some debug stuff, but let's see, let's see what happens here. So we'll log in, we'll set this, confirm Legions. Your allegiance has been set. Welcome to the fellowship. Okay, awesome. So we solved the problem. The faction was selected. I should be able to go to my database and I have to actually leave it and come back to my data for it to refresh, but I should— yeah, my faction is now set as Rivendell. So that recorded. I can check the database to make sure that worked right. And then, but the problem is it left me on the login screen. Great, I logged in and set my faction, but it left me on the login screen after authentication. Okay, so I'll Let it— I'll let it figure that out. Okay, Sam, what's up? Um, yeah, quick question. So when you're previewing your app here, do you need to be in like incognito mode? Is that a thing, or is it fine just public, uh, public URL. It's fine. No, no, no. So you're— are you worried about— what are you asking specifically about that? Well, I had, I had something come up when I was in this window that was like, use incognito mode. It—

I don't—

that's like paraphrasing, obviously, but I was just curious on like if you need to be in incognito mode. No. Oh no, you definitely don't need to. Got it. Okay. Um, it should be— the preview should just show right here in the, in the app. Now, if you don't tell it the JWT authentication, like I said earlier, and it uses Replit Auth to where you can log into Google, log into Facebook, things like that and other stuff. Then it will force you to preview outside of Replit, right? And so it will make you use basically this button right here, open in a new tab, where you can open your project on this weird domain in a totally separate tab. And the problem with that is you then don't have the ability to do the element selector and do all this cool stuff here and, and talk to it the same way. So that's why, that's why RipletAuth is really dumb. Really? Can I share my screen quick to show you what's kind of popping up on what I'm— I'd love to see it. Yeah, show us. Let me see. But George, You just asked about the create a database. Just say, say to Replit, don't click the add a database button. Just, uh, or unless this is coming up in the agent. Does that come up in the agent? It is coming up in the agent. My choices are add a database. Yeah, maybe later. It is sad, but sad. Click Add. That's fine. Do what? Yeah, yeah, click Add. Click Add. That's fine. Hey, you want that? Well, I'm clicking on— so, Sam, I can't, I can't see it yet. Just says Sam Gibbous has started screen sharing. Oh wait, let's see. But it wasn't actually showing it yet, so maybe, maybe you hit something wrong. Does this work? Yep, we're in. So you went— again, it's pulling this up to me. Yep. Yeah, you said it. You didn't, you didn't do this. So I'm— go ahead and escape out of that, and then let's talk to the agent for a second. And just go message the agent and say, I want you to replace the Replit auth with JWT. With what? JWT? Yep. That's all we need. Okay, let's see what that says here. Let's just watch it for a second. So it's going to refactor your app. It's going to build some new stuff here. Um, Integrating. I'll replace this. So yeah, it's going to break a couple things, but hey, your site's looking awesome so far. Um, all right, so let it finish that and it won't make you leave and go into another, into another browser or another tab anymore. Yeah, sweet. Thank you. Yeah, that JWT auth, guys, is critical to being able to work inside. I really don't like Replit auth system. That's something I'm going to be telling them to fix. So Guys, back into my app here. I was able to get past my initial login problems. They're not login problems. Really what it is, is it's incomplete steps. You can't expect it to just write 30,000 lines of code without any direction. It writes 7,000 lines of code without any direction in your first step. Then you got to start taking it down and building those little things in the right way. Okay, so one of the first things I always like to check here because it's kind of related to, to everything else is logout. Can I click the logout button? What happens? Hey, actually worked well, so that's cool. Now I should be able to log in. It's already set my faction, so it shouldn't set it again, and it should take me in. Perfect. That's exactly what I expected. If it did anything different than that, then I would have said, hey, when I did this and this, this is what I wanted to happen. So you're really just at this point talking to it, trying to figure out what's going on, right? So I don't know if I love landing right on the dashboard. It'd probably be cooler to land on the map. Um, this is kind of crazy. Um, there's no map here, obviously, so we could obviously work on that. This is a lot of visual stuff. This is pretty hard stuff, so I don't know if I want to work on work on this right now. I might work on that a little bit later. We've got these artifacts. These are all the artifacts that were uploaded in the document. Now, none of the artifacts right now have images. We might, we might start working on that in a little bit here to just kind of play with that. We got rare ones, got all these. Yeah, and they all say not collected, which is cool. Now there's some quests, no active quests. All right, so complete your first exploration of the interactive map. Map. Start quest. So, okay, I started it. Choose your faction, by the way. Select your allegiance among the peoples of Middle-earth. I did that, but I did not, um, but I did not get the— I did not get this. And so it's just not connected. These ideas that it has, what it tries to do is approximate the functionality, and then I have to go and actually test it. And, and the reason— the reality is this is fake data. Like, when I click Start Quest, it just autocompletes it for me. Anyway, complete your first— actually, maybe, maybe it didn't. Let's see this. This says in progress. Let's see if I go— what happens if I go to the map? Map. I click here. Explore, explore, explore, explore, explore, explore. Okay, I can't do that. So that button doesn't even work, right? So obviously I'd have to go in, I have to figure that out. Figure out what's going on there. If I go back to quests, yeah, it hasn't changed. So here's what I'm going to do. I'm going to say during the login process, I selected my faction, but this quest was not completed. This should be the— this should be part of the onboarding process, receiving the first, let's see, yeah, the user receiving their first artifact should be part of onboarding. Let's build this out. Okay, so again, guys, you could sit here and watch me build out this fake Lord of the Rings engagement platform all day, but what I'm trying to show is you screenshot, you select, you talk to it, you describe what you want, and you just click on stuff. And you're like, okay,[No speech] Hey, you know, AI chat, what is this? Does this work? Hi. Okay, first of all, I can't see because the text is bad as well. They did the same dumb design idea here, um, so we need to make the text here black, right? So I can come here. I wonder if I can change the text color. That's just the background. By the way, guys, as you go throughout your plan, I really want you to have the AI focus more on— I could change the background color for now. That's an easy fix. Um, let's just make it on the business side as opposed to the technical side. It will say, oh, you should build a website with this technology and this thing. I'll just be like, don't talk to me about tech. I just want business money. Let's make it that. Okay, save changes. Hello. Oh, you know what, it probably— okay, it'll do that in a second. Okay, now I understand the quest completion system. All right, so it's working through this other thing. Okay, so you saw it was gray here for a second, and then all of a sudden it turned white. Vision, mission, purpose, that type of stuff. Okay, leave the tech for it. Um, when it turned white, it, uh, it knew what was going on. So you can manually go in, guys, and you can change things. Like maybe you don't want this to profile. Actually, maybe I can't edit that text specifically here, but I could change it here. All right, greetings young one, I am Gandalf the Gray. What wisdom do you see about Middle-earth? Maybe I want to say, you know, I am, I am the White Wizard. White Wizard, right? So I want to use Gandalf, right? So I can change that and I just click Save Changes. Oh, and I can update that thing. Okay, I know it's saying failed to update text context, but but it's not. It's just sending the message over to here, so I don't know why it's saying that. All right, so this quest thing where I'm choosing my faction, um, first of all, I, you know, I can click refresh here and see if it's completed that quest for me. Choose your faction. It still says it hasn't been done. The problem is I told it to build a system where it only works as I'm logging in. It's the onboarding journey. So the really— the only way to test this is to log out and go through the onboarding journey again, right? So I have to create Create new account. I have to use a different email. I can go in and delete my old, my old stuff. We'll just call this Jon Cheney 2. Create account. Sure. And then, you know, recent artifacts not showing up, active quests not showing up. Quests, choose your faction. Still didn't work. It didn't fix my problem here, right? And so I don't know if I can Used it by asking it to do a few other things. I probably did. I just tried creating a new user and selected the faction. But this quest isn't completed as part of the onboarding journey. I'm going to be more specific here. This is important. I want the— I want the first place for a user to go to be the quests. And to show them that they have this available request that is already started and they have already completed, they should be able to click a complete. Ah, complete quest button. That then gives them the associated artifact. Then they can click on the artifact and it pulls up a page for that artifact. Talking about it and showing it. All right, so, so now that's a user journey, right? That's like a full user journey. I need— okay, this is what I want to happen. I do this, this is the onboarding. I want to go here, I want to go here, I want to go here. Your onboarding, guys, takes a long time. You guys got about 5 minutes left here, right? It takes a long time to think through, and some things are easier than others, right? Like, I've built other apps that are really easy, like that Cougar IQ, the BYU one, the fan dashboard. It's like, hey, just chill, man. Welcome. Click on anything you want. Okay. And you could have a, you know, a quick tour through like, hey, click here for this, click here for this, click here for this. You could do that kind of a tour if you want, or just let people figure it out. I have another one. Called Scribe, which is a really nice app for writing books. I may— I wrote my own app to be able to write books. I'm an author and I like to, I like to write a lot. And so I was like, hey, you know, so one of these classes I built something called Scribe and it's super easy. There's no onboarding. You just log in. It's super clean. It just, you know, it's very intuitive where you're supposed to go. But a game like this, right? All of you have downloaded that game on, you know, your iPhone, whatever version of it is. There's all these crazy ones and you're like, oh yeah, you click here and then do this and then click this and they'll take you like through 30 or 40 different tasks, you know, little buttons you need to push. And then you're like, okay, now I know what to do. They're just training you by actually helping you. So you have to consider whether your app needs more of a game-like, you know, onboarding or if it's just intuitive or it's a dashboard. Order, it's whatever it is. So that's up to you. And maybe it's just a website, maybe it's just a marketing website. You just want people to click, you know, submit requests or, you know, make a purchase from you directly. Okay. But, but this process is just an iterative one where you are acting as the user the whole time and you're just taking that user journey further and further and further and further and further until you've clicked on everything and you've verified that it works, right? So here's the profile and you don't have to do it all at once, right? You just go step by step, right? The profile here, like, what would you guys— like, you guys do my job for me for a second here. What do you see on a profile page that you would want to check here? Like, what doesn't look right to you or what feels weird? Just blurt out things. Oh, it just got rid of it. We'll go back to it in a second. But, but still say them, right? I know. I'm sure you saw a couple of things before it went away. There you go. I feel like I should have like a, a character. Yeah, yeah, exactly. I think there's like an icon. Yeah, I think this J is my character right now. Like, it's like, okay, we took the first name here, and this is probably my profile picture right now. So I probably need some sort of a character, like choose avatar. Yeah, right. And then the avatar could be like, maybe there's 5 pre, you know, or 20, you know, pre-created ones that I could create, and then, you know, Midjourney or ChatGPT or whatever it is to create something there. Upload all those into the system, or upload your own profile picture. You can do whatever you want. Okay, I might— one thing that whenever you let people upload pictures, you have to worry about pornography, you have to worry about, you know, bad stuff being on there, or just, you know, hate stuff, you know, all kinds of— you don't want swastikas on your platform. And so, uh, you know, choose wisely. I have some platforms where you can upload pictures, and, you know, it's part of the risk of having a social platform with pictures. But I agree, what else? So yeah, some sort of an image there might be good. What else do you see here that might be, you know, I don't know, important? I see a pretty critical one. Anybody see anything else that jumps out at them? I'm gonna click change faction. Rohan. Confirm. Yep, I can't change factions. Yeah, I would push that button. I gotta try. So obviously that little pathway has not been set up. That update, that PUT request, right? That PUT request is not functioning yet. So I don't see, guys, any way to edit anything. What if I want to edit my name? What if I want to delete my account? I don't have those CRUD functionalities in here, right? I want to be able to edit things. I, you know, what happened? How does it measure artifacts? Are these— is this— this is where I have, um, when I come over to the artifacts section here, I don't, you know, I see not collected, not collected, not collected. I see common, rare, epic, and legendary. That's super cool. So these are just basically filters that work, and they work. That's cool. That's, that's awesome. First of all, there's actually only supposed to be two of these right now, but whatever. It made— it was overeager and it built a bunch of stuff. It also got all my points wrong, right? And so I might have it go back and, and listen to my artifacts document. In fact, I should look at my artifacts document. Rare, 100 points. Map Walker point. Yeah, I just kind of made up a point system and didn't actually listen to me, so that's fine. But another thing I would probably add on here is mine or owned, right? That'd be a good filter. Then I just click owned and then I can look at all my cool, my cool artifacts that I own here. Okay, now again, there's no right or wrong answer for any of the stuff that I'm saying here, right? The achievements, activity, all this stuff works well. Activity, what is this supposed to show? I don't know. What do you— what would you want it to show? This is your app, right? Well, maybe I want to see like the most recent finished quest or my most recent action or my most recent, you know, whatever. I probably don't need to keep a record of all the chats right here in activity. So maybe the activity is just related to this. Maybe activity is overkill. Maybe it's not even important, right? So just be like, dude, like, this is cool, but get rid of this. Get rid of this tab. Okay, I just simplified. Now I don't have to figure out what activity is, right? I wasn't dying to have that. That wasn't even in my mind about, about being there. Achievements, I think that could be, that could be useful potentially, but, but maybe not because achievements is related to artifacts and quests. That's how I have it set up. So achievements isn't even important. And artifacts is listed here and here and here. Well, you know, pick a lane. Maybe I just need to really simplify my profile so it just has a few little things, and then when I click on artifacts, it just takes me right over to the artifacts page, right? So I, I get to design all that. I'm the boss. I get to figure out— and on the dashboard, more artifacts, right? It's crazy. I should also, by the way, be able to click right here. If I click this, right, that should take me to the artifacts page, right? When I click this, or let's see, I want to, I want this to be clickable, um, and it should go to the artifacts page. Okay, so I just kind of go through and I'm like, okay, what would I do? Hey, quest completed one. How did I do that? Did it finally complete my quest? Choose your faction. Hey, That's cool. Active, available, completed. Select your allegiance among the peoples of Middle-earth. So now it's built some sort of cool system that actually tracks completion of quests. I need to understand what that means exactly. Theoretically should have given me an artifact, but it didn't. So, okay, well, let's figure out where the artifact is. Um, Completing this quest should have given me an artifact. By the way, this is a fairly complex type of system I'm building here, this game system where it's like all this logic. It's, it's not crazy complex, but it's more than, you know, a CRM. Um, should have given me an artifact, but it didn't. Fix this. And update the profile to reflect— update, let's see, I'm going to update all users to reflect current quest and artifact status. Okay, then part of my idea was, you know, when you have artifacts, you get points, you get, you know, you get artifact points, and then those points unlock unlock different things. And I don't see a place in here where it talks about what those artifacts actually unlock. Like all the legendary ones, like one of them was supposed to unlock— oh, here you go. Yeah, unlock action ultimate mastery. That doesn't do anything. Unlocks complete platform mastery. But what does that actually mean? What I really wanted was one legendary that actually activates a voice AI. AI in the AI chat that sounds like Gandalf or whatever, or wizard or something like that, something cool. So then we could set up, we could set up an API with voice. Um, all right guys, I want to pause for a second, let you ask questions. But do you— does this make sense? Because I don't, I don't want to just keep building out my platform. What I'm trying to do is have you guys learn by watching me build and just think through my user journey and fixing different things and telling point at how I want it to act and, oh hey, you didn't do this and so you need to do this. Any questions on this? Yes, I have a question. Go ahead, George. Um, are you able with this agent, does it work much like Claude or ChatGPT? We can also query the agent, ask it questions and dig into what it's doing? Absolutely. And I love, I love that you asked that question. And just like I said earlier, you know, I, I, I'm someone who asks questions to my CTO and, you know, my developers because I want understand. And so I do the exact same here, same exact same thing here. Look, I got my one artifact point, that's cool. I've got my recent artifacts, that's cool. My artifact is here, all— hey, I got collected, right? Boom, I got it. And I can click on it, uh, point value, comment. It doesn't indicate when I'm clicking on it that it's, that it's there anyway. So, um, so let's ask it a question. So I would say something like this, I'd say, um, I want you to adjust the artifacts— whoops— artifacts to match the artifacts.md document exactly. Points, capabilities, and everything else that's on there. Don't write any code. Just tell me your plan first. Okay, so this is the answer to that question, George, right? I, I want it to just talk to me. You have to be very specific. Don't write any code, just tell me. Okay, so now it's going to do some research, it's going to look some stuff up, right? So it can look up things. Um, so here's my plan to update this, right? Currently using a 1 to 5 versus 5, 15, 35, 100. The artifact names. Okay, this is moving too fast here. So if you just, if you ask it a question, then it might write code if you don't tell it. Oh, it absolutely will. It is an overeager, action-oriented AI. It wants to get busy, right? And so, um, so anyway, I, I like it to plan first. So here's the plan. Here's the current issues. Here's the step. Phase 1, database schema updates. It's got to update a couple things and clear the current artifacts replace it with these. Then it's going to do a data migration, update existing user artifacts to maintain their progress. Right. So it's going to give me the new one that it makes, recalculate user point totals based on new values, map old artifacts to new ones where possible, update any references. Okay. Everything I'm seeing here looks awesome, right? So would you like me to proceed with this plan? Do the plan.

Okay.

Now I can— Go ahead. I've also noticed that Sam's and yours, the design was something other than black text on a white background. You have graphics, you have pictures, images, and mine did not. Mine was that way until after the agent was done. And now it's back with no images and nothing that looks good. Yeah. So what I would do is I would say, so this is something I love to do. It is able to look at any website you load. You love. So maybe you love apple.com, maybe you love qualtrics.com, maybe you love Netflix, whatever, whatever design you like. Just say to the agent, hey, I need you to do a complete redesign of the entire website. It looks terrible right now. I want you to follow the best practices and design patterns and structures on apple.com, right? And then it'll just look at apple.com, see everything, and you'll have a beautiful site. Right? You might still have to make a few adjustments and tweaks to it. Right? But, but you can have it pattern things after whatever the heck you want. But if I tell it to do a complete, complete redesign, but I have some of the functionality I'm seeing on it I like, I don't want to lose that. So I have to tell it that as well. You can, when you say design, it's probably not gonna mess with it. Right? It's probably not gonna mess with functionality. You can say, you can, but you can be sure and you can say, Hey, be sure to not mess up any functionality. Just redesign everything so it looks better. Yeah. Make this a visual update. Don't, you know, don't break the functionality that you already have working. Am I the only one who came up with a website that— can I show this real quickly? Yeah, please do. This happens to me all the time, by the way, George. You're not alone. All the time it will do something weird. Like the first time, the first, the first app I made with Midjourney, it was like, I didn't, I didn't love it. Now I did do a bunch of pictures. Yeah. Turn one photo into the something. This is all here. That is white text on a white background. That doesn't work too well. Yeah, that's ugly. Whatever. And then you've got a, I don't know, something here on the right. I'm not sure what that thing is supposed to be doing. Well, that was a picture frame. Originally, and then when it regenerated after the agent was done, it turned into this black box and that's it. So you just have to address that one issue at a time. Say, you know what, I want to replace this picture and I want to do this and just make it, make it better. Right. And one of the things that's obviously struggling with is contrast, right? Make sure there's good contrast. So everything's readable, whatever. Right. And it will calculate those types of ideas into it. Right. But you just write out that thing and say, hey, I really want it to be like, you know, you're in the photo world, right? And so say, hey, I want to take all the best things from— I don't know, Vistaprint is probably the wrong thing, but just pick something in the photo world that you like and, you know, Kodak.com, make it feel like that, you know, whatever you want to do.

All right.

You don't have like already like a UX general document that you use right now, do you? No, I don't. And the reason is because everything I've done is unique. It's different. And, and the one thing I'll talk about is what I've just been talking about with George here is you can use any website out there as your reference file, right? You just tell it, hey, this is what I like, this is what I want, go, right? And you can say, I hate that, try again, use this reference or do more of this. And you just, you just talk to it. And the other thing is it's so personal, right? It's so like I don't think there's a single design that works for everything. I'll take a minute here and show you guys around a few of my projects. We'll close down Middle-earth for a minute and I'll open up some other projects.

I've got— Can we also see the— Does it only show us the web version or is there a way to see devices?

Yep. I will show that right now. It's a great question. Fantastic question. Okay, so here's again, sneak peek behind the scenes with Tony and Dean. So this is the thing I made, the AI Accelerator. This was also kind of a dark theme that went with— it's been doing dark themes lately a little bit more, I feel like, right? It didn't for you, George, obviously. But, but anyway, I went with the dark theme on this one and this took me about, I'm going to say 3 hours, guys, to get it to this point. Okay. And it's pretty nice. Most practical and useful daily AI resource on the planet. It's gonna be completely free to, to launch. It'll be awesome. Right? Dean's department by department AI checklist. I can look at this on mobile like this. There's this little button right here and I just click whatever device I want and I can, you know, I have a different image there. You see how Tony and Dean are now really close together as opposed to over here? They're actually a reasonable distance apart. So I had to have, I actually had ChatGPT create those two variations. I grabbed an image of Dean and Tony on Google, and that, that was used for something else completely. And I said, ChatGPT, change the background, make it dark, make it match this color scheme. You can see here there's all these, you know, kind of floating little particles here that kind of make it come to life a little bit, makes it feel kind of cool. But anyway, so this is the design I have here. Everything works. I've got like software if you want to build software. We've got news, you've got quick labs where you can just like copy this, try the lab, you click this, this little open Replit actually in this case because I'm a big Replit fan. So I've, I've, you know, I'm defaulting to my biases in this platform. Top 5 tools, different things like that. So it's just getting started. This is a very much a rough draft, but, but this is the type of stuff that's being built. So that's one, one example. Let's come back into here. My Chinese Mastery School. This is one I built last week, the Flowts platform. Um, yeah, yeah, this was the one I worked on. So workspace is loading, just give it a second. And a lot of these guys I've worked on, you know, I've kind of gotten it through the initial, the initial pieces here. Um, and so I have the ability to, um, wait for it. I have the ability to finish a lot of things. You know, I finished a lot more things in some of these than obviously I have so far in this, just because I spent a few more hours on it, right? It's just time. This was my Floatzel one, right? Start learning free. I want to get in here. I have to log in. Log in. Got all that working. Takes me to my dashboard. I can do daily lessons. I can have it speak Chinese for me. Um, anyway. Traditional. That's cool. Switch between those. That's pretty, that's pretty neat. Character tree. I haven't— I didn't even build that out yet. Here's the daily lesson. Maybe I didn't even build that up. Maybe, maybe I did build that up. Maybe I just didn't do that. Um, there is— oh, here we go. Yeah, yeah, I can have it say it. So it's actually reading it. It's reading it incorrectly. I actually set up an AI that can speak to it, although I don't think I actually completely integrated that. Yeah, I got it. I got it kind of just started here. But anyway, so you could see just a couple ideas there. Let's try signing out here. So a couple different, couple different things there. Let me show a couple more complete ones that are like really good. Let me click on some that are awesome. This Fitness Wanderer one is pretty dang cool. I wonder if there's a quick way for me to open up the entire view cover page. That's not what I wanted. That's okay, we'll just open it up, but we'll do it in a new tab like this, and then we'll go to deployments and we'll just click on this. That'll be faster. So this one was called Fit Trip. I had somebody ask me— well, not really ask me technically. What I did is I went on LinkedIn, I was like, hey, um, anybody got any ideas for a piece of software they want built? I'll build it for you really quick, right? And so I spent about 90 minutes on this one. This is an AI travel coach for healthy adventures. And so you can come in here and you can say, all right, I want to go to New York, I want to be there for 3 days, and I'm intermediate. I want, you know, weight loss and strength training and endurance, right? And love protein, let's go low carbs, and love cardio, cycling, hiking. And then swimming. Okay, generate my trip, sign in. Oh, I don't know if I actually saved this. That might be the wrong password. Yeah, yay, there we go. So now what it's doing here behind the scenes actually is it's, um, analyzing all of my answers. Unable to trip. I wonder if I didn't connect something up there. That's fine if it doesn't work completely, but I should be able to look at past trips. Finalizing your personalized Fit Trip plan. I wonder if that error was fake. Sometimes it is fake, and so you have to, you have to check that too. But, but basically what it does is it puts together an entire plan for you. Um, let's just pick one up here. So like Sugar Land, Texas, View Trip, right? So it put together, you know, hotels that are in the area that have like an extensive gym, right? Cardiovascular equipment, whatever. Here's the different places it wants me to go. You want to get gluten-free buns here, but this specific type of, you know, diet. This was a different diet that I picked. Here's how you can shop. Here's some of your fitness activities. So it finds and like creates everything I can do on my trip to eat, stay, you know, activities that, that are in line with my fitness goals, right? Pretty cool. Okay. Whipped that together in not very long. And again, it was kind of, oh, here's, here's something. So for Outpost X, right, with, with Travis Chambers, right? Travis Chambers has his Outpost X site here. You know, go stay at this place. It's so cool if you're anywhere near Utah or whatever, if you're ever near Zion, right? This is kind of— and maybe not super close to Zion. It's like an hour and a half from it, maybe. But anyway, really, really amazing. Basically feels like you step into the middle of Star Wars. And I built this assistant right here where you can actually talk to it. Welcome, traveler. I am Amira. How shall I guide you? Yeah, can you tell me about the speeders? Do I have to rent them? The vehicles you speak of are the Sandcruisers, high-performance relics from the Nirmajan Resistance. They are available for you to use during your stay, first come, first served. The keys are in the cruisers, and we simply ask that you sign them out via the on-site QR code and return them to their original parking spots. Will you be wanting to take one out on the player, perhaps to engage in a training session to hone your skills? Anyway, so I can talk to it. It's well trained. I'll teach you guys how to train that, you know, this type of thing, probably tomorrow, maybe today. I might even get into this today. And then I created a separate AI to, to do it text-based, right? Because they're like, yeah, not everybody likes the voice. Some people want to just be able to type to it. And I was like, okay. So I set up a different one. I have actually two totally different agents that are not connected to each other, but they're trained on the same data. They understand the same lore. They respond the same. They understand all the same information. It would give the same, you know, if I asked about the speeders, it would, it would give a similar answer, right? Although it would be unique every time you ask, it's gonna be a little bit different, but it is going to follow this, you know, hey, the sands of Namajin and, you know, all his, you know, cool lore that he's built at his, you know, nice experiential stay here, right? So pretty cool. That was built on Replit, totally different. And it's just like a little feature, right? That's kind of integrated into something here. Let's see another couple of these. A couple, um, apps in here. I showed you Cougar Companion, the Cougar IQ, that was pretty cool. Um, Outpost X, Road Rover Roadshare Rental. This is kind of a fun one. Um, this was more of a, of like a whole new business that I was— I, I don't actually want to start it, but I had just, I had just literally gone to Outpost X actually in an RV, so, and I hated the experience of renting it. And so the next week I was like, okay, Okay, I'm going to build a better RV rental, you know, service, right? RV Match, right? And so I built this thing where you can come in, you can, you know, choose your destination, check in, check out, search for different things. People can upload stuff, sign in, sign up, featured RVs. You know, these, these were created, these images were created by Midjourney and different things like that. So I didn't, I didn't replace that one. But, but anyway, so you can see a few of these, you know, projects are just kind of started here. Let's show another one that's really quite, quite well put together. Let's see here. Homestead Manager is pretty cool. Cinematic is pretty good. Here's Scribe. Scribe is the writing tool I was talking about. I actually really like this one and I haven't shared this outside other than like on a LinkedIn post where I was like, hey, look at this cool app. This is kind of a demo. And I shared like 20 apps. And so, but this, this could actually legitimately be a product. I think it's actually quite reasonable. By the way, it's taking a minute to do this, to load up, because it's on an auto-scaling server. And so the server shuts down when it's not being used, which means no one's using it, which is what I want, right? I don't want people using it yet. Let's see if this works. I don't know what my password is. Yay. Okay. I mean, save that. So yeah, so this is cool. I can come in, I can create a new book. And I can just say, all right, you know, Cohort 16, and these guys are awesome. Whoops, not— okay, awesome. And status, you know, I'm in the idea stage here. We're going to aim for— we're going for 30,000 words on this one. And I don't know, AI, we'll add that tag. Create the book. Okay, so I can come in here and edit my book. I can open the editor. And I just can create first chapter and I can start typing, right? We talked all night long. Okay. And you know, I can, I can change, you know, edit different things like that. Pretty fun. Okay. So I can change these things. I can add a chapter, you know, the next night, create chapter. We talked about sales and marketing. And lots of other things. So this is a pretty cool— I, I like this app a lot because it's just clean, man. It just takes almost no training. I know you just put this in front of somebody and you just know what to do. And so it was really well designed. By the way, my prompt for this, my coach on this one was Steve Jobs. I actually had Steve Jobs. I was like, I want simplicity, clarity. I want you to— I just want to be be so user-friendly. And look at the result, like, it's, it's awesome, right? Um, I can't— I've got a corkboard, I can drag stuff around. I don't know if I can drag stuff around yet, maybe that's just there right now. Yeah, I would easily be able to create it where I can just drag those around and move things around if I needed to. I can add characters, right, where I can then, you know, um, integrate them and put them into the thing. There's lots of really cool stuff. I didn't create the templates thing, so I never did that. Bookshelf settings, OpenAI API key. So this is a cool thing I did here, guys. Instead of putting my own API key in, I said, hey, if you want to have the assistant work for you, then here's how you create your own API key. And they could put their own one in here. So they use their credits. And now I don't have to worry about like crazy usage bills or anything crazy like that. I just let them drop their API key in. I've trained the prompt, so it's gonna be my prompt and everything in there, right? But then they've got this cool thing where while they're writing, um, it should be able to— maybe I don't, maybe I didn't, maybe I didn't turn it on. Um, add card editor view. Yeah, there was supposed to be a little, a little assistant, a writing assistant right here, or maybe it like slid out. I don't know where it is. I don't know where, I know where it went. Let's look at the, uh, mobile view. Yeah, there it is. Why is it not showing up on desktop? So yeah, there's a problem. But yeah, here's my little AI copilot, right? And so yeah, I want to generate some ideas. What would you like help with? I can ask it this. And I didn't actually set it up, but this was the idea, is that they could come in and, um, and use the AI to help them write, and I wouldn't have to worry about my credits being used while they're doing that. So pretty neat, pretty neat little thing there. And Scribe was going to be— I think I was going to make it free, um, like in, you know, for, for a long time, um, and launch it free and just like get tons of users. They're all using their own API credits, and so I don't have to worry about tons of, uh, tons of cost to run this thing. So I really like this one. I think it's really fun, um, super useful, super simple. Um, here, here's a, um, this was a backyard. I was like, hey, you know, I'm thinking about getting a pool and So I'll build a website, you know, as if I was a pool builder, right? And so I built that here. This is kind of just a sales site, you know, Verde Pools created some cool images and things like that. Anyway, I created one of the, one of the main things I worked on a lot here was the Estimate Your Dream Pool, right? I can just click this and, and then I can, you know, choose how big I want this. I want a convertible indoor/outdoor. That's so cool. And terrain type is rocky. Yeah, it's kind of flat-sloped. I don't know what it is, but I want this one underwater window and I want this, you know. Anyway, and it's kind of building out. It's pretty cool. So I worked on some of the pricing and stuff. I actually had a pool guy that was on, on the course. I didn't even know he was a pool guy until I started my idea and he's like, hey, I actually do pools. That's kind of fun. Um, so Luxury Aquatics, uh, I'm gonna show just a couple. Actually, you know what I'm gonna do Let me show you GenAIPI. Okay, let me show you guys inside GenAIPI. Lots of cool stuff in here that I built. So one of the things I built here, I'll talk about this a little bit more tomorrow, but the onboarding style that I decided to go with was a very similar style to if you've ever taken like an IQ test online. You take the IQ test, it's free, and you take like this, you know, 20-minute test. And right at the very end, it's like, pay $3 to see your score. And you're always like, come on, man. Like, anyway, so it says like it's free and it is free, but if you want to see your score, you got to pay, which is stupid anyway. But I said, hey, I'm going to do the same thing, but I'm just going to say, all right, you go through the test and, you know, as soon as you're done, it will ask you to enter your email. To see your score for free, right? So it wasn't costing anything, it's just, hey, I just want your email. So that was a way for me to get, get, you know, emails in the system, obviously get people to test their knowledge and to get them some sort of benchmark. Because the, the whole, the whole process was set up so that no matter what you get on this, you're going to want to keep engaging with my site, theoretically, right? Because if you do bad, then it's going to suggest, um, to take courses. And if you do well, it's going to say, well, why don't you get a certification and prove it? You know, put it on your, you know, show it off, put it on your LinkedIn, right? So anyway, I would create it. You'd create an account, continue. When you log in, you know, that would then take you to your dashboard right here. And then you could either browse courses, come in here and check all these courses, which I'll show in a second, or you can do the certification, pay $39 for this, and then you can, you know, you've got an industry-recognized certification industry-recognized, guys. I, I, it said that on day one when I launched this thing. Now it's actually becoming an industry-recognized certification. People, real people, are turning to it and actually doing it. I've had a lot of people take these certifications. But guys, this is how everything— this is how the ACT started, right? They had one user at some point, right? Everybody did. And so you just have to remember that as a, as an entrepreneur, like, you're making the rules.. And you can't be— you shouldn't be afraid to just be bold and go for it. I stated something, I put this out there, and it became it, right? And it was— it's obviously gone really well, um, but, but building out a lot of the stuff was really cool. So I built all these modules, each of them have a module quiz, right? There's 4 lessons, so you continue the course, come in here and just say, all right, I want to do this module here. I've got all these different lessons. Lessons. They all work. Again, I had to click so many buttons. I built all of this in that first 3 days, guys. It's crazy. Okay. But I could take, I could do a quick, a quick lesson here, read through it, read through it. And by the way, I, all of these lessons are terrible. Don't buy these from me. I've had a couple people buy them and I just feel bad for those people. I need to improve all of these. They're just text ones. The better, the better lessons are being worked on right now. Um, man, this is a long lesson. Probably like 8 slides. All right, which of the following best describes how AI-powered email management differs from traditional email filters? Yeah, that's different. That's different. That's different. I don't know. Let's try that one. Submit answer. So hey, boom, I got the right answer. By the way, if I got it wrong, it would have told me why I got it wrong, right? And so, but, but the point is I just want to show all the functionalities that I built. Okay, the functionalities are great. The only thing that's bad about the course is the content. I just need to upgrade the content. I'm working on it right now to do that. But, but this, this is beautiful, man. Like, I click continue, takes me on to lesson 2, takes me right back to the next thing. I can come back to the module if I want here. But now I finished this. It's, it's showing progress here. This is complete. I can review it, go back if I want. I come back in lesson 2 that I didn't complete. I can't take this quiz this module assessment until I've completed all the sections, creating all that logic and asking it and telling it, okay, hey, I want, I want, I only want to be able to do this once all of these tests or quizzes have been passed, right? And I want this progress bar to match this and 1 of 4 lessons complete. I want it to state it properly here. There's a lot of different things in here that are really important. Here's another, here's another critical one. So this general AI proficiency test is $39. Um, one of the things that you want to probably manage in your own database is your pricing. Like you might set up Stripe or something, and Stripe actually has a place where you can set up products. You can set up products and prices and all that stuff. And if you want to manage it through Stripe, you can, but I just don't love that, right? I love having control inside of GenAIPI. I can jump in here and I can actually come in and go to my— by the way, I have 78 tables. I have a lot of different things in here. Um, but anyway, so, uh, let's go to, uh, let's see, certification tests. I've got this one certification test here, but this is really, really critical. You want to update things, guys, and have them be driven by the database. Let's say I want to change the price of this test to $59. So I changed it to— this is all in pennies, right? So I change it to 5,900 and I click save, and I come back to here and I just click refresh. And all of a sudden everywhere where that price was $59 or was $39 now is $59. So it shows it there, it's going to show it right here. I think it shows it on the home page. If I sign out, I think there's a spot here where it talks about it. Yeah, $59. So I had to make sure that anywhere where the price of the test was mentioned wasn't what's called hard-coded. Right? It wasn't just hardcoded in. It's database-driven. Right? And so when you guys are building out, you know, maybe a product page or something like that, or even the court, you know, like my courses page here, right? This is database-driven. It's pulling these from the database instead of hardcoding it. So if I want to then edit it, I don't have to go redeploy my site. I just have to go into my database, go into my courses, and update them. If I want to change the name, I just click on this, change it, and click save., right? If I want to change the description, the price, whatever, right? Any of the content, whatever I need to do, I can do it all because it's all database-driven. Super important. Course test questions, test progress, right? All of these different, you know, all of these different elements here are, um, yeah, quiz, quizzes, sessions, slides. Yeah, the slides. There's so much data in the slides here. And by the way, maybe I don't want Maybe I don't want to adjust stuff here. Maybe this database thing is kind of meh. Well, great. Let's create an admin. So I did create an admin section for myself where I can control everything on a nice front end. Here's my user management. I want to look at my name, Jon, right? Here's all the people named Jon that are in there. I want to edit course content, modules, lessons, live courses, right? Participants, course content reviews. I want to manage my reviews. I can do that right here. Here. All of these reviews show up on janaipi.org right here. Live courses, view details, uh, reviews right here. They're all right there and they're right here. Oh, oh, well, I can change those and edit the, the order of them. I don't like that that button moves. I need to edit that. It's like have it be a static spot. But anyway, um, so yeah, just little things, right? But that's just a tiny, tiny, tiny fix. That I can, that I can improve with, right? But all of that is edited right here. I can add a review. I've got this. And so I'm even making my job easier for me. I'm a user of my platform also, right? I'm making my job easier to be able to go in and edit things and manage all kinds of really cool stuff. Test content, test results. Test results was an interesting one. I really needed a, well, I had somebody reach out and said, hey, I took your certification, but I can't find the email with the certification. I was like, okay, I'll email it to you. And I jumped in the system. I was like, I was like, crap, I have literally no way to show that certification. And so, so I built a system where I can now just search for user and look at test results and view their certificate right there. Right. And then I can download that PDF and send it to them. Right. And it downloads that PDF. And so I had to build a PDF thing and that's great. I had to build out an ID system that follows the structure inside the data was actually taken. And this is all accurate, right? And by the way, did I have this on day 1? I did not. I waited until somebody asked me for it and they're like, hey, it didn't work. And I was like, I haven't actually built that out yet, right? I'm just going, I'm just building and launching. There's certain places where you can build. There's also a couple things that are weird here. If I'm logged in as, sorry, let's see if I can get this to work right. Yeah, so see how like all this spacing up here, the admin and things like this, it just doesn't look good. Who cares, man? Like, yes, I could spend the time to go through a couple prompts and probably fix it in 10 minutes or 5 minutes or maybe even 3 minutes with one prompt. Or I can just say, you know what, I have bigger fish to fry. I need to go get more customers, right? As an entrepreneur, you have to be like, okay, I need money. I need, I need to turn this into money. Eventually I need enough money that I can hire somebody else to maybe manage this so I can focus on sales, or maybe I need them to focus on sales so I can focus more on this. Whatever your strengths are, do the thing you love. Let AI fill in the gaps. Okay? You got to get to money fast. Okay? So the point here, guys, with Replit and all the things we've been going over tonight so far, and you know, we've got about 30 minutes left here. We'll do a couple more fun things, but Replit can help you get something working that is way more than way more even than an MVP, right? Your traditional MVP like barely works and whatever. This can be really complex. And as long as you click all the buttons and go through all the user flows and journeys, it's going to take you 30 or 40 hours, maybe 15 hours if it's a relatively simple thing. GenAI API was complex because it had quizzes, it had certifications, it had tests, it had all these different things. It had payment systems and different ways to do all of that. I originally, when I launched, did not have live courses. That was not part of my business plan. Plan. My idea was, can I launch a business that's like totally self, you know, self-contained? I don't do anything. Obviously, I'm having to work a lot with you guys with these live courses. And after about a month of running Jenny API and talking about it, people are like, dude, how are you making money like that? How did you, how did you do this? And people asked me to start a course. I was like, all right. And then I decided to throw the course on here so that, you know, people were just originally Venmoing me $1,000, you know, like that's how, that's just how I was doing it. But again, that's the entrepreneur in me that knows like, okay, I just got to, I just got to make it happen. It doesn't matter if it looks awesome or if it's perfect or if it's Venmo or if it's an Excel sheet or whatever it is. But with Replit, you can look pretty dang official really fast and you can whip up things like that Scribe, that writing app like that app's ready to go almost. I bet another, I bet 4 or 5 hours of just really solid work on that thing., and I would be ready to throw that up on like Product Hunt and just say, hey, if you want to write books, here's a free tool, go for it. Bring in your own API key from ChatGPT and go nuts, right? Write books, do cool stuff, and people will vote it up or vote it down or whatever. Like, I don't care. Like, I'm just putting it out there because that's how you make money. You get it out there, you get it in front of people. All right, um, any questions, guys, or things you want me to show you inside of Rep, whether you're just like, man, I'm trying to figure this out, or maybe you're working on your project right now and you you can't quite, can't quite get through it. Hey, Jon, just a quick question on— I see in the upper area you've got preview and console and secrets and database, and I just see preview and console even when I said like, hey, I need you to make secrets, I need you to make database. I'm just wondering why that's not popping up for me. Yeah, it's just a matter of closing or opening the tabs. These are just tabs. Apps. Okay, so just come in here and just click this little plus button that's right here. Um, okay, see that? Ah, yeah, all those. Okay, so yeah, I do have a database, it's just not pulled up here. Okay, yeah, awesome. Yeah, very good. Other questions? I think George had a question. Yeah, a very simple one. I'm, I'm panicked that we get off this call with me not understanding how I simply save this file because it doesn't look obvious to me. Is it automatic? It is, it is saved. Um, one thing we'll go over right now, I'll show you guys, we'll go over this right now. I'll go back to our current project here. Let's go to apps and we'll go to OneFandom and I'm going to deploy it, right? So deploying, guys, is, um, we're basically going to have— once we deploy, we're going to have two versions of our app. We're going to have our dev environment, which is here in Replit, all this stuff, right? And then we're going to have our public, right, or production. And so we're going to come here into one of the tabs here and it's going to say Deployments. Okay. When you come into Deployments, I recommend you just go Auto Scale. Okay. Now you saw this Reserved VM Always On Servers. Okay. That would have loaded my sites faster, right, on that first click. Right. Auto Scale means it's going to turn off when it's not being used. And it's going to turn on when it is being used, and then it's even going to scale up if 1,000 people click on it at the same time, right? And so that's probably what you want. It's the cheapest version and it's never going to get you in trouble. If 1,000 people accidentally click on your site or 10,000 people accidentally click on your site, it's just gonna be like— now your, your costs related to the site will go up for a minute while you have 10,000 people on it because now you're running a bunch of servers, right? And I'll show you guys how that works. But anyway, Auto Scale is great. Set up your deployment. This is like the number of machines. This is the max number of machines. If you're worried about, you know, tons and tons of people getting onto it, you could scroll it up. I'll just say this, all of my sites are set at the default settings. Okay, so approve and configure. You can rename it, right? I can call it, you know, onefandom, onefandom.replit.app. Now this is a.replit.app. You can add it. I can change it to onefandom.com. Later. I can do that, OK? But right now when you first deploy, it's just kind of this temporary domain. This has all your secrets and you don't have to do anything here. You just kind of rename it if you want or don't, you know, and then click deploy and you're done. OK, so once you've done that, now you're, you're pushing it up. It's going to go through this 4-step process here where you are then, you know, it's then public. You could share it with your friends. You could share it with your wife or your business partner or whoever, right? You know, as soon as you're ready. And you don't have to do this tonight, right? Ashton, I'm sure you're not going to send this over to, you know, your girlfriend until it's ready, or I don't know, like maybe you want to wait until it's perfect, or you can be like, hey, look what I built in like an hour, like this is so awesome. And so this is how you would share that, right? You're going to get it up to where you can quickly share this and put it in a public location.

Now there's a few things that we haven't done. Sorry, go ahead. Just one question on that is, so for example, if mine is like the journal that I'm building that has my personal data in it, when I share, I'm sharing a version that will, in this deployment stage, it will make them create their own account and their own data, correct?

Yeah, if that's how you've built it. Yeah. So if you have a login that's required, then, then your personal data should theoretically be behind what's called an authentication token, right? And if they don't have a token with your username, they're not going to see your data. It'll load up and just be blank and empty for them. Now, one thing that I love to do, right, is before I actually deploy— now I did deploy already and it's fine, and it's actually okay to deploy and then run this, but you have this cool thing here called Run Security Scan, okay? And so if you click that, then it's going to check really quick. It's going to say, okay, You know, are you, are you accidentally exposing an API key? Are you accidentally, you know, letting people in without having to actually log in? You know, is something not properly set up right? This should find, you know, anything, anything really big, and it did. No vulnerabilities found. I'm not surprised. This is— I kind of know what I'm doing, and I know how I built it. I know that I did the JWT. I know that, you know, when they land on it, they're going to land on this. Like, if you were to go to this link right here right now, boom, you're going to land on that. Okay, this is the ugliest site ever. I hate it. Make the login route page way better looking. Completely redesign it and think like apple.com. But it's that simple, huh? Lord of the Rings, um, theme. Also, don't break anything. Okay, I mean, that's it, George. Like, if you just— wow, you just tell it to do it and it will do it. It'll make it better. Now, will I like what it does? I don't know, because I didn't give it, you know, crazy things. But I just, I just hate the way that this thing looks, and so I need it to not look like that every time I quick on it. So kind of a very thing. Most of them, when I, when I talked about saving a file, you, you jump to deployment, but we wouldn't deploy until we get the thing ready for anybody to see it. It's still going to save what I'm working on without me clicking on it? Correct. Yeah. So if I were to literally just close this browser right now, like, all of my progress is saved. I haven't saved anything. Everything you do, every move you is automatically saved. Yeah, perfect. Yeah. So you don't have to worry about that. This isn't Microsoft Office, Microsoft Word. Anyway, so yeah, that's not that hard. Okay, guys, tomorrow, let's talk just a little bit about what we're going to talk about tomorrow. And then I might, I might, you know, dip into a couple things that we can do right now to just kind of get ahead of ourselves on some of that. But tomorrow we'll do images, we'll do video, we'll do sales, we'll do marketing, right? All AI driven for all of this stuff. We'll do voice. Maybe I'll knock out voice tonight. Voice is actually not that hard and you guys can start playing around with it. We'll also talk tomorrow about building your own GPTs and how to really train your own agents and AIs, how to understand, you know, what your user needs and what it wants and how to get it to produce outputs. Kind of like, kind of like you saw with the Outpost X assistant, right? How did I train it on all that data and have it, you know, be consistent? If I ask it the Wi-Fi password, it'll give me the answer. If I ask it whatever— hey, we're already doing better there. That was already better. Um, okay, yeah, apple.com inspired landing page. That looked just like Apple just for the second. It crashed. It's trying to install something here, and that's fine. We'll let it, we'll let it, um, we'll let it work through its problems here for a second. It knows it's crashing, so it'll fix myself. But anyway, so those are some of the primary things. We'll actually go into like Meta ads and different things like that and show you how I'm making money, how I'm making money right now with GenAI API, what my strategies are. You know, one of the things, guys, that I do that any of you can do, right? It's not something that is hard is, you know, I'm jumping ahead a little bit on the sales and marketing but to get to build software now, guys, is almost free. Like, I know it's not quite free, but it's almost free. Like, it's crazy how free it is now. And so because that— and by the way, this won't be a secret forever. You guys are learning about it, and that's awesome. A lot of other people know about Vibe Coding and stuff, but, but everyone's gonna be able to just develop stuff really fast, right? Um, yeah, so I need to obviously do a little bit of redesigning here, but it is, it is a little bit better. It is a little bit better. Um, let's go refresh. Um, but you just got to tell people about it, man. I, I got myself, um, when I, um, let's see, it's here. Yeah, KUTV, Jon Cheney, Jenny IPI. So I, um, I just reached out and I was like, hey, I built a six-figure business in a few days. Right. Told my local news station, KUTV, Channel 2, Channel 2 News, and they did an article on me. And not only an article, they actually came out and did a— they like sat right here in this room and filmed with their crew. And I was on the news, right? I had this clip, shared it, all this stuff. I got easily 100 signups. I probably made $100,000 from that, from that little news story. Right. It was free. Took me one email, right? People, people are looking for like people as in David Ochoa at KUTV. Writers are looking for things to write about. Reporters are looking for things to focus on. Those are Good Morning Utah, Good Morning Arizona, Good Morning Maine. All of those sites are looking for stuff to talk about, right? If you can show up with something to talk about, you're doing you're helping them, you're doing them a service, right? Um, anyway, I need to work on this, but now we've got kind of a marketing side here, which is— it's good. It's not, it's not just a sign-in page, but it's all kind of too black. Uh, it's too black.

Add some LOTR-inspired images and make sure the contrast is good on all texts. It kept making my stuff super black too.

I finally said, I want a white background, why are you not giving it to me? Yeah, I love asking it like, why are you so dumb? Why can you not figure this out? Like, what the heck is wrong with you? Why does it want to— why is there a watch preview button here? In order to explore factions to join, I thought there was— I thought there were 6 factions originally, but that's okay, that works. Works. AI characters. Yeah, see, that's kind of cool when I— oh, it's already, it's already improving it. It's already making it. Yeah, there. See, it's already, it's already working on it just as I'm talking about it.

Wow.

Anyway, again, guys, it just takes time and being in patience, and you have to understand that it's not going to get it all right the first time. See, like, fixed it, then it broke it again. Like, anyway, whatever. Okay, so we'll work on that. Um, I want to teach one thing. So one thing that's really cool is you're getting ready to deploy and you're getting ready to share with with others. Something that a lot of people don't think about is, you know, just kind of little things like the favicon, right? Anybody know what the favicon is? Right, favicon is just that little thing right here up in the top of the browser, right? There's GoDaddy and whatever this is, Provo Beach and Google and whatever, CallRail, I don't know what that is, Google Drive, right? So I recognize these and I can quickly obviously be like, oh, here's GenAIPI, that's my site, right? Because I've got a nice favicon set up. Another thing a lot of people don't think about is what a link looks like when you share it, right? Here's my, you know, conversation with, with Dean Graciosi. I'm talking to him about the site and look how beautiful that looks. The most practical daily AI resource, right? That looks really pretty. If I were to share right now, even Outpost X, right, this is not going to look good. Let's just share it with myself here. Paste. Yeah, see what it looks like? It's just this ugly nothing with no logo, and it's just not cool, right? Um, I'll text this to myself so that I can actually see what it looks like here, right? But you can see it just— it doesn't look good as compared to CBS Saturday Morning, right? That's a YouTube link, right? So that's, that's got properly set up favicon and a properly set up what's called Open Graph share settings, okay? And so if you share, if I were to share this link on Facebook or LinkedIn or something, it's going to look crappy. If I share this link, it's going to look great. This is what general AI proficiency is. GenAI API looks like this, right? And so setting up those little types of things, guys, you just have to know to think about it. And by the way, maybe you don't know to think about it, or maybe you don't know that it's called a favicon. Then just say, hey, you know that little image that shows up in the browser? Can we get that working on this? It knows, right? So just talk, just talk to it. Ask it, ask it how to do it. Okay. All right guys, we're gonna do a crash course in ElevenLabs. This is super fun. ElevenLabs is one of my favorite things. It is an AI voice platform and I'm gonna teach you some cool stuff about it tomorrow, but I'm gonna demo it a little bit tonight. We're just gonna play with it for a few minutes here. Okay. So, oh yeah, that's awesome. We got all kinds of stuff going on. People are using a mirror. Yeah, Amira, mindfulness, Chinese tutor. A Chinese tutor was cool. I like that one. Look at that, Amira, 66 calls, 43-cent cost. That's awesome. That's replacing, guys, like, an agent, a person. A person used to have to answer those questions. So cool, so cool. 76 minutes of call time for 43 cents. I, I don't even know. I don't even know, man. That's so cool. But anyway, guys, in here you come to voices and you can listen to different voices. Um, okay, so this might sound silly, but I just got this weird sparkly feeling in my chest, and I think, I think I'm actually falling in love. It's actually a pretty good voice. I really like the hesitation and some of that stuff. It felt, it felt pretty real. Here's Hope the Podcast. Welcome back to The Balance Podcast. I'm Hope, and today we're diving into something that's been coming up a lot. So cool. So anyway, guys, you can go through and you can pick a bunch. They have a bunch of ones that are created. You can also create your own voice, right? So we'll do like an instant voice clone here. Record audio. 3, 2, 1. All right. This is Jon's voice, and I'm gonna say all the letters. A, B, C, D, E, F, G, H, I, J, K, L, M, N, O, P, Q, R, S, T, U, V, W, X, Y, and Z. So now you have a full range of how I talk. Okay, and click next. And I've created a voice, right? So we'll call this Cohort 16 John. And this is John. I confirm I have the rights to this. By the way, guys, maybe I need to add a label. Oh, I need to do this English. And I don't need an accent. Okay. So by the way, guys, you could absolutely like recreate Morgan Freeman's voice. Just download the YouTube video that has his voice and drop it in there. Now, you know, are you allowed to? I mean, yeah, probably not, but you know, if you're going to use it for something internal or whatever, like who cares, man? Like don't sell it. Don't break any laws that way. Right. But, but the bottom line is, you know, you probably want to respect other people's voices. This you can, you can, you know, clone other people's video, like their likeness to lots of cool stuff. OK, so let's narrate this now. Let's just, let's just speak with myself. So let's see if I can find my voices. And here I am. Let your light shine brightly and brighten dark paths. OK, so it's kind of fast. I'm kind of talking fast a little bit there. I wonder if I can come in here. Let's see, let's see voice. I don't want to try out the new clone. I don't want to try that out. I want to actually let your light shine. Instant voice clone. I know there's a way to edit like how this works. Maybe I have to go into agents to do this and that's fine. Let's go into the mindfulness agent here and voice. Yeah, yeah, here we are. So yeah, this is, this is good. So I can come in here and I can switch it to cohort 16, right? Then I can optimize the streaming latency. Maybe I want it to respond pretty quick and I want it to be pretty stable. I want to slow down a little bit because I was actually talking pretty fast when I did that thing there. I want the voice to be pretty similar to me. Let's even go 90%. We'll save that. This agent, by the way, is set up to Let's see here. Your name is John. You're a mindfulness coach, right? So it's going to help me relax as my own voice. Let's test it. Hey there. Thanks for taking this moment for yourself. Let's begin by introducing ourselves. I'm Joe. What's your name? You're not Joe. You're Jon. I just read your prompt. You're right. My apologies. I'm Jon. It seems I got a bit mixed up there. Thanks for pointing that out. Now, where were we? Yes, I was about to ask your name. Yeah, I'm Jon also. By the way, I like your voice. It's a pleasure to meet another Jon, and thank you. I appreciate that. So Jon, what brings you here today? Is there something? Anyway, so the reality is, guys, my voice clone that I just did with only 12 seconds of audio is going to be far inferior to another voice clone that has 30 minutes of trained audio, right? And so you can do a quick one like that, just kind of test it, see if you like it. Or, you know, I honestly like using a lot of their other voices, right? Like there's some, there's some crazy, like this is like whisper. This is gonna feel like maybe a little bit inappropriate almost. Maybe if I save it as Clara here, let's go agent and let's see. You are Clara, by the way. Why did it see Joe? Why did it call me? Oh, I'm Joe. There we go. I'm Clara. Yeah, there we go. That's the problem. Joe, there's Joe right there again. Clara. Okay, no more instances. Now it's actually Clara, although she's going to sound a little bit seductive here, right? Because that's the voice. That's the specific voice. She's kind of whispering. And anyway, let's see how this works. Taking this moment for yourself. Let's begin by introducing ourselves. I'm Clara. What's your name? Hey Clara, will you stop whispering? My apologies, I can adjust my tone. Is this better? No, it's exactly the same. I understand, I'll adjust my settings. Sorry guys, that's funny to me, man. I don't know why that's funny, but you can actually— there's, there's some cool variables you can add into different agents where you actually have it change its voice based on, you know, so you'd have to upload a couple different voices to it. I don't remember how to do that. Detect language, transfer to agent, transfer to number. I've seen it somewhere. I don't remember exactly where it is right now. Oh, here it is. Voice. Oh, add voice. Multi-voice support right there. Yeah. So you can say, okay, so it's going to start out as that and then we'll have it go into, um, in the middle of every Born in Italy. No. Cherish the earth and its— Sure, we'll use Eve. And we'll just say Eve 1, add voice. Okay, and then we'll save it. So now I should theoretically be able to say, hey, stop whispering. I want to see if it actually gets that. Hey there, thanks for taking this moment for yourself. Hey Clara, will you go ahead and stop whispering? Of course, I can speak at a normal volume. Is this better for you? You know what? I think I actually like the whispering better. Will you go back to that? Certainly, I can go back to whispering. Does this feel more relaxing for you? Nope, you have to go back to whispering still. My apologies, I thought I was whispering. I will whisper now. So it clearly, it doesn't know that it can whisper, right? And so the way I could fix this, guys, is actually pretty easy. I could say, hey, when the user asks for whispering, use Clara's voice, right? When the user is asking for non-whispering, use Eve-1, right? And then it would— it should be able to get it by then, right? But anyway, setting up an agent, guys, takes like as long as you saw me do it, right? Like, obviously I clicked into one that was done, but new agent is super easy. Blank agent. And, you know, create an agent. You select a voice. Eric. Gratitude is riches. Complaint is poverty. Okay, thanks, Eric. All right, it's ready to go now. Now, the only thing I haven't set up is what's it going to talk about today? It's set up as, okay, hello, how can I help you today? That's your first message, right? Like, yo, what's up? Okay, add variable. Interesting. That's cool. And then I can say, hey, you know what? You're just here to have a good time. Okay. So that's the agent.

Describe the agent.

Yeah.

I mean, that's all I need. Where are you putting in? So you're talking about having agents who are prepared to speak about certain things.

So like, where is context that you build in, so to speak? Yeah. So number one, it's using an LLM right now. So I can set it to use ChatGPT 4.1 Nano or Mini or any of these things. It's going to use Gemini 2.0 Flash because that's going to be the cheapest one right now. Okay. And so it's like you're talking to ChatGPT. You're just talking. So it can talk about anything that ChatGPT or the Gemini might know. So you could just say, I want to talk about Lord of the Rings. All right, man, let's hang, you know, you know, you're just start to have a good time, act like a surfer dude. Okay. And, and so that's my prompt right now. If I wanted to train it on a bunch of other things here, then I'm going to come down and I'm going to use this right here. Agent knowledge base. I can add documents, all kinds of stuff. And then I turn on what's called RAG, Retrieval Augmented Generation, increases the agent's maximum knowledge base size. The agent will have access to relevant pieces of a attached knowledge base during answer generation. So if you want it to be trained on your, you know, system, whatever, or maybe you want to just answer your phone, right? You can connect this to your phone system, okay? And then you could say, hey, gives the agent the ability to transfer call to another AI agent or transfer to a number, right? As long as they get past the agent, they can then transfer it to you and your phone can ring, right? You can give it all kinds of different instructions and capabilities here, it's very, very easy to integrate. Um, and I usually put it all together inside of Replit. So I'll take the agent, I'll take the— my API key that I grab right here, right? Here's my API keys. Jump in here, come in here, create an API key, and then, you know, go back to my agent. I'll grab the agent ID right here, and I hand those two things to Replit and say, all right, build this, this is what I want it to do. Just go, right? And so if you want to do— build something like I did with Outpost X where you're just talking to a mirror, that's awesome. Maybe you want it to be answering phones, and so you got to connect it to your phone system, Grasshopper, or, you know, whatever system you use, and it's ready to rock. But anyway, this agent— let's go back. Why did I call it Chinese Tutor? I didn't mean to call it Chinese Tutor. I want to edit name, Server Dude. Um, let's see. Yeah, what's Steve. Okay, now it's got a name. Save. And we'll test the AI agent. And we'll just ask— I'm going to ask it about just something general. I don't know. Yo, what's up? I'm Steve. Hey Steve, could you tell me about the moon? How far is it from Earth? Hey, how's it hanging? The moon, that gnarly orb in the sky, is about 239,000 miles from Earth. Did you know that it's the Earth's only natural satellite? Do you have any other questions about the moon, or are you interested in hearing about other stellar stuff? Stellar stuff. I see what you did there. That's pretty, that's pretty clever. Right on, man. Glad you caught that. So what other stellar stuff are you curious about? Anyway, so I can tweak it and play with it, but, but you saw how fast I set up Surfer Dude. It's like 3 minutes, guys, right? And so for your own business, your own things— by the way, guys, what do you think this tool costs per month? There is a monthly fee. What do you think it costs? $10. Must be $11. You know what? They should be. It should be $11. It's $5, guys. It's $5 a month. It's great. Absolutely insane. Okay. With $5, you get up to 10 different voices. You can create custom voices, whatever. You can swap them out and whatever. Whatever. So, and then if you go up to like $17, then you get like 30 voices and it's just crazy, right? The value that you're getting out of these things, guys, is absolutely insane. So you said you share over, you connect, you need the API token and then what else for the agent? Yeah, you just the agent ID. And by the way, it would ask you, right? Replit would say, hey, I need the agent ID. And so you'd be like, okay, Okay, agents. Okay, uh, I want my Surfer Dude one. Oh, there it is. I got it. Or if you don't know, be like, Replit, where do I find it? Where do I find the agent ID? Oh, click on agents. It's going to be right under the name of the title. You know, it'll, it'll, it'll tell you, right? If you don't know, just talk to it. Just ask it, right, guys? By the end of tomorrow, anytime you guys ask me a question, I'm gonna be like, dude, you could have just asked Replit that. Like, in the same amount of time it took to ask me, just ask Replit. Right. Turn on your— I don't know if you guys use dictation much on your computer, right? Just use voice to talk to it. Right. But just like turn on dictation and, you know, talk to it and let it do stuff. So, so anyway, guys, we've got about 2 minutes here. I'll let you guys go to bed. I'm going to go to bed too, and we'll see you guys tomorrow night. But we got a lot done today. Being able to build software, like I said, software isn't everything related to business. But it's a heck of a lot, right? It's usually how your customers find you in the beginning. It's often how they engage with you even while they're using your product. It's how you get leads. It's how you have them purchase things from you. It's, you know, websites are critical. Platforms are critical. And so we're going to not spend quite as much time in Replit tomorrow. We will spend some time. We'll integrate like, we'll integrate Surfer Dude into my Lord of the Rings thing, which won't exactly match, you know, the feel, or maybe we'll switch it to to a wizard that sounds more— yeah, we'll probably do that. But anyway, we'll integrate an AI voice into there. We'll show you how to do that. We'll build some more ChatGPT, like show you how to really build your own custom GPTs and things like that, get prompts right. And then sales and marketing, like I said. We're also gonna spend probably the first part of tomorrow working on or talking about agent, like agentic systems. Systems that can like do a huge amount of things for you. Like Replit is an agentic software developer. It's all it does. But what if you want it to build a PowerPoint system for you? Or what if you want to do a bunch of research for you and create some reports? And what if you wanted to, you know, do other things? Um, there's a bunch of cool agentic systems that you can get to work for you. And, and guys, when I was building GenAIPI, you know, I didn't just sit there and watch my agent work in Replit. I would be Okay, do this. And then I'm, you know, 10 other tabs, you know, I'm setting up my Facebook page and my Instagram page and setting up the LLC and setting up my Google Workspace and then getting email configured and setting up SendGrid and Twilio and all the different things I felt like I needed for this to actually feel like and be an actual real business. And I've done all those things a million times. And so, I was kind of just on autopilot, just like bouncing through. I created, you know, a couple dozen social media posts, right? Images with image generation, all the posts, all the content. I created ad content, got my ads set up and all that stuff. It's all just ready to go so that as soon as Replit was done and I couldn't find any more problems and I was ready for a customer to land on it, I could just click go on the ad and it was done, right? So the actual process of sitting down and saying, I want to launch a business in a, in a quick sprint in a day or 2 or 3 or 4 or 5 or whatever it is you do,, you know, you can be doing a ton of stuff and having all these agents just kind of go to work for you.

So any questions before we finish up today, guys? Okay, I had one little one. Let's just go ahead, but it's, it's more specific if, if you want to do it after other people. No, go ahead, just go ahead.

Yeah, it was just like the MD files, they're just still PDFs instead of MDs. They're PDFs inside of the attached assets folder or what? Yeah, inside of Replit. Go ahead and share your screen real quick. Let's just look at the file structure really quick. It might be there, you just don't know where you're looking. That might be an easy fix. Yeah, guys, I like learning by watching other people learn. Your MD files are there. I see them. Your attached assets folder is is garbage folder. They don't matter. But look down on the left there. You see progress and README.md?

Down. Yeah, right there. Yeah, click on one. Boom.

They're open. Okay. So, these are the— that's the MD file created.

Yep.

Those sit over there. Yeah, the attached assets folder, guys, is not going to load up to like when you deploy it and stuff, it's, it's, it's just kind of a temporary storage file for your development, right? And so that's why you want to put it into the project, get it into that MD file, and it'll move it down into the normal structure. So, and it changed the— gotcha— changed the titles of them versus up here. Yeah, and you don't have to— I mean, you can keep it called, you know, README and whatever, just whatever you want to call it. But yeah, don't call it business plan, just keep it README. You want it to— you want the replit.md file though to know the names of those so that it knows to constantly reference them.

Okay.

Yeah. All right guys, um, that's it. Have a great night. We'll see you tomorrow, same link, same time, and, uh, we'll dive into part 2.

Awesome.

Thanks, Jon.

Have a good night. Appreciate it. See you guys. Thank you. Good night. Thanks.