Cohort 18 - Day 2 - 8/20/2025

Jon Cheney leads developers through building AI applications with Replit, OpenAI, Claude, and Grok. The workshop covers integrating LLMs with Freshdesk and SendGrid, building custom agents with memory, and deploying with CI/CD pipelines. Jon emphasizes that mastery requires hands-on practice and iteration.
Chapters
A lot of people, but then we'll just get some of this conversation recorded here. But, um, oh yeah, that's crazy, Jeff. Um, super, super excited for you. You know, some people like dive in and just like, I, I got this, like, let's roll, you know? And it sounds like you're catching on quick, so that's cool. How about you, Michael?
And then the thing that got me so excited too was with, uh, one of our divisions with Blue Wave and the residential kind of tough nut to crack that it is, just because it's just the, the scale isn't there, right? Individual homes and, and dealing with individual homeowners and HOAs and all the other difficulty there. But I was like, man, this is— this seems rife for, uh, you know, for change. And I could— with this process, I could probably rebuild a website that easily 6 figures was put into. And I said I could probably have that whole thing up and going, you know, within a day and have something that would rival that, probably function better, and have all the workflow in terms of automating the emails and all of that, which has always been such a difficult thing for us. But, you know, copy, connections, uh, infrastructure could all be done.
Yeah, yeah, for sure. That's awesome. Um, I love it, I love it. Welcome Stephanie and Lisanne. Michael and Natasha. All right. We've got a, we've got a good group this morning. I think I'm going to text Taylin because I know he just like doesn't look at his clock just so that— let's see here. Taylin and Chris. Chris is actually going to be a few minutes late. He texted me. Okay. All right. So I know we have a few people that are still missing, but we're just going to kind of dive in and get started. Chris is a little bit late. Emily mentioned to me that Corey's going to be a little bit late today. And so that's fine. And again, this is recorded, and so people can go back and kind of catch up. Where I want to start this morning, guys, is I just want to open it up for questions, right? We went over a lot yesterday. It's very easy to be like, oh my gosh, I don't feel confident in this thing, or you kind of talked about that and I was so lost, I didn't know where to go, right? And Stephanie, by the way, I just got to point her out. She texted me last night and said, hey, look what I did, right? And she sent me, you know, her first working version of a website. And it's not pretty or perfect. But it works. It's there, and she's got through the, through the first step. And so I, I love that. And, uh, and she said something that was important too. You know, we all have different learning styles, right? Um, some of us can just like follow along and do it right there. Some of us need to see it a couple times, and some of us need to like pause, think, and then work it at our own speed, right? And those are all great. They're all great ways to learn. And, um, and it's, you know, don't, don't Don't feel bad if you're like, oh man, I just, I didn't get it. Or he's going too fast. I am going fast. Right. And that's why we record it so that you can go back, you can watch it again and then internalize it. Um, and, and then lastly, before I open up for questions, I had a, or I watched yesterday, has anybody ever, ever watched, um, the, the show Limitless on, um, not the one where you like take the pill, but like the one on National Geographic with Chris Hemsworth. Mm-hmm. Thor. Right. So cool. It's on Hulu or Disney Plus, whatever you wanna watch it on. Right. My kids love it because it's, It's Thor, and, you know, but he's, he's going through, he's trying to figure out how can I live longer? How can I be smarter? How all this stuff, right? And, and it's in season 2 now, I guess. And the first episode I watched last night was about him trying to learn. Well, he kind of talked about how our brain, you know, we have our two hemispheres in our brain, right? And then there's this thing in between them that connects them and that, that breaks down. Over time. I don't remember what it's called. One of you probably remember science or biology better than me. But anyway, the thing, you know, that these fibers break down over time. And as we get older, our brain shrinks. And this is one of the things that happens, right? And so, activities that we can do that use both sides of our brain are especially good. And not only that, when we learn new skills, that our brain just is strengthened. And those, there's more fibers and more connections and more synapses that happen. And so he decided to learn an instrument. That's what the doctor said, "Hey, you know what?" You know, he just kind of talked through a bunch of things. He's like, "You know, I'm not very good at music." Like, "Perfect, I want you to learn an instrument because instruments, you use, you know, your ears, your eyes, your brain, like coordination, it's tactile, there's a lot of different things. It's using a lot of senses." And then it also, you know, stretches you to learn something, especially if you're not good at it, right? If you're not great at that thing. And so anyway, he goes through and he decides to learn the drums, and he does it. He goes over to his buddy's house, Ed Sheeran, and, um, and, you know, goes into the studio, plays all these different things. Like, Ed's like, okay, you're gonna try to learn the drums. And, and Chris Hemsworth sits down at the drums and is just horrible at it. Just, I can't even describe, like my 8-year-old could do better than him. He has no sense of a beat. He can't handle it. It's just, it's not one of his strengths. He can do everything else well, but he can't do music. And at the end of the episode, his goal and the kind of, you know, kind of self-imposed or imposed by Ed goal was Ed said, hey, you're gonna play a song with me on stage in front of 70,000 people. And so that's obviously like, and by the way, you only have 2 months to learn it, right? You gotta learn 1 song on the drums. You'll be able to do it. And I'm telling you guys, I did not think he was gonna pull it off, but go watch it. It's amazing. It's amazing what our brains can learn when we stretch them and we use them in these different ways. And I feel like what we're doing here, guys, is a new skill. Learning how to talk to AI, how to do all this stuff is a new skill. But the more you use it, the better, easier it'll get. I just got a message from Sharon saying, hey, I can't get in. So we're gonna take just 2 seconds here. I'm gonna try and, um, just send her the invite link here again and see if, see if she can get it to work. Try this. You guys all did it, um, so let's hope, let's hope she can get into. So we'll watch for her. All right. Okay, and then let's also turn on focus mode. Okay, questions? Let's open it up. Mitesh, I saw your hand raised early. Go ahead.
Yes, thank you. So my question will be followed by a little bit of sharing, if I may, about what I implemented last night. Is that okay?
Amazing.
All right, thank you. So I've been having this idea of— I use Freshdesk for emails for customer support. And I've been wanting to hire some kind of an AI agent that can learn from my emails and respond to emails automatically instead of the employees doing it physically. And I found a lot of tools out there which were like $500 a month, $600 a month. And I tried all the demos, nothing really worked. So I gave a very simple prompt based on what you taught me yesterday to ChatGPT, created a project saying, step 1, I want a small AI sign within my Freshdesk emails. If I click on it, It should read my email, step 1. 2, give me a suggested response. I will read it. If I like it, I'll send it. The send button should be within the Freshdesk platform. If I don't like it, I'll edit it and whatever I edit, the AI should learn for the future reference and there should be a knowledge base for that. That's it. I give that prompt to ChatGPT. I said, create a prompt for me for Replete and Lovable. And I added the same prompt in both Replete and Lovable. And you were right, Replete was damn good. And believe it or not, the whole thing was ready in 5 minutes. And I mean, I can show you, it's mind-blowing how the—
I'd love to see it.
It is beautiful. Like this is what it got created there. So this is Lovable. It looks more pretty. Yeah, but I didn't really get to work the actual end product yet. But when I really got to Replete, oh my God, it really got things working. So yeah, here's the Replete thing. It created this in about 5 minutes, 10 minutes. It took my OpenAI API and then it said, okay, so here is an email. And this is exactly how the Freshdesk actually looks. It created that for me as a sample. So here's the sample email. He says reply. Here is your AI reply button. So when I click on it, OpenAI is actually reading through my email right now. Creating a suggestion. And if I like it, I use it. If I don't like it, I can edit it and it just creates it based on my database. So if I say, okay, use this response, that's it. It is using it and sending it right now. That's all it says. And then I can just hit send here. So I've reached here so far.
Yeah.
I don't know how to go. So that's my question. How do I go from here? But this is my flow. Like this is exactly what I was envisioning.
That is so cool. I'm, I'm really impressed, uh, Mitesh, first of all. Like, I, I just love this, right? The thing that I love about this course— so I've taken other courses from, from people, uh, you know, selling courses online about, hey, do this, right? And I swear 90% of those courses are like, take a course and then, okay, let me teach you how to make a course to do this, right? Like, that's all it is. But the types of solutions that are coming out of this class are so broad, and this is— and so specific to your needs. And that's what I love about about AI is it can just help you with exactly your problems. So, Mitesh, yeah, you've gotten, you know, the core skeleton is there, the core functionality is there. Now you just gotta tweak a little bit. Like, so first of all, I saw that when you did that, it didn't actually fill in the reply, right? No. And it probably should. And so you just say, hey, after you do this, you know, make sure it shows up right here, right? And use your little edit button down there, the little selector, right? And say, hey, make sure it shows up right here. Okay? And, and so you get that, right? And then you say, okay, now what happens when we click send reply? Like, tell me, like, so, so don't, don't just click it and try to figure it out. Ask the AI, say, hey, I want you to tell me what happens when I click send reply. And then it will say, oh, it's not actually connected to anything. How would you like to connect this up, right? And so then you'll have to find, you know, whatever API you can API into, into, you know, your Google Mail or whatever it is that you might have. You can API into something like SendGrid or something else if you want to do that. I'll show, I use SendGrid. So all the emails you get from GenAI API are all coming from SendGrid, you know, through that API system. It's a very powerful email system that's capable of this kind of stuff. And really it's just a matter of asking. And this is where, you know, some of you might be like, I don't know what SendGrid is. I don't know what all these— this is so hard, right? The really important thing that Mitesh did is he just read what the AI told him, right? That your developer, your AI developer, will just help you. It will guide you, right? You don't have to feel like you're overwhelmed. If you don't understand it, say, I don't understand what that is. I don't know what that is. Can you help me? And it will respond and it will just walk you through it, baby steps. Explain it to me like I'm a 5-year-old, right? And it will do that for you. And so that's the beauty is you have this AI working right there with you. That's very powerful. The AI that is inside of Replit is actually Claude, right? It's Anthropic, right? It's Claude. 4.0, right? And it's very, very powerful. It's one of the best AIs that is out there. Out of the top AIs, you've got, you know, ChatGPT, Grok, Claude, maybe Perplexity. Those are really kind of the ones that, you know, Gemini— Gemini is really actually not very good, but, you know, it's fine. It's— but it's not as good as these other ones. In my opinion, Claude speaks the most human-like, right? It just feels the best across the board. And so the fact that Replit uses Claude and Claude Code is just primo, right? They made it. That's one of the reasons why it's a very, very good product. Replit is very, very solid. They're making the right decisions in a lot of places. But today, guys, we will go over more APIs. We'll help you. I'll show you how Mitesh did what he did, right? How do we get AI integrated? How do you make AI, you know, think about that, right? Mitesh built an AI-powered email assistant in a day, not being a developer, right? That's crazy. That's awesome. That's awesome. Um, and, and if I would have asked Mitesh, you know, a week ago, hey, could you build that for me? He'd be like, I don't know, like, I've never done that.
I was going to pay $500 a month for that. That's like $5,000 a year expense.
No, you know what, so there's another— you remind me of another thing. So I I was looking at one of the things that's important for this new AI LLM-powered world is discovery, search, right? Traditional Google search is, I don't want to say on the way out, but it's adjusting at least, right? The ratios are adjusting and more people are going to AI directly for answers. And that can work sometimes, sometimes it won't. But the bottom line is they're starting there. And even in Google, when you search, there's an AI overview first, right? So how do you teach the AIs and the LLMs to know what your business is about? How do you get them to suggest you, right? Hey, I need help with AI in my business. How do we get— how do I get GenAI API to show up, right? How do I make that happen? Who do I target and how do I structure the data properly, right? And so a company just raised about $30 million in San Francisco. That's doing this, right? And they're doing it well and they have a lot of really cool stuff. And, and I looked, I was like, that's cool, I want to try it. And I clicked on their free trial and it's like, okay, it wasn't actually a free trial. It's like, fill out this thing. Okay, now $699 a month. It was really expensive. And I'm like, I'm not going to pay $700 just to try something. Like, no way. So I jumped on Grok and I said, Grok, tell me about this company. What are they doing? Like, what, what are the technical things behind the scenes that they are doing, right? Look at their website, look at reviews, look at Reddit, figure it out. What are they doing? And it came back and after, you know, a few minutes and said, okay, they're doing this, this, this, this, this, and this. I said, perfect, put that into a nice guide on how to prepare your website to be better, you know, scrapable and findable by AI search engines or by, by, by LLMs. And it said, great, put together this guide. I took that guide, I handed it to Replit, and I said, Replit, update my site with all of this in mind. It said, great, I'll set up this new AI SEO structure and boom. So I just saved $700 a month, $10,000 a year by simply asking an AI to tell me, you know, hey, what, what, what is this thing doing? Okay, great. Here's what it's doing. It's a bunch of technical stuff. I don't know how to do it. I hand it to my other AI who knows how to implement it. It implements it. Boom. I've saved $10,000 and now my site is extremely well structured and now I have this cool document that I move from project to project now. If I need, you know, my AI SEO structure set up properly, I just drop this document in. I say install all this, follow these guidelines. Great.
I'll do it.
That's cool. That's cool. Awesome. Great, great, great discussion there, Mitesh. Other, other questions? Go ahead, Stephanie.
Okay. So you mentioned Claude was in Replit, which I didn't actually know, but we were doing everything yesterday, getting all the information we needed in chat. And then migrating it over to Replit.
Correct.
Can you do what we did in chat within Replit, or do you always need to bring something over that's ready to go first?
Yeah, Replit doesn't have a great planning tool, right? And one of the things I like about ChatGPT is they have this project, right? This project feature, right? And so you have this project, you can give it all these files, all this reference, and then it remembers— remember context, how important context is, right? Needs to know what are you doing? What is this all about? Right? And so I really like using ChatGPT as my project manager, right? And ChatGPT-5, guys, is no, no slouch. Like, it is a solid product. Very, very, very, very well done. But there's no, there's no reason you couldn't do it inside of Replit other than that the user interface would be bad. It would be really tough to do, right? But when I, when I created GenAIPI, just share my screen for a second here on this. So I can just kind of demonstrate what's going on here. Let's see here. I've got 100 things open, guys. I'm sorry. All right, here we go. Um, GenAI API, all of these courses right here, um, all of the content in these courses was written by Claude, right? I just kind of gave it a structure. I said, okay, I wanted to have, you know, a few different lessons or modules inside of each one. There's 4 lessons. I want a quiz based on the content. And then at the end, I want a final course test, right? And so all of the content was, was created by AI, and then I set it up to automatically update itself every month. I basically said, hey, every 30 days I want you to go and I want you to research and I want you to update the content to make sure we're sharing the latest and greatest content. And so now I have completely taken that task off of my hand forever, right? This content will update itself forever now. Is this the best course in the world? No, I don't think so. I think it's just okay. I think it's fine. It's definitely useful if you're trying to learn AI and, you know, you're a beginner. This is a great place to start. I think it's okay. I personally like videos and things like that, and those are harder to update automatically, although not impossible. There's actually plenty of ways to have an avatar of yourself just deliver things. And so, but guys, use AI to build that content, right? And so even though, you know, we didn't do that, Stephanie, inside of Replit, Replit is still very capable of building out content. It could do that strategy. It could do some of that research for you. It could do all of that stuff. But because simply, simply because of the user interface, right? We're gonna choose ChatGPT. And, and quite frankly, I do always love to have multiple models working on my project. Right? It just refines the content. And so if I start in ChatGPT and then I implement with Claude, and then I say, okay, what can we do to the content? And now you built the homepage. How can we make this stronger? Right? It's gonna still, you know, you know, follow your, your business plan that you've worked on and you've agreed on, you know, you say, hey, this is a good solid business plan, here's my why, here's my monetization, here's all my different things I need to focus on. Um, anyway, uh, Claude will, Claude will be able to help you. Uh, it will stay within your, your, again, your bumper lanes, right? So you don't fall off into the gutter, but you will, um, you will, you'll be very, very efficient and effective there.
So, but also Replit has no memory, correct, of past projects?
Correct. It does not know who you are. Yeah. Every time, remember I said yesterday, every time you push new chat, it's like hiring a new developer.
So that would be the biggest reason not to do.
Yeah. The memory element. And I actually just, you know, 2 minutes ago before we jumped on here, posted on LinkedIn. I said, hey, ChatGPT or Sam Altman came out yesterday and said, hey, well, we're moving on to ChatGPT-6, right? Because GPT-5 is like, it wasn't as big. I don't know. Some people liked it. Some people didn't. There was a lot of controversy about it. But the big kind of heralded new feature of GPT-6 is going to be a dramatically improved, even beyond what it is already, memory. The memory is going to be the next thing, right? Because what Stephanie needs, the type of advice Stephanie needs is different than my, than what I need, right? I need more help on my writing than Stephanie might need, and she needs more help on AI than I might need. And if it knows that, then we're going to both get better advice from our AI if it understands that context of our lives, right? Sharon, go ahead.
Thank you. Sorry, I was late. I couldn't log in for some reason. So that's partly the question I had in the sense that at the moment, the one I built previously, it uses an API to connect to my GPTs, my custom GPTs. So the reality is my custom GPTs and the solution I want, I want 10 of them that sort of daisy chain from one output to the next. You sort of like, it's like a relay race. You take that content, put it into the next one, into the next one, into the next one. So one of my questions was, should I stay doing that in ChatGPT and pushing that and connecting that with Replit, or should I be taking my prompts that are my custom GPTs and then creating that in Replit? But based on the memory issue, then that's probably not right, because the memory of chat creation and design and planning is different to the experience at the other end of the user interface, I believe.
So yeah, can you help me? Did anybody not understand what Sharon just said? It's okay if you say yes. I want to, I want to make sure everyone is up to speed. Does everybody understand her question really well? Okay, Stephanie, thank you for, for shaking your head. I just want to be very clear here. So Sharon, we'll work on this today. We're going to build some custom GPTs and you guys are going to understand how you can do your own GPT, how you can build prompts and how you can do all this stuff so that you're building something maybe for your own business or maybe you want to create a, you know, Stephanie knows chatbot, you know, or a Jeff, you know, the painter, you know, AI, whatever. You can do whatever you want and you can build these custom things and they become really powerful and, and can almost take on a fun personality. Sharon, I'm guessing you created those, those GPTs in ChatGPT, correct? All these custom ones? Yeah. So one of the things that's really important to understand is you actually can't API into those. They're only accessible inside of ChatGPT. You can rebuild them.
I did do it with the OpenAI though. I did it on an AI PR, AI, whatever that acronym is.
API.
Allow me to do it. So I did do it and then it broke.
Yeah, you can, it might, you might have thought you did it, but you can't actually API into those GPTs. You can API into OpenAI's API. You can certainly do that, right? I can show you how to do it. What you're talking about is 100% possible, right? We just need to change the structure a little bit. Okay. And we're going to go over that today. And I will use your specific example as we go through it. So you're going to get exactly what you're looking for there. So I'm excited. I'm excited for you on that. But, but the bottom line is, guys, there are— and you know what, we're going to skip ahead maybe just a little bit, but, but I'm just going to show this really quick. If you go to OpenAI API, right, you click on the API platform. Okay, what you're going to see here is if I click login, there's a ChatGPT login and there's an API platform login. Okay, and so Sharon's using this, Mitesh already used this, and that's awesome. But they are two distinct products. They're both made by OpenAI, but they're two distinct products. And actually, ChatGPT— I hope you understand this— ChatGPT is simply an interface built on top of their own API. Okay, so ChatGPT has an API, and an API again has all this capability, right? And in this case, it's an LLM capability, a large language model, right? It's GPT-5 or Mini or Nano. It's one of these. You're connecting to it, and then you're building functionality on top of it. So ChatGPT has now built this beautiful interface, right, that has all of these cool things here. Right? It's got all these custom GPTs. It has a project functionality. It has all of your past conversations. It has, it has your memory that we talked about right here in personalization and all your, your memories. It's built all this stuff, right? The API platform has access to none of this, right? You're pointing directly when you're using their API platform, you're pointing directly at the LLM. You're not pointing at ChatGPT. ChatGPT itself does not have an API. It's just an interface, right? So, Jon, just— Yeah, go ahead, Sharon.
So does that mean— sorry, so that's really important because that's a really important thing I've just learned there is that when I'm creating my configurations for these GPTs, I need to do it in a vanilla way that's not me because it's got my memory mixed into that, my behavior and my knowledge. So whilst that's helpful to a point, from what I'm trying to create as a product service solution, I don't want it referencing my behaviors and me.
Well, no, what I'm saying is the API, the OpenAI API does not have access to your memories, to your custom GPTs or any of those things. What you have to do is just rebuild those.
That's what I'm saying. I need to create a clear memory custom GPT that's not associated with my memory within my— I know you can't do it in OpenAI, I know that's just the backend tool of feeding all those other GPTs and everything else. But what I don't, I don't want to clear my GPT, my ChatGPT, too many bloody GPTs. I don't want to clear my own experiences with ChatGPT.
You don't need to do that.
But I also recognize if I'm doing configuration prompts for my own custom GPTs, I don't want ChatGPT to be mixing up me in that.
Yeah, and it won't. It doesn't have access to it. That's what I'm saying. What you're going to do, what we'll show you how to do later, is you're actually going to store your prompts and your instructions for your custom GPTs in your own Replit database. You're going to build those prompts in the Replit database and then you're going to say, okay, we're going to now build an interface. Here's your prompt. Here's the LLM, get to work, right? And take this output, put it into this input, take this output, put it into this. And so as it goes through your, your daisy chain of prompt question-answer things, right, all of that information is going to be stored in your database. Because ChatGPT doesn't— or, you know, OpenAI doesn't have a database. It's just a language processor. It's just processing for you. You're, you're attaching yourself to the calculator.
All right, guys, can I ask one more question?
Yeah, just while we're on it.
Sorry. So because one of my queries was always that, does my audience intended use end user have to have a ChatGPT account or a Pro account where they can use my custom GPTs to the extent that I want them to, or they might want to? And what I want, still unsure on, which is why I went down the OpenAI route and trying to connect something that then isn't connectable. Was because I wanted people to be able to use Replit on my cost or on their subscription charge within that rather than, which is how I even came across Replit, if that makes sense. So am I correct on that? They're not, they don't need an account with ChatGPT to use my custom GPTs within Replit.
Yep. Let's share, let's share screen here. You know what, guys, since we're talking about this so much already, why don't we just do this right now? Normally I do it a little bit later, but it's okay. We're not, this is not super advanced. I think you guys can handle it just fine. So I'm going to open up ChatGPT here on the browser. Okay. And, um, the— so, so one of the cool things here, inside of here you got new chat, search, jazz library, we've got all kinds of cool stuff in here. There's this one section here called GPTs, right? And GPTs are basically, let's say, fine-tuned, uh, prompts. That's really all it is, right? It's just a fine-tuned prompt. This one is ready to help you write better. Okay, awesome. And someone named puzzle.today created that, and that's one of the top picks. This is one of the ones that's better out there. This one will help you research better. This one will help you with your astrology, right? This will help with your fitness. And it's just tweaked. It has some more information. I'll show you how, you know, what kind of extra information it has, but it's just tweaked and pointed at, okay, your job isn't to use, you know, you have access to all knowledge of all humanity from, you know, 10,000 years ago to now, right? That's what, that's what ChatGPT has access to. And so sometimes when you ask it a question, you're like, okay, well, this is too much. What if we just narrow it down and we say, let's focus here on just this information. Maybe it can process a little bit better. Okay? So lots of cool GPTs, go and explore them. I think that's really fun. Where it gets really fun is where you start to create your own. Up here on the right, you see My GPTs and Create. Right. So I click create right here and now I can talk to it and it can kind of walk me through or I can just manually configure it. And I like to manually configure it because it helps me understand how it works. Okay. So first of all, we're going to name it. So let's say I'm going to build a kayaking one. Okay. We're going to say, yeah, Kayak, Kayak Master. Okay. That's going to be its name. I'm sure I can come up with a better name than that, but doesn't matter. You are a— let's see. Hold on. Short description. Okay. I help you become a better kayaker. Okay. So now it's kind of giving us a preview of what our product is gonna be. I can publish this to the public, by the way, after I'm done.
Right?
If I think a lot of people are gonna use it or I build a really good prompt, maybe you end up in one of the, one of the featured things. Okay. Then we're gonna come back to the instructions. This is the prompt. This is the most important part. We're gonna have a couple conversation starters. I want to learn how to kayak. And I want to learn my role and how can I progress to Class 5 whitewater? OK, so that those are a few conversation starters that are just kind of, you know, pre-canned. Hey, this is a good starting place. A lot of people want to learn their role. A lot of people just maybe they want to start and maybe, hey, how do I really push to that next level? OK, so then down here, you know, you've got some model stuff, you have some capabilities. Can it search the web? Yes. Can it use Canvas? That tool. Yeah, that's great. Can it generate images? Sure, why not? Okay, can it do code interpreter? And we don't need that for the kayaker, so we won't give it that. And then we give it some other actions, but we're not going to get into that today. Then I have this knowledge base here, right? The knowledge base is okay, what is this extra information that we might add on top of ChatGPT's model? Maybe I have a bunch of, you know, guides or PDFs or things that have a ton of information on how I do this well, right? Or maybe I'm a therapist. And I've, you know, I've written a book and I want this GPT to answer questions like I would, right? And so I upload my book to its knowledge. I just click upload files and upload a PDF. Simple as that, right? And so I upload a few things and now all of a sudden here's all the kayak thing. Then the prompt is, okay, what does this GPT do? How does it behave? What should it avoid doing? Now, guys, I use— when I'm building a good prompt, I actually will use ChatGPT, I'll use an AI to help me build out a nice solid prompt right now. I could say, you know, your job is to guide people to become better kayakers. You are always encouraging, never negative, and always pushing them to be better. You know, refer to the styles of Dane Jackson and Rafa Ortiz, um, and all of their unique approaches to reading. Okay, whatever. So I can go on and go on and go on and go on. A good prompt, in my opinion, is a couple pages long. Okay, a lot of people would stop right here, right? Or they would stop right here. I can't tell you how many people make a ChatGPT that literally is this long, right? And that's not very effective, right? You're leaving too much, you know, it's going to make up too many things. Now, it might be, it might not make it up in terms of it being false information, but it's going to be filling in so many gaps because you haven't really instructed it, right? Some prompts are 15 pages long. I'm not kidding. They're really, really detailed. They say, if you run into this situation, do this. Like, think about this. Let's say we're building a a marriage therapist GPT, okay? And I'm a therapist. I've studied for 20 years and practiced for, for just as long. And, and I, um, you know, I have a very unique approach to solving marital problems. And there's different types of marital problems, right? Maybe, you know, the husband cheated on his wife, okay? And now I'm talking to the wife, okay? Well, I'm going to approach that very differently than if the husband came and he feels like he's not providing enough, you know, money for the family, and he feels guilty that his wife has to work also. I mean, that, that could be, you know, those are two very different marital problems, right? And you don't want to just have the same generic approach to it, right? And so maybe you have a very specific way to address these types of things. So it could say in the prompt, if the user asks about, um,, you know, providing, you know, for your family and feeling like they're not doing well enough. This is the approach and the line of questioning that I want you to go through. Here's how we're going to work through that and help them, you know, build them up and, and, you know, show them that they're doing their best and they're doing awesome, whatever. Right. I'm not a therapist, so don't listen to me. But, and then, oh, hey, if there is a fidelity problem and if it's, if it, you know, if you're talking to the wife and it was the husband that cheated, here's how you're going to do it. If you're talking to the wife and she's the one that cheated, here's how you can do it, you know. Whatever, right? You're going to have very different approaches for each of those scenarios, right? And so you can provide that very explicit if this, then that instructions to your prompt, and then it's going to fill in the gaps. It's going to make it all feel really nice. And you can say, use this language, don't use this language, never suggest this, and whatever, right? If you, you know, if they say that they're suicidal, immediately try to transfer them to somebody, get them to call a hotline. Whatever. Like, there's so many different scenarios where, where you could, you could do that. Now, you don't have to be exhaustive, right? You probably want to just hit high levels and things that are going to make it feel really nice. But, but anyway, that's what the prompt is for. But to kind of bring this back down, all GPTs, all custom GPTs have these elements: a name, probably, sometimes a description. They all have a prompt, some have conversation starters, and they all have knowledge base, and they all have a recommended or a model that they use, right? And so, you know, GPT-4 or whatever, it's just gonna default to whatever it is they're doing. But now that you know how to build a GPT, now you know how to tell Replit how to build one, right? Because you can say, okay, let's set up a prompt. Let's set up a GPT that has its own prompt. And, and here's the knowledge that I want it to refer to. And here's this. Let's go to, yeah, we'll do that. We'll do that a little bit later. We're gonna do some audio stuff later. It'll be really fun. Okay. So a couple questions here. Let me, let me make sure. So Lisanne, is what Sharon talking about using an N8N workflow to connect tasks? It's not, Lisanne, although you certainly could. You certainly could use N8N. For those of you that don't know what N8N is, we'll just pull it up really quick. N8n is an automation workflow tool, and there's all kinds of things you can do, and it's very easy to kind of visualize what you're doing. I would say N8n is a little harder to use than Replit if you're talking about difficulty scale and needing, you know, how well you are, how good you are at dealing with code, right? Even if you're not writing the code, but understanding like what it's asking for even, right? Like, hey, we need some JSON input here, and we need this, we need structured data on this, you know. It might ask for things you're like, I just don't know what that means. Now, can you— could you guys figure it out with AI, with the help of AI guiding you through it? 100%. You certainly could. One that's a little bit easier that you might consider, I don't make money off of any of these, right? This is just me suggesting, but make.com is, I think, in my opinion, a little bit easier to do. Let's see here. I want just an example. I want just like a product overview. Yeah, anyway, it's, it's, yeah, here, here's kind of maybe a, an example, but you visually orchestrate automation. That's what it's saying here, right? That's, that's what you're, that's what you're doing. So you're, you just say, hey, you start this and then it does this and then it goes here and then here and then here and then here. And so could, um, Sharon, could you use this? 100%. And it might, it might be the right way, it might be easier actually, to just use a prebuilt system like this. You just say, okay, one input, then an output, then the output, then the output, then the output, then the output, then the output, then the output, and then send it back over wherever. Right? And by the way, you can create one of these, one of these workflows, and you can, you can then just use, you know, the system, the API that it has and put it on your Replit site. Right? So it's not like you're using one or the other. You would just potentially use this to create that chain and that automation. And it'd be very visually easy to do because if you build it in Replit, what you're going to do is you're going to have to explain, here's what I need to happen, this, and then you're going to have your structured output and all these different things. And you're not going to be able to visualize it like this. Replit can't show you. It can just say, yep, here's, you know, you might be able to ask it in the—
you could create it in this and put it in and put that in Replit though, couldn't you?
Absolutely. That's what I'm saying. Yeah, that might be the way to do it. So that's a great question, Lisanne, and a great tool to bring up. But n8n and Make are 2 tools that I think we're not going to go over them in this course. We're not going to do anything in them. They are more advanced and we just don't have time. We could do a whole 8-hour course on n8n easily, and we're just not going to do it today. Okay. Lisanne, you asked a question, what's the difference between a chat, a project, and a custom GPT? Great question. So a chat is just going to be your standard, hey, I'm here talking about a specific, you know, here's my, here's my River app, right? I'm just talking about this thing, although I'm having this chat inside of this project, right? So let me click on my General AI Proficiency project here. This is my, this is my main project. This is where I started, you know, months ago when I started GenAIPI. My very first prompt in this project way back here. Okay, this was my first, first prompt. This is how I started my business on February 27th. Okay, I said I want to create a website that checks the proficiency levels based on an intelligence test that shows the aptitude for an individual to succeed in a world with AI. This means they need to know how to talk to AI, how to be good prompt engineers. They need to have a good brain to figure out how complex systems can quickly be put together. Can you help me build this? That was my first idea. That's where I started. And then we said, okay, great, let's do this. That was really the first idea. That's why when you land on my— on GenAIPI today, the first thing you see is, hey, take the free AI Mastery Predictor test, right? That was it, right? Hey, we're going to measure how good you are. And then based on how good you are, if you're not great, we're going to take you to some courses. If you're great, we're going to take you to some certification. Right? Either way, you got something to buy from me, right? But the idea was, hey, let's see how good you are at AI. But inside of this project, I have, you know, my original prompt, and then I have dozens and dozens and dozens of other conversations about many, many different topics. Like, what's my short slogan for GenAIPI, right? What's this? Hey, let's, let's design a certificate here. What's this build a business guide? You know, I need to build a GPT for this, whatever. So all of these different things, right? All inside of here. This is my project. Okay, so the chat can live either on its own over here, or it can live inside of a project. If it lives in the project, then it has more context around it, right? Now it's related to all the other stuff you're talking about. Then you have a custom GPT, and a custom GPT is totally different. It's separate from these other things. You create a custom GPT, right? By here, you know what, I'm actually going to go back to the web interface for this because it's just a little bit better. But this is the custom GPT and what we're doing here is we're creating a tool. We're going to have it create an image and then we'll publish this thing. Let's just use DALL-E, it will automatically create an image here. DALL-E is the image generator by OpenAI. You can just generate images now inside of ChatGPT inside the chat. But anyway, so, so right here, we're gonna, we're gonna now have, we've got a name, we've got instructions, we've got conversation starters, we've now built a GPT. So now we will click Create. Okay, and we'll, we'll share it with anyone with the link can use it. Okay. So I could share this link with you, but, and this is going back to answer part of Sharon's question. When you build a GPT inside of ChatGPT, you can only access it as a, as a user. You can only access it if you have a ChatGPT account, right? It's inaccessible to everybody else. But if you build your own solution on your website using the API, then every time someone asks it a question, you, the API holder, are going to get charged. For it, right? And so then it becomes your— it's your account always, right? That it's getting charged. And so if they're paying you, you know, $30 a month, you know, for whatever, then— or £30 a month, then they'll, you know, that's what they're paying, and they're covering, you know, some of your charges. So welcome, Chris.
Sorry I am late.
You're good. We are building custom GPTs, talking about that. So, so anyway, guys, we've now built this, the Kayak Master, right? And so we can test it, we can use it, we can, you know, I can share it. You know, anyone can copy this link now and just go to it, right? Just paste it and hey, welcome to the Kayak Master thing, right? And so let's say, how do I— how can I progress to Class 5 whitewater, right? So that's an awesome goal. That's great to get there safely and confidently. You want to think like Dane Jackson and Rafa Ortiz, precision, creativity, and absolute respect for it. But they're two of the best kayakers in the world. Okay, so Now it's just, it's just going to start teaching me. And so now what I've done, Lisanne, is I've now started a new chat. I'm not in a project. I'm just in a new chat talking to this new GPT. So the GPT itself is the, is the tuned model, right? That's pointed at a specific task, right? You're— if you just click new chat right here, you're just opening a general model that can do everything right. And quite frankly, I could use this, this model right here. I could say, Let's pause on kayaking for a second. Teach me about the sun. Right? And it'll just start teaching me. Okay. So it still is using a general model. But I wasn't, by the way, I wasn't explicit in my instructions to tell it not to do this. I could have said, hey, you know what? Let's not let them do that, right? If the user tries to talk about something other than kayaking, guide them politely. We'll say politely guide them back to kayaking. Okay, done. Update. Okay, so now if I were to go back and try this again, All right, we're just gonna, we're gonna refresh this here. Let's, um, teach me more about the sun. I'm wondering if it's updated already, if it already is looking at it, at its prompt again. Yeah, it hasn't, it hasn't. So it's just going to start talking about, about this, and that's okay, that's okay. So let's try this again. New chat, um, teach me about the sun. So I'm curious if it's going to finish and then say, let's get back to kayaking, right? Because I wasn't very explicit. I didn't say, yeah, kayaker's connection. There we go. So it is politely connecting me back to kayaking. So it actually did listen to my prompt. I wasn't explicit enough, but it said, hey, want me to break down how the sun's energy actually creates the river systems we paddle? So it actually very intelligently pulled me back to kayaking. It did a good job of listening to me. But what— do not answer anything unrelated to kayaking. Yeah, so let's try that now. Then I could go back and do this. This is really important, guys. I know it might seem like I'm kind of wasting time on this a little bit, but I'm really not. The important thing here is that you don't put an a GPT, an AI bot, chatbot on your website that people discover they can just use for their homework, because then they're just going to eat up all of your API credits and charge your credit card a bunch working on, you know, calculus, right? Instead of going and paying for their own ChatGPT account because they discovered, oh hey, this AI is not trained to go back to its intended purpose, right? Teach me about the sun. Boom. Now it won't do it. So I've been a little bit more explicit in my prompt, in my instructions on here's how you're supposed to act. Okay, sun is definitely powerful, but let's tie it back to kayaking. When you're on the river, the sun is one of your most important factors, right? It is right there, right? Let me try to use the sun to your advantage when reading a rapid. Cool. It's actually a pretty cool topic. There's some cool stuff you can do there with light. And Reading Rapid. So anyway, I hope, I hope that clears up a little bit, guys. But, but building a GPT is not hard. Now let's actually do it. Okay, we're going to now build a GPT. Okay. And you're just going to watch me do it here. We're just going to tie an OpenAI API into my HighWater thing, right? So I'm going to create a chatbot. I'm going to ask it to do a little chatbot down here that can talk about kayaking. Now I can't tie into my custom GPT that I already created. All right. But we'll, we'll, we'll work on this. We'll get this working. Let's add a, an OpenAI-powered chatbot in the bottom right-hand corner of the website. Its purpose is to talk about kayaking, help users find the best rivers in their area according to their skill level, and to always stay focused on these kayaking-related topics. If the user tries to vary from this and use the chatbot for other purposes. Um, don't answer those questions and guide them back to the relevant topics. Okay, so I basically just said I want a chatbot that's powered by OpenAI and that does this. Now I gave it kind of a very high-level prompt And I'm going to show you how it actually structures it, right? How it actually structures it in the code. It's going to write, it's going to build exactly what we just did here. Okay. Going back to the structure here, it's going to have instructions. It could have a knowledge base. I didn't give it a knowledge base because I don't have a knowledge base of kayaking to just hand. I don't have a document to give it. But, but I certainly could. I could say, hey, you know what, refer to the knowledge base when wanting to answer, you know, when answering these questions. Okay, so let's see what it does. So first of all, first off, this is important. So we haven't shown this to a lot of you guys yet. I know Sharon's done this and Mitesh has done this. But right here it says, hey, to build the kayaking chatbot, I need an OpenAI API key. Okay, this will power the AI that helps users find rivers. An API key is a secure token that lets your app communicate with OpenAI's AI service. You can get one by going to here, create an account, create a new API key. Super easy, right? You just click, you create a new account, and you know, you have to log into this, right? So I'm going to go over actually to this other browser because I have access here. OpenAI API here, and then I'll log into the API platform. Please. Continue with Google. By the way, this is a totally separate account. If you already have a ChatGPT account, you're going to have to sign up for a new account. It is a totally separate system. Okay, remember, ChatGPT is an interface built on top of OpenAI's API platform. They're just using their own product, right? And so you can start building, you can grab an API key, and, and you just kind of dive in. Now you click here on your settings, you should be able to grab your API keys. Your API keys there's a, there's a security feature. I couldn't even show these to you if I wanted to. They're impossible for me to see. Okay. And so I can just click create new secret key and it's going to be a long string of just random letters and numbers. All right. But I can create a new secret key very easily and just say kayaking. Whoops. Kayaking. And then project will just default project. I can obviously, you know, do that, create secret key, and then it gives me a key. Right. There's the long key. I'm actually not going to use this one. I'm going to use a different one. And it's again, super long and there's no way I'm going to be able to remember that or see it all. But as soon as I'm done with it, as soon as I click done, I will never be able to see it again. Now, no big deal if I forgot it or I lost it, just create a new one. You can have unlimited, right? So I'm going to finish this one. I'm not going to use— I'm just going to delete it. So we'll just revoke the key because I already have one of the cool things that Replit does is it helps you save Your keys, right? And so it had— this is— we're gonna introduce a new feature here in Replit called Secrets. Okay? So here's Secrets. Your secrets are your API keys. And I have a ton of account secrets. Account secrets mean— so this is my app, what I'm building now, my Riverflow, my HighWater app. My account secrets, because I've built dozens and dozens of apps, I have tons of API keys. I've got Cloudinary and Bright Data and ElevenLabs and Mapbox and OpenAI, OpenWeather, lots of stuff. It's already there right now. If I want to do a, a, a kayaking specific and I want to separate my API usage, that would probably be a best practice to just do different things. But since I'm just doing this as a demo, I'm not going to worry about that. I'm just going to use my OpenAI API key. It already knows it. If I didn't know it, this would just be empty. I would copy it from OpenAI, paste it here, and then just click continue. Then it would come up and say, hey, do you want to save this to your account? Just click yes. I've already done that, so you didn't see that little prompt there. But that's it. Now it's added this API key, this OpenAI API key from my account secrets. It copied it over and now there's a copy of it in my app, in this specific app. Now it's going to create the Kayaking chatbot. It'll build a little bubble here or something. I'm guessing it's not going to work perfectly the first time I test it. We'll go through and we want to tweak it and make sure it's a smooth thing. Show some animations maybe so it knows that it's answering. There's a lot of just little different things you'll want to do. It might show it in the wrong format or show a markdown format or something like that that just feels weird. But this is really all it takes to add AI to your app. You just tell it, hey, I want to add a chatbot that's powered by OpenAI. Here's what it should talk about. Boom. Okay, that's all it takes. All right, and then literally follow the instructions. Copy the thing, paste it over here. Okay, and so now it's creating the routes and all the different things it needs to do in order to make this chatbot work. And, and by the way, I also could give it additional instructions to say, hey, um, the chatbot only works if you're signed in, if you have an account, right? So I could do that too. Right? So you can also, when you're, when you're building, you can say, hey, I only want this to happen if some condition is met, right? For example, another, another common use case for this type of a functionality is an admin functionality, right? So here, if I go to GenAI API, I see enterprise solutions, live courses, courses, certification, about, whatever. But then if I log in, I'm an admin, and because I'm an admin, I get a little admin button, right? No one else has this. If you go and log into your GenAIPI account, you won't have this. You'll have dashboard and you'll be able to go down and see all your stuff and that's cool, but you don't have this admin functionality. And here in admin, I can manage users. I can manage course content, live courses, test content, test results. I can look up, you know, a past result of a test that I took or that another user took. And here's my certificate. I can download that PDF for another user on behalf of them, right? I just built out these tools that are useful for me, right? I just built out this new referral system. It's really cool. I haven't actually finished building out this, but the referral codes actually work. If I build a referral code and say, you know, $100 off or $100 off, right, then I could say this is $100 off unlimited. It's used— you can use it one time. Expires on the 30th, status active, create code, boom. All of a sudden I've got a, I've got a referral code and, or a discount code, right? I built a discount code system and then I should have a list here and I just, when I built it, I didn't have time to go back and do that. So this is just kind of a pending feature that doesn't matter for me, right? It doesn't matter for us, right? It doesn't matter that much. It's, it's kind of important, but it's not that important because, um, I'm the only one that sees it. Right? The only thing that really matters is that when I put that referral code in, right, um, if I go to, if I go to buy something, um, let's see here, let's log back in. Uh, let's see if that's just like another random user. Yeah, yeah. So if I come in here and I go to buy something, uh, reserve your spot, register now, and I type 100 off, apply code. Oh, I didn't save it. I didn't save it. That's okay. I think there's another one. I don't know. 50 off, is that one? Buy code. I actually think I made another one for somebody. What was it? What was his name? I don't remember. Anyway, I know it works, but I can put that there and the price will change. I don't remember any of my codes. I'd have to log back out and log back in. I can actually go in and check my database to check those different things. Now that this is done here, let's look at it. Let's watch it again. Let's play with this piece here. So hey there, I'm your kayaking guide. I can help you find the perfect rivers for your skill level, suggest gear, answer any paddling questions. What's your kayaking experience like? I'm— or I live in West Mountain, Utah, and I'm looking for some Class 2-3 whitewater. Where can I go? So let's try this. Let's see. Let's just first see if it works. Okay. All right, so it, it kind of did something for a second and then it stopped. Okay, so guys, this is, this is kind of the next part of Replit, right? Some of you have already experienced this and tried to kind of push a little bit farther. But, but this is what you do when you're, when you're kind of creating new things. We've created kind of like if the entire experience is click, open, ask a question, get an answer, go back and forth. We're about half of the way there now. I was able to click open, ask a question. Now we need to actually make sure the answer displays. I'll just say this. I just asked the chatbot a question and it seemed to think for a second and then displayed this timestamp as the answer without the actual answer. Let's get that— let's get that piece working. Okay. So it's gonna work on that for a minute. While we're waiting here, let's see if I can find this. Apps index, or maybe— Do, do, do, do, do. What did it build here? Installed OpenAI. Yep. 17 more. Here we go. Okay. Rivers chatbot. Okay, here we go. Source components, chatbot. Let's look at the chatbot code here. Okay. So, all right. So this is super cool, guys. So basically all I did was I, I tried to identify, I could have asked the agent, hey, where is the chatbot prompt, right? But right here in chatbot.tsx, it's just this little component that it's built. Okay, I can see, hey, this is what's happening. Content. Hey there. Right. So it's going to start. Hey, I'm your kayaking guide. So I can edit this. I could change this right here. Now, I don't have to edit this here if I'm uncomfortable editing code. Totally fine. Maybe you accidentally delete a comma or something and you break it. So maybe you don't want to mess with that. Okay, so you can modify it with the assistant or something, or you can just tell your agent, hey, I want you to update the message. I want you to update the message to be something different. Now it has a few other things. Sorry, I didn't receive a proper response. Let's see here. Let's see if it has anything else here about the actual prompt. Yeah, I'm not seeing the actual prompt here, which is okay. I want to see if it has any instructions. It's going to clean this up. Found the issue. Function returns a response object. That's fine. Perfect. I found and fixed the bug. Let's clean up debug. Now works. It should work. Great. Awesome. Let's try it again. We're going to copy the same question. Whatever. We'll just ask it here. Go. We ran into an error. No big deal. We can look potentially what that error is. You've got DevTools, this little thing right here. We have a chat error right here, error 500. Super easy. You could just say this, this is the error that came up this time, check the console error. Console error, you just tell it. It can access that, it can see this error here, so you don't have to even copy paste, you just be like, check the console error. Sometimes I'll copy it, whatever, and just be like, hey, what's going on here. Now, I'm actually building this in a way that I probably, I probably would be a little bit more specific about doing some of these things. I'm kind of just letting it go through the process in the way that I built this the first time to kind of walk through the way that you guys might do this. Like, hey, build me a chatbot. Hey, I got an error. Hey, this didn't work, right? And so you just keep working through those things just like this. And I think this is probably the best way to do it. Notice I also did a couple things there, guys. I did a screenshot, right? It can see screenshots if you just, if you just screenshot it on, on Mac, your Shift+Command+4, right? Just do that and you can select a piece of the screen and then you just drag it in. Super easy, right? Um, you can remove that later. But anyway, so now let's try it again. Let's see if we get this same question. Let's go. Thinking. Okay, so we're having another error again. I want to see if it's the same error. Yeah, so, all right, so here, so now, now that we've run into the same error a couple times, guys, I'm gonna do something a little bit different this time. Okay, I got the same error. Um, tell me what is happening and why this isn't working. Working. Don't write any code, just tell me. Okay, so now I'm asking it to really think. And explain back to me. If you just tell it a problem, it's going to fix it. Unless you say don't write any code, just tell me. That is a trick I use all the time. Now it's going to sit here and think for a second, looking at the error logs, here's what's happening, here's the problem, here's why it didn't work. Now guys, I don't know how all these things work and fit together, but I am capable of reading and understanding that something's going on, right? So, all right, can you tell me specifically what you can do to fix it? Don't write any code. Just tell me. The, the Replit agent is trained to be an overeager, you know, action-oriented worker. Which is great. It's great that it does that. But if you tell it to just slow down, now it wants to add detailed logging, log the incoming data, fix approaches, add array filtering, frontend message cleanup, fallback content. Basically saying most effective approach is to do the logging first. Great. Add the logging and let's figure this out. I basically said, okay, let's do your plan. There's a couple other things to consider here, guys. Right now, if I click on my database, remember, my database is, is what allows me to have persistence of data, right, to go between sessions, right? There is nothing here that is related to user conversations right now. Obviously, if you're not logged in, you can't have a user conversation. But at some point, if you feel it's useful, it might not be useful in this case. It might not be that important. But if it's useful, you could say, hey, I need you to log the conversations in the database. I need you to create a table for the database. Sharon, I think probably you're going to want this, right? You're going to want to know what the users have asked, you know, the different chatbots and this whole chain of things, right? Because especially if you're trying to daisy chain them together, right? And again, if you're using N8n, you might not have access to all that data. It might not send it back in the proper format.. So you'll have to consider that. But if you're doing it inside of Replit and you're building your own little daisy chain of, okay, input, output, input, output, input, output, it might be easiest for it to just save it, save it, analyze it, save, analyze, save, analyze, save, analyze, save, analyze. Here's the final output, right? And to have it store it in the database there. All right, so let's ask this one more time here. Whoops, that's not that. Let's ask this question. Copy. You know what, we're actually going to refresh this just in case somehow that was some instance that wasn't resetting all the way. Hey, that worked. It might have been— that might have been my problem. So that's my fault. I might have— it might have been working properly, but I was using the old code. Anyway, West Mountain, Utah is a great spot to access some fantastic rivers. First of all, no, it's not. But anyway, you might want to check out the Provo River. Great. Another is the Weber, Ogden. Yeah. Why didn't you suggest the Spanish Fork River? I'm just curious what it says here. Okay, so it's working now. The chatbot is working now. I want you to do It's kind of working. It's thinking really hard here. Chatbot is working out. Um, will you show me the prompt? Yeah, it worked. Spanish Fork River does have some sections, but it often depends on the flow conditions. That's commonly recommended for consistent Class 2 to 3. Yeah, that's probably right. Chatbot's working out. Will you show me the prompt, um, you are using for the chatbot. So, Mitesh, you're asking if the chat is connected to OpenAI API. Yes, that's what's happening there. That's—
this is— I meant it's not answering from Claude, it's answering from OpenAI.
Definitely. Yeah, it's using— we put, you know, using the secrets, we've put that OpenAI API key. And that chatbot is using that API key to answer questions. Yeah. All right. And then let's say, can you get a specific API for a custom GPT so it's not just accessing OpenAI? Of course. I mean, if you want a Claude one, then the way that you would do that, guys, is you don't remember Claude is inside of Replit, but you can't connect to the Claude API here. You're going to have to go and you're going to have to log into your own, you know, Claude API, right? And, you know, build with Claude right here, Anthropic API, you know, start building, continue. I swear I have— I swear I have an account here, but maybe I don't. Maybe I haven't done that with Anthropic. With this is my FlowState Leadership one. But anyway, it's fine anyway. But this is— this is what I would do. FlowState. Good. Continue. Individual. It's fine. I could have done organization, I guess, but whatever. But yeah, if you wanted Claude to be the one answering, or maybe you want Perplexity to be the one answering, or maybe you want Grok to be the one answering, you're just going to go to their system and get their API key, right? Create a key. There it is. I can grab that key and install it right now. Now, it's not going to work yet because I haven't added a credit card yet. It does cost money to use these APIs, but I want to talk about that for just a second here. The APIs, guys, are insanely cheap. Okay. The amount of money you're going to be spending on these, we're going to go OpenAI API pricing and we'll look down at this. So, so there's a bunch of different types of models you can do here, but you have your input cached input, which is like part of that conversation, right? And then you have output. Okay. Now GPT-5 is basically their biggest, most expensive model right now. It's already only $1.25 per million tokens. A token is roughly a word. It's a little— it's more like 0.8 words, right? But think about how much talking back and forth you have to do with an AI to get through a million words, right? Or even 800,000 words, if that's what it was closer to. Insanely cheap. And by the way, you probably don't need to talk to GPT-5. You could probably use GPT-4.1 or 4.1 Mini or heck, GPT-5 Mini. Is $0.25, guys. That's crazy. That is seriously nuts. They are being insanely aggressive. It's cheaper than, than these. It's cheaper than GPT-4.1 or 4. Like, look, only 4.1 Nano is cheaper than that. That is crazy. And GPT-5 Mini is more than capable of doing everything that you guys are trying to use it for, unless you're doing really high-level, you know, research or math or like like crazy, like PhD level beyond math, you don't need to go beyond GPT-5 Mini. This, this model would be perfectly acceptable. And then if you want to use, you know, this, let's just click continue here. Manage API keys, billing. Yeah. Okay. Hey, gave me a $5 balance just to start here, at least that I can use for a year. Free credits. Cool. But let's see here. I need, I need pricing. Show me pricing. Cloud API pricing, or it's actually Anthropic API. But anyway, um, no, I don't want that one. I want that one. Okay, so here's Claude, right? If you're using Claude Opus, $15 per million. If you're using 4— 4, by the way, is what, um, is what, uh, Replit is using, although it is using I wanna say a mini version of it or something like that. Maybe it's Sonnet 4. I think it might actually be Sonnet 4. And then you can use Opus 4. So Opus 4 is the best one. I guess they have 4.1 now in here in, in Replit. We're getting ahead of ourselves just a tiny bit. Maybe this is kind of right on schedule. This is fine. So agent tools here. If you want, if you have like a really complex problem that's just not fixing itself, you can go dynamic intelligence and you can click Higher powered model, Claude Opus 4.1, right? If you do this and you start talking to it, just be ready to pay money. You're going to be spending $100 an hour. Okay. It's very pricey. It's 5 to 10 times more expensive depending on what it's doing, because it's going to— you can do extended thinking, you can do higher power model. I rarely use it. I will only use it sometimes to maybe make a plan that's complex, right? Where I do one of these ones where I'm like, hey, don't write any code, just tell me your plan. And then I will turn off HighPowerModel and I'll just let 3.7 Sonnet take over— or sorry, Sonnet 4 right here take over because it's going to be a fifth the price, right? So bottom line is, guys, all of these different systems have their own API. Rock API. Okay. There it is. Okay. Model that fits your needs. Ready for your business. Quick start models and pricing. Here we go. Grok 4, $3, right? Output, $15. So it's a little bit more for outputs and less for text inputs. You see that across all of those different models. But depending on who you want answering your question because of whatever reason, you can do this. Now I have a system. I have one app, CougarIQ.com. I built this with Replit kind of as a— I was kind of for fun and, and I didn't, I didn't perfect it at all. There's a lot of things that I did not do well on it, right? But I built this Cosmo AI and I was like, you know, tell me about the schedule or the, the football schedule for 2025. Okay, so checking ESPN for the latest data, searching the web. So that's using an ESPN API. Then it's searching, you know, the, the web anyway. So, uh, I can't give you the official schedule. Why not give me the schedule? It should be able to do that. I don't know why it's not working, but theoretically it's supposed to be searching Perplexity. Yeah, I don't know why it's not doing that. I'll, I'll work on that later. I'm going to show some more things on this later. But, but what I did for this, I swear I had it working. Maybe it just broke. Maybe something changed. I haven't pulled it up for a month. But what it's supposed to do is check the API or the ESPN API first. If it finds the answer, great. If it can't, then it moves to its second task. Now I'm going to look at Perplexity because Perplexity can search the internet well, right? If it can't find the answer there, then at last it will go to ChatGPT and it will search and find what it needs to do and then it will display. Then it's supposed to analyze and display what it's doing here. Somehow it's running into something and it's not, hitting all of those steps. So I need to go through and check on that. We'll look at that, that side a little bit more later. Okay. All right. Good questions about the APIs and the pricing. Sharon, if you daisy chain GPTs, you need a secret key for each API? No, you do not. The secret key is just going to reuse the same one each time, right? It's just going to make that happen. Okay. Is it sensible to instruct the agent to determine the model used based on cost or cost based on use? I would not. I would not tell it to to change your models. It can. It most certainly is capable of saying, all right, I'm going to use Mini for this and Nano for this and this for that. It's too complex. Just install Mini and roll with it. That's going to be your best bet for that, for what you're doing there. Yeah. So how much— how does it know which GPT to use? You'll tell it. You'll tell it. It's part of the prompt. It's part of the setup here. Okay. So, so we'll ask it that question too, Sharon. So here we are. You are a helpful kayaking and river guide. Assistant. And by the way, it did this in routes. So this was actually put in the routes, uh, right here. Now routes is really long or whatever, but it told me where it is now, and I'm sure we're going to find it towards the bottom here. There it is. So this is the actual code that— and I want— I hope this— I hope this makes sense, guys. This is the actual code that is showing the prompt. That's the same. I'm going to bring this back. That is the same as when we were building this. This is the instructions. This is the prompt. Okay. What we're looking at over here is the exact same. You've just built your own. You gave it a name, right? Kayak. Kayaking guide. Ask me about rivers and paddling. There's the description, right? And here's the prompt behind the scenes. Only discuss topics related to kayaking. If asked about unrelated topics, politely redirect. Do not provide information about politics, personal advice outside of paddling, coding. It's weird. We're going to edit that. And, you know, or other non-kayaking topics. Always be enthusiastic. That's so funny that it said coding in there. I have no idea why I did that. You gotta check your work, guys. This is part of the, part of the game right now. I doubt there'd be a lot of coders trying to come here and I don't, I, that may, that problem may never rear its head, but that's an easy fix. I'm just gonna delete that. And you can, you can edit code. Be careful in here, guys. If I delete all this stuff, it's not gonna work anymore. Okay? Keep responses concise, 2 to 3 sentences max, and practical. Use a friendly, encouraging tone like a fellow paddler. Anyway, so that's the prompt. That's how you build a custom GPT. So guys, this was a little bit ahead. Of when we normally do this. I normally do this in about an hour from now, maybe even 2 hours from now. But any more questions on how to build this? Do you feel like you could do this inside of your own app?
Go ahead, Sharon. I put it in chat, but it was some— so what I can't work out in my head is I've got 10 GPTs that are part of a process. So I appreciate it's one API to access, those GPTs, but how do I make Replit know when one finishes and they're taking the output of that one into the next one if it's in the same interface?
Yeah. Chris, you listening to this?
Chris might not be.
Yeah, sorry, what's up? You're good. So I want to ask you this question because I know how I would answer it, but you're a developer that's much more experienced than I am. So Sharon, wants to take a daisy chain, basically create a daisy chain of 10 GPTs. So you ask it one question and then an output happens and then that, and then the system automatically takes the output and then uses that as the input for another GPT and then an output, input, output, input, output, input, output all the way until, okay, now we've got, now we have this question that's been asked and it's been gone through this kind of filter of processes and to come out. What is the best way from a code perspective to set up a sequence like that that sits and waits and waits for the answer and then pulls that in. I'm just curious what your answer is, how you would build that in your head.
Yeah. So that's like called, well, assistant chaining, which we actually use it in Minify. Cool. And the best way to do that is to basically have like a relay assistant, an assistant that knows about all the other ones. And then it just Like, based on the context, based on like what the— I guess if you have different assistants that do different things, is that really the use case? I mean, it really depends on the use case. Like, are you trying to train different assistants on different—
Yeah, effectively, each assistant is considering an aspect of something contextually. And then ideally is bringing those together as it joins and adds another dimension of another agent.
Another— Can I ask a specific question, a follow-up to that? Are all of your GPTs analyzing the same data, or are they actually supposed to be doing, like I'm saying, input/output, and then they measure, they change the— or they analyze the new—
Well, we're trying to build a context. I'm trying to build a context flow, if that's even the right phrase, where It's a career guidance tool and it's a career clarity. So it's around understanding motivations and different elements. And so whilst they stand alone in their own right in terms of a conversation, they actually collectively— I want them to have a collective output at the end of it, considering all these aspects.
OK, so I think more clearly then, Chris, it's not a daisy chain of them. Like one after another. It's we get one input to 10 different GPTs. They all put their 10 outputs out and then a final GPT asks the question or, you know, looks at everything and then answers the question in, you know, a holistic way.
But each one of those GPTs is a process that we go through so that, but then they're building up iteratively. It's like adding more spinning plates.
Well, so you gotta kind of— I mean, you gotta kind of understand the way that assistants, uh, work. Like, assistants have instructions, and instructions allow them to do a certain thing, but they're all trained on kind of the same base information, which means they don't like you can give them custom instructions, but they're all going to be capable of the same thing. And then you use an assistant to do, to perform what's called a run. And the way a run works is it takes that assistant's knowledge base and it performs an action, right? So you could have the same assistant like perform a run, um, like you could have 10 runs running at the same time on the same assistant, if that makes sense. So like, if you want really specific training, the other thing is like you can provide an assistant contextual training, which means one assistant can know a base set of instructions, but on each individual run you can give that assistant like different sets of instructions so that single assistant then becomes, you know, 10 different assistants processing different data in different ways, just slightly different.
So, and then, Chris, could you— Chris, could you then set up the way that I do it in my mind? Now let me actually skip to the end and then go back and ask this question. Sharon, the right way to, to structure this is to explain exactly what you're saying here, either to Replit or to ChatGPT and talk through it until you have a nice solid plan and then say, okay, let's take this plan and let's execute it in Replit and Replit can then build it out. So that's that. But, but Chris, what I was thinking is one way to do this would be, let's say you have 10 and they're maybe they're all running concurrently. So that's fine. Super easy to do that. But then could you just set up like an oracle or something that's just watching for completion? And as soon as it sees completion of each of those processes, then it takes all of that and drops it into a new one, does the final review and does its final output.
Yeah, I mean, but the, the, I guess the key is whether you're using like 10 different assistants or one assistant, the output, the result is going to be the same essentially.
Cost is going to be different though, Chris, right?
What? Not really. I mean, the cost would probably—
it's the same amount of text. Yeah.
All the costs— if they're all using it, does it only— do they reference it from one place, then they don't— it doesn't get dropped into each of them, and then they're charging you for every time each one is processing the same stuff.
Well, yeah. So I think part of— let me, let me maybe clarify a little bit, guys, and I think this is important, and then I wanna, I wanna move on from this pretty quick here so we can go on to the next stuff. But There's— AI is more than capable, Sharon, of— you could just ask it one question, or you could ask one assistant 10 questions and then have it respond all at once, right? Just do one thing. From a coding standpoint, you should do that. Like, that is easiest for sure. But you can say, I want you to consider this, then I want you to consider this, then I want you to consider this and answer each of these little pieces, and then take all of that and summarize. I mean, you can, you can have one, one assistant is way more than capable of, of handling that. It really is. If you, if you do, though, have very specific lenses through which, like, like let's say you have a 5-page prompt for each one of these different types of questions. Okay, now I understand why you're needing to run separate things. But if their question or their prompt is a paragraph, well then just build one long prompt that's a paragraph, right? But if they're all really long, then yeah, build out these other things. Talk to ChatGPT about how you put it all together and then, and then do it. But, but you, you don't have to run them in sequence, right? You don't have to sit and wait till the AI is done thinking, right?
Imagine I have— I think I want— I think I just want to make sure that the AI agent is clear that we need a final agreed outcome of each part of the conversation. It's like we're going through a quiz and each quiz question needs to be complete so that all of those questions can then be considered as an output. That's what I'm trying to— Yeah.
So you have—
—do the most efficient way and that you can reiterate and re-update all of that if something new is learned that's appropriate. Yeah.
Yeah. So you just need to talk to the AI about how you would structure that, right? I would just explain it, right? And by the way, guys, one thing that's really cool about AI, more information is better for the AI, right? So just turn on your— I'm not kidding— turn on your chat function here. Just go like this. Oh, okay. What we want to do next, right? And just start talking to it, right? And it's just going to record. You can see it's recording what I'm saying down here and just talk and say, okay, here's what I want to have happen. I need 10 different assistants to all be working together. They're all going to run in sequence and then at the end I need all of the outputs to be put into one new thing to analyze it and then we'll run the final assistant that's answering this question in your way, whatever your product is. I love doing that, just talking. Just talk to it, just give it data and it's going to do that. Anyway guys, that's the prompt. Let's also ask this, which model, A, uh, yeah, which, uh, OpenAI model are you using for this chatbot? I'm gonna guess 4.1 mini or 4.0 mini. I don't know, maybe it switched to 5.1 or 5.5 mini. I mean, I'm curious, curious what it says here. So this is configured somewhere in there. It's very easy for it to do that. Yeah, it's using 4.0. There you go. And it's even set, you know, max tokens and the temperature and, you know, different things like that, right? It's going to keep everything to under 200 words roughly, right? And then you can change the temperature. You can say, you know what, I want it to be more creative, or, you know, I want it to be more consistent, right? And hey, let's change the max tokens to 120. Let's keep the answers even, even shorter. You can just change this stuff by just looking at it. Just ask it the questions and it can help you and say, hey, you know what, I want this reconfigured to use GPT-5 mini. Cool. It'll do it for me. If we look at that, we go back. Sorry, I have a million things open here, but oh, for— do, do, do, do, do. There's more. There's another one. Flagship models, detailed pricing. Borrow. Yeah, so it's doing $2.50, right, which is 10 times more expensive than GPT-5 Mini. So let's use GPT-5 Mini for this one, right? I think that's definitely a good choice. So pay attention to this stuff, guys, right? If you don't set it up right, it could cost you a little bit. First time you get a bill for $80, you're like, wait, I expected that to be $8. You'll, you'll fix it, right? Perfect. Up to use GPT— what, you're using 4o Mini? This is a smaller— more— no, no, I asked you to use GPT-5 Mini. Super curious why it's, why it's disobeying me on this one. It's like, okay, it may— know that if the chatbot encounters an error when you test it, maybe because it hasn't been released yet by OpenAI. That was my last update. Do you need to update the OpenAI API? I'm curious if this works now. Let's just click refresh. Sorry, this is taking way longer than was expected. Yeah, it even searched and it knows that it's now available. It should know that it's available after doing that web search. All right, let's just ask a question. Where can I go kayaking? Close to San Antonio, Texas. Yeah, so it's having problems with the model that it's trying to set now. Um, it isn't working now. Make sure the name is set properly. Okay, anyway, um, so we don't need to work on that a lot more, but APIs, guys, are super, super powerful. I'm going to add one more really quickly here while it's doing this. I'll just say Let's also add a weather widget in the top right of the screen or of the website, just under the nav menu based on the user's current location. Now it's going to do that. The, the issue, GPT-5 models use different parameter name. I've updated the code. Max tokens to max completion tokens should work now. All right, cool. So that's cool. Hopefully that works now, but now it's going to add a little weather widget. Okay. And again, it's going to need, you know, a weather API key. And I already have one on my account. Remember my secrets over here? I've got my weather API key right here. And remember, I get a million free, you know, free requests a month on that. So there's no way I'm going to be hitting that, especially during testing. Most systems, guys, give you a free test. Or a free, like, limit because they don't want you to have to be paying a bunch of money while you're developing and trying to figure it out, right? So if you want to use like a map feature, you want to use— and some of them, they won't allow that. They'll require you to put in like a $3 or $5, you know, kind of credit, and it will take you forever to use those credits. A lot of them are like fractions of a penny, you know, per, per request that you're trying to do. So, okay, it's adding this. Create the weather component. I want to show you one other thing. Be sure to use the weather API key that I already have in my secrets account. Secrets. I'm going to click Add to Queue, but because I want it to know it right now, I can hover over this and click Send Message. I can tell it something right in the middle and say, hey, I need you to know this while you're building this. Don't use that other thing, use the one that I already have here. So now it's going to check my secrets, and you don't see the weather API key. Well, there it is. Continue. Done. Actually, I don't even know if that was the right one. Weather API key, maybe it is. OpenWeather, it used OpenWeather instead. That should still work. That should still work. Add it to navigation. Anyway, so now it's building another API and connecting that. Now we're starting to turn this into something that's useful. We haven't connected, by the way, the core functionality, the USGS, So let's look that up, right? If I go in here, USGS River Data API, APIs right here, latest continuous values, that's probably what I want. Read the docs, browse the data. Do, do, do, do, do. Yeah, let's just do that. So this is cool, guys. There's a lot of crazy stuff in here that you're like, I don't know what I need here. Okay, and so you can, you can do a couple things. One, you could just ask Replit to say, hey, you know what, I need RiverData, right? I know USGS has something, just use that one. You can also just copy this and say, great, now let's install RiverData. Great, now let's install, uh, let's go back to home, install RiverData. And up-to-date river levels using the USGS API. Now here's some— something that I want you guys to think about. In order for— when I click Browse Rivers, in order for some data to actually show up there— I know it's restarting the app, it's working on stuff trying to get this weather stuff to work. Um, in order for rivers to actually show up here, guys, we probably don't want it to just do a call every single time. I call that API every single time. If there's 15,000 rivers and it's having to search through those, if it has to do an API call every single time, that's kind of inefficient and it's going to be slow. It's just going to be a slow user experience for, for the user. And so sometimes you might want to say, okay, Let's pull in all of the rivers right now into our database. We'll get it all into here. Then it can do a whole big scrape of all the data and pull it in. Then all it has to pull is, if I click on the American River or if it's displaying that as one of the featured rivers or something, then I need to do a little call on just that data instead of having to do these huge calls. You don't want it to do that, especially if you're paying for something. Luckily, USGS is free, but if you're paying for like your weather one, right, you don't want it to pull all data from all locations every single time someone clicks on something. Um, that's going to be very, very expensive and end up, you know, end up being very inefficient, right? Not only is it expensive, it's slow and you want fast. People want fast stuff. Okay. Um, all right, let's see if this works. I want to just see if the, uh, the weather widget works. I don't see it anywhere, so we're going to go like this. Cut. Um, I don't see the weather widget, and I'm going to show it a screenshot. Maybe it's hidden by something. Okay, then we'll give it this next instruction. Now let's install RiverData. Um, In order to save on API calls, let's do an initial scrape to populate the database for rivers. Let's just start with Utah for now. Add that to the queue. And by the way, guys, I haven't finished the weather widget. Um, you have to— oh, it's only configured to show an authenticated navigation. Let me add the weather widget to the landing page as well. Weather unavailable. It's possible that weather's unavailable because it doesn't know my location. Right? So one button you can push here, guys, this is one that says open in a new tab. So this is gonna open it outside of Replit and let's try it. Let's see if we can get it to, yeah, see now it's asking for the location, right? It's automatically set that up to ask. So allow while visiting the site. Great. Now it says weather unavailable. So now I gotta figure out what's not working right, but at least it asked for the location. We know we're moving in the right direction. Now we'll just say, I gave it location, still says weather unavailable, and I'll say, oh, sorry, let me fix whatever I have to fix. You just got to keep working, clicking it, trying it until it works. All right, guys, that's a lot of Replit this morning. Let's take a break on this. The river data might keep going and we'll let it just work for a bit. For this section, guys, if you want to participate in this conversation, we're going to do another conversation like we did yesterday. You can turn on your cameras and just kind of be a little bit more present on this one. But it's going to be a fun little discussion about surf— surfing versus swimming. This concept, guys, is made by a good friend of Chris and I, actually, and Taylor, if he was on, named Jeremy Witt. He's writing a book called Surf Don't Swim, and he's coming out with it in a couple months here. And it's going to be really, really an awesome book that you're going to want to— you're going to want to buy. But let's, let's kind of back up a little bit here. How many people here have never swam in the ocean? Anybody? They never swam in the ocean before? Yeah, everybody has. That's good. Does anybody here think they can outswim the ocean? They can beat the power of waves out there? Yeah. Like I said, I'm a kayaker. We've been talking about that a lot, and I have never won a battle against water. It's going to win every time, even when it's 30 CFS coming out of a small thing. It is crazy how powerful it is. It just never lets up. It's always there. And it has unlimited strength, it's going to win. So let's ask this. How much— maybe there's a kinesiologist or somebody, Chris might be able to answer this. I don't know how much, how many calories, how many kcalories do you burn in 10 minutes of swimming in the ocean? Roughly.
Anybody?
Probably. Let's see. 100, maybe. Boom, bingo. You're the first person in 18 cohorts to get that right. So fantastic, Chris. Chris is a, is a health nut too, right? Aside from being a brilliant coder, he runs like 10 miles like it's nothing and, and he reads all kinds of books and stuff. So I knew he was going to be the likely winner of that one. So prize goes to Chris. Awesome. All right. Now let's think about a tsunami, right? We just— I don't know if any of you follow the news on this big tsunami. The big earthquake just happened. Tsunami comes towards the West Coast and all this stuff. I love weather. Like, I want to be a weatherman, kind of, kind of, right? And so maybe in another life I'll do that. But how much energy do you think a tsunami creates compared to that 100, you know, that 100 calories in 10 minutes swimmer? How much energy? How much more energy? Just as a ratio, right? 10 times more powerful? 100? 1,000? 1,000,000? What do you guys think?
Trillion.
Whatever's bigger than that. Awesome. A trillion is not a bad guess for sure. It's obviously so much bigger, almost hard to measure. Any other guesses besides a trillion? There's no actual prizes for this, so if you're wrong, that's fine. The answer, guys, is 4 billion. It's 4 billion times. Now there's obviously variables. How big is the tsunami? How strong is the swimmer?. And so whatever, but on average it's about 4 billion times stronger, right? There's no way, there's no way you're going to beat it. Now what's interesting is I have a good friend that, um, she posted a video yesterday of her. She was, you know, heading out to the surf on the East Coast. Hurricane Erin is off the coast and it's pushing this surf and, uh, tsunamis actually create pretty cool surfing conditions. Um, uh, now you gotta be careful. You gotta know what you're doing and you know, if, If you're a beginner, maybe don't start in a tsunami. But, but bottom line is it creates a pretty unique opportunity, right? So when the surf is coming in, sorry, when the tsunami is coming in, the surfers are excited, right? And they stay on top of the water, whatever. If you're swimming, however, during a tsunami, you're going to get smashed against the rocks and you're going to— bad things are going to happen. Let's just say that. Okay. So we want to be surfing. We want to learn how to surf the tsunami. And, you know, to bring this back inside to AI, AI is the biggest wave It is the biggest tsunami that has ever hit this planet, bar none. Nothing comes even close. Like I said yesterday, electricity, the internet, the Industrial Revolution, the wheel, nothing is going to transform society in a bigger way, in a faster way than, than AI will right now. It is the biggest tsunami ever. So we need to learn how to surf it. We need to learn how to surf AI, this big tsunami, so that we don't get smashed against the rocks, right? No one wants to get— anyone want to get smashed against the rocks? I've done that before. It's not fun. I've been smashed into rocks in my kayak on the bottom of the river or wherever. And yeah, I have a good bruise from last year that still kind of hurts a little bit. Actually, it's kind of— maybe it's kind of gone now. Took about a year. Took about a year to get rid of that thing. But I got smashed on the bottom of the river. I'm, I'm pretty good. I know what I'm doing. And so when I get messed up, It's because something went really wrong and a really powerful something really smashed me and it hurt. Dane Jackson, who's the best kayaker in the world, he hasn't swam in about 6 years. I mean, sometimes you swim because, you know, the wave knocks you over and you get pulled out of your boat or something happens, or you have to just pull out because you can't get up. And he swam this year and he broke like 5 bones. Crazy, crazy. So anyway, I digress. But anyway, so let's talk about what it means to surf. How do we surf something? Okay, here's the definition that we're going to roll with that's used in this book by Jeremy Whit. If you're— and so we're going to bring it back to our businesses, right? So maybe it's your own personal business, it's the one you work for, it's whatever it is, right? But if the technology upon which you are built, right, If your business improves as that technology improves, you're surfing, right? So in this case, if AI gets better and we're surfing— if we, if we— we're going to learn how to surf in a minute, but if we're surfing, AI gets better, our business gets better. Okay, if you're swimming, as AI gets better, you start to sink, right? And you're gonna, you're gonna lose. All right, okay, very simple. The technology upon which you're built If it gets better, you get better with it. Okay. If you are not getting better and it's causing you to now fall behind, then you're swimming. Okay. So let's examine a couple of companies and, you know, ones that we're all familiar with. Let's look at Netflix. Okay. What does Netflix do? What is their—
what's their business model?
They personalize content. Nailed it. Sharon also gets the gold star. Most people would say they're a streaming service or they're, you know, a studio, and they are, right? They are those things. Those aren't wrong answers, right? But actually, Netflix's core technology is personalizing content. How do you know that, Sharon?
Did that just come to your mind?
I'm a marketing person, so you're, you, you're sharp for getting that one right away. That is exactly right. There— did, did Netflix invent the internet? No. Did they invent streaming? No. Do they, do they even have their own streaming technology? No, they don't. I'm sure they do some optimizations on top of other stuff, but they are surfing on other people's technology, right? They're surfing on the internet. As the internet gets better, does Netflix get better? Absolutely, right? Now we can stream 8K and 16K and 32K, right? As streaming, as Akamai, the service that they actually use to do their streaming, as Akamai innovates and spends money, do they get better? Yeah, Netflix gets better, right? Without having to do that. Now, when they spend money on their content studio and lose $1 billion a year, that side of their business isn't so much surfing, right? It's just something they're doing. I'm not saying it's wrong, it's just, it's not surfing, okay? As AI content, As AI video content becomes easier to make and becomes more Hollywood ready, does their content studio get better or worse relative? It gets worse unless they're using it, right? It gets worse, right? Because right now they're not using AI, they're just making movies, right?
I'm assuming they're going to ride the surf with that.
Yeah, hopefully they will. And they actually did. I actually saw that Netflix, and one of the things they did one big fight scene all using AI, and it was, they just wanted to get it out and it came out, I don't know, a few weeks ago. So they're definitely moving that direction. But most of, for most of what they're doing, they're still hiring actors and, you know, building sets and, you know, that's the swimming way to make movies. It's great. We love it. I hope we continue to have that in the world. And so do many movie, you know, people in the movie industry that are terrified of Veo Tech, of Veo 3 and other technology like that. But, um, But anyway, so they are surfing. Netflix is surfing. How much money do you think Netflix invests in sending workers down to the bottom of the Atlantic Ocean to work on the cables that help connect the internet between the continents? Zero is correct. Yeah, they do not spend money on that. But as other people are investing millions, billions of dollars on making the internet better, Netflix continues to rise. Okay, awesome. So we did the, we did the Netflix comparison or analysis here. Now let's talk about, um, let's talk about AT&T. Is AT&T a surfing or a swimming company?
What do you guys think? I don't know enough about their infrastructure, but they're built on something. So I'm— given the size of them, unless I'm really out of date, which I might be, I imagine they're actually surfing.
Okay, I, I agree with your analysis on, on what you said. The only thing I will add in here is they build their own infrastructure. They're the ones going and building towers. They're the ones going in and upgrading from 3G to 4G to LTE to 5G to 6G to whatever, right? Right. So are they surfing or swimming because of that? They're swimming. Okay, this is hard to understand, by the way, guys. This is not easy to just grasp immediately. Okay, it took me like 3 hours of conversation with Jeremy Witt to really understand this. And AT&T, because they're investing billions of dollars, they're the ones carrying all the risk, right?
Can anyone think of a technology? They're not doing both, Jon. Are they not doing both technically? Because once they've done something well, they're surfing.
No, you can't surf your own tech. That's one of the rules. Yeah, if you're the one building it, You're, you're swimming, right? Because it's your, it's your equipment at risk. If a hurricane comes through and knocks over the towers, your technology stops working, right? And you can't depend on someone else to go fix it. You have to go fix it, right? It is yours. It's your responsibility. When that surf comes in and smashes it, you're dead. Can anyone think of a technology or a company out there that could—
hold on, just really quick, like, is it fair to say that they're swimming if they're inventing the the ocean? Like, if they're inventing the tech, they're putting up the infrastructure, like they're creating the ocean.
So how can you— they potentially are if they were the only ones doing it, yes. But they're not the only ones doing it. There are other people helping build the ocean. So I understand what you're saying. Um, and you're also pointing out something that's really important. Can everybody surf? No. Someone's got to build the stuff to surf on, right? That— so that's, that's a critical piece that we'll get to in a second here. Let me ask this question. If you are— sorry, can you— can anyone think of a technology or company that could make AT&T obsolete?
Well, I'm guessing someone like Starlink or Elon or something.
Yeah, Sharon's getting all the gold stars in this one. But that's exactly right. Starlink— Elon could tomorrow, assuming everybody had a device that could actually, you know, connect to the satellite, right? So there's, there's a little bit of work that still needs to be done there. But theoretically, if that, you know, that barrier were overcome, he could say, I'm going to give the internet for free to the entire world for the next month, right? Now everyone in the entire world has free connectivity. Nobody needs AT&T or Verizon, Vodafone, or any of these companies at all, right? I mean, maybe, maybe they do a little bit. Maybe they'll just sell phones. Sell phones. S-E-L-L phones. Anyway, um, so, so anyway, that, that is a really critical piece here, guys. Um, this surf versus swim element. There's risk involved in starting a company. A huge amount of risk in starting a company. If you surf on other people's technology, you can reduce your risk substantially because you don't have to build cell towers, you don't have to send things into space, you don't have to build streaming technology and figure out new ways to do whatever. You can just say, I am going to surf on these other people's technology and I'm going to ride on this. And as that gets better, my company is just going to get better and better and better and better. Right? And not only that, most technologies in the world have more than one provider. You don't have to get your phone service from AT&T. You can get it from Verizon, or you can get it from someone else, right? Depending on what country you're in, right? And, um, and so if you're surfing— because one of the risks of surfing is, well, what if the thing I'm surfing on goes out of business? Okay, no big deal, just move, right? Move to another service that's competing with that. And keep surfing, right? In, in what we're doing here, switch your API key. Maybe OpenAI goes out of business. Okay, I don't think that's gonna happen, but let's say it did, and now all of my chatbots are broken. No big deal. Switch to a Groq, switch to a Claude, and it will still work, right? It'll still have the prompts, it'll still have this, it'll still have those LLMs. It'll all work, right? And there's a lot of ways you can approach those problems. But guys, being able to surf on top of technology is incredibly powerful. Powerful. So, now, let's think about it. Let's bring it back to our businesses. Let me give an example of one of the businesses that Jeremy is building right now. He had a big win in the late '90s with the dot-com bubble. He was part of a company that almost went public and then they ended up getting acquired. It was $120 million or something like that. It was a good buyout and he got a piece of that and that kind of started him off really well. He was very young. He was in his 20s then and was able to turn that into a lot of wealth and he's done very well for himself. But one of the things he decided to buy was a senior care center, bought one in Kansas City. And he's just like, look, people are always going to get old. They're going to need to be taken care of. I think this is pretty recession-proof and almost everything-proof. So I'm going to buy one of these and I'm going to do it really well. And it's been a very great investment for him over the years. But as AI started to come out, he's also a developer. He's a developer and a business guy, sales guy. Very broad, very smart guy, but he was a developer in the '90s, and that's why he was part of an early company in the tech world. And anyway, so he's always like right there at the edge of tech and trying the new stuff. And as soon as AI started coming out, he said, you know, I need to start building some stuff with AI. And he looked around at his senior care center. He said, hey, you know what? One of the things that we suffer from at the senior care center is a lot of people get Alzheimer's and dementia and problems of the mind. And one of the primary causes of that is not the bacon you eat. It's, it's the people you don't talk to, right? It's not using your brain. You're not talking to people. If you don't talk to people, your mind is going to just kind of fade away, right? And so there are literally services that, that he would pay for and that insurance actually would pay for. Insurance will pay more to— they'd rather pay to prevent than pay to cure, right? It's way cheaper to prevent. And so insurance would actually cover people coming in to just talk to their people, right? Super smart, right? Hey, let me come in and talk. We'll, you know, we'll charge you $20 an hour per person and we'll come in and we'll do this. And anyway, so they've, they've figured— or he said, what if I just use one of these AI voices to just call these people and talk to them for hours on end, right? This is infinitely scalable. I can have 100 people talking to him at once, right? And so he called his insurance companies like, hey, I'm putting together a thing. Here's a demo. Like, talk to it right now. See how real that sounds. That's awesome. All right. Here, I want to, I want to have a service that, you know, talks to our senior care folks. And, you know, and just like you pay for real people to come in, will you pay for this service? And they said, hey, we love it. Sure, let's do it. And he got them to agree to $90 a month per person. Right. So he has a senior care center. Let's say it's 200 people. That's $18,000 a month in revenue. Amazing. Okay, but here's the cool thing. He has AI do the outreach to the nurses to say, what time do you want to do this? And then it auto-schedules on their calendar and it auto-configures and sets up the call to auto-dial. It auto-dials. Betty picks up. Hello? Hey, is this Betty? Yeah. Hi, my name's Bob. I'm just here to chat with you today. How are you doing? Right? And I'm an AI, right? And so, and so now all of a sudden the AI can auto-dial. All of the things are set up. AI is doing everything from the scheduling to the communication to the initiating the call. It records the call and puts it in a database and stores the call and the conversation in a database so that next week when they call, I say, hey Betty, how was your grandson's recital? Right? Now you're following up and you're able to actually have a conversation with this person over and over and over again, right? But it's all automated. It's all done. He's got all these systems that are just API'd together and he's now making about $100,000 a month on this. Right. It's completely automated. He has one part-time person that helps a little bit with billing questions. Right. Anytime somebody emails, hey, we have a question on this billing or hey, we need help processing this invoice or whatever, a manual person does that for now. He could probably replace that with an assistant pretty soon, an AI assistant pretty soon. But he does nothing on it other than if maybe 11Labs, right, which we're going to play with in a minute here, maybe 11Labs goes under. Well, he'll switch to another voice provider. Maybe that one goes under, he switched to another one. It's just an API swap. That's all it is, right? And so he is surfing, Jeremy is surfing on all of these different technologies. He didn't have— here's the cool thing. How much money did he have to invest to build all of that? Took him like a day to put it together. It's super simple, right? He's using AI to put it together. He knows code, he knows this stuff, and so that's super easy for him. But he puts it together. All of you guys could build that same business right now with what you know now. You might not think so, but I promise you it's true. You absolutely could do it. All you would have to do is describe it to ChatGPT, make your plan, take that plan over to Replit and tell it to build it, right, until it works. Right? You absolutely could do it. You don't know how to integrate a phone system? Okay, who cares? You know who does know? Your AI assistant, right? Your AI assistant can walk you through. Here's a provider, go sign up for an account, give me this API key, give me your phone number that you signed up for on the site. Perfect, done. Hey, you just set up a caller. Now you can set up an automated sales team that's completely AI-powered, right? And so, um, Mitesh, you're asking, is Repliq capable for end-to-end building from user authentication to database management to everything? Yes, absolutely. It can do everything. If a developer can do it, Repliq can do it too. That's, that's really the short answer. Go ahead, Mitesh. Oh, I can't hear you. Um, you're, you're unmuted, but I can't hear you for some reason.
Yeah, I'm sorry about that. Uh, can you hear me now? Yeah. So is it also capable of having a database and user authentication and everything right in Replit? Like what happened in Lovable was it asked me to go to Supabase, create an account for database.
But in Replit, can everything happen in one place where I don't have to do anything else? That is why I love, that is why I absolutely love Replit, right? Right here, my database, right? Here's my database inside of Replit. Now, if you look at the settings and you dig into the details, it's actually using Neon database, right? Um, which is a Neon serverless Postgres. You don't need to know what all this stuff means, but this is the database it's using. But what it did, what Replit did for, for you and me, us non-techies, right, that don't know how to do this stuff, it said, hey, you know what, I'm just going to give you the database right in here. You can use it. Now if you want to use Supabase or you want to use Firebase or you want to use something else, totally fine. API into it and use it. You could totally do that. But what I love about Replit is all of these things we haven't clicked on half of it, right? Like app storage. Maybe you want to store pictures and videos and other things like that, right? Documents, all that stuff. Hey, let's set up app storage. That's actually a new feature. I haven't even tried it yet. Super excited about it. Does it do auth? Sure. Right there. Right. Can you have it build the auth system? Yep. Totally. I like to use my own auth that I build in here, or I ask it to build a specific way. But it has everything. Hey, do you have a cybersecurity team? Yep, security scanner. You want to check and see if you have any API keys exposed or anything weird? Run the scan. No vulnerabilities found. Last run at 11:01. We're up to date. Perfect. That's where Replit is beating everybody else. They are the most complete solution. You don't have to leave, you can just do it in here. Now there's still plenty of other things you might want to do. Another good system that I'll just shout out is Base44. Base44 is not as powerful as Replit, but it is simpler. They have a lot of these APIs already figured out. The idea behind this one was instead of forcing everybody to figure out how to use API keys, let's just do it for you. It just automatically has these things set up, Giphy, Bright Data, X, Twilio, Resend, ElevenLabs. Now these are a lot of the things that we're using, right? I'm going to show you 11Labs in a minute. Super cool. Zapier, amazing, right? And it has all of these kind of APIs already set up for you. And so it's just a lot easier. Um, but I don't think it's that hard, guys, to listen to its instructions, hand it the API key, and then you've got your own custom system, right? Base44, you'll run into some limitations when you really try to go outside of their pre-built integrations. It's not great when you get outside of that, but if you just have something simple, you need a basic website, Base44 is going to be easier. Um, it's when you want to do more powerful things that you're going to get a little bit— you're going to get a little bit stuck. Um, awesome. Um, okay, I still don't see my— oh, I do see the weather app. Um, let's— I'm actually just gonna— oh, you know what, I want to click on Browse Rivers and see if it scraped data for me. Let's say, okay, you know what, I'm just seeing no rivers found. Okay, so we'll let it just keep working, see what's going on there. Okay, yes, Sharon, I said just use JWT. JWT is just— I forgot what JWT stands for. JWT. It's token. It means a token, but what is it? Yeah, JWT meaning JSON Web Token. That's right. It just basically pulls a specific type of— it's a specific type of very easy authentication that I feel is just more than sufficient for most things. It's secure, it's doing great stuff. That's what I use on GenAIPI. Everything is done through there. Okay, guys.
Jon, just a quick one, if I may. Go ahead. Would you recommend using something like Retail AI versus Replete for building a voice agent?
For a voice agent, I'm going to use ElevenLabs. We're going to do that next. That's actually where we're going next. But what we're going to do, guys, right now is we're going to take a 10-minute break so you guys don't melt. So it's 11:04. We'll go to— actually, we'll make it 11 minutes. So 11:15, see you back here. All right, guys. Welcome back. I need to grab a water bottle. My water bottle stash is empty. Give me just one more second, guys. Grab some water here. All right, sorry guys, I said to take a break and I just sat here and worked the whole time and didn't actually get up and take a break, but here we are. Okay, so guys, we are— we've got about an hour and 45 minutes left, um, You know, I said we might go over a little bit today. We might, we might not. We'll see how fast we zip through some of this stuff here. But, you know, just to review really quickly what we've done, right, in terms of how to build a business in a weekend with AI, there's a lot of things that we still haven't gone over yet. But we have gone over, you know, hey, let's start with a strategy, right? What are we going to do? What are we going to call it? What is our branding going to be like? Who's our customer? How are we going to make money? And we worked through and we built a beautiful strategy. Then we said, okay, let's go and execute that strategy and let's put it into Replit and actually start building something that can work, right? Hey, look, I've got some Utah rivers now. That's kind of fun. We're not pulling data yet, but it's at least got the, it's at least got the river names. Um, and you know, then we tweaked it and we learned, okay, how, let's, let's work with Replit. How can we now start to push this to where we, we turn this into a real product that someone would actually to pay money for. This is getting to the point where as soon as we start connecting this, it's almost a very useful site. I might just finish this and use it for my own purposes. There's still a lot that I have to build, a lot of notifications, things like that. But now we've built out something like this where we have something that's working. Now we need to maybe pretty it up a little bit. Some of the things that we're going to do to make it a little bit more powerful might be, hey, maybe we need some images, Maybe we want a video or two in there. Um, you know, part of, part of launching, when I was launching GenAI PI, one of the big things that I focused on, um, I sat down, I started building and I got, you know, my first version and then my job, my goal in my mind was, okay, I have to be able to get all the way to the end of the user journey and have nothing break the whole way. Right. I had to be able to say, okay, can I click on the test and can I start the test and then can I finish the test? Does it send an email with the, with the result? What happens after that? Does it then push them towards taking a course or paying for a certification? If I click take test now on this, does it, you know, get them to sign up for it or to pay for it? Right. You know what, what happens here exactly? Right. And so, you know, take test now. Okay. Now it's going to ask me probably it should ask. Oh, you know what? I think I gave this one. I gave this account one. But anyway, does my assessment work, right? Can I click start assessment? Um, can I click through a next question? Can I finish? And I had to literally click through 40 questions, right, um, to, to see if it actually worked, right? Can I sign out? Can I click on, you know, every single button? Is there a contact form? Does the send message work? Like, you have to test. It just takes time, right? But what you're trying to do, guys, really, you're trying to not die a death of a thousand paper cuts. Right? Because when a new customer lands on your site, everybody is kind of in a default state these days of, am I getting scammed, right? That's just like something that's in the back of their mind. It's not maybe at the front, but if I click on like the about page and it goes to like a 404, an antenna is going to go up and they'd be like, oh, that, that doesn't feel like a company that's really like— if I'm going to give them my money, are they going to take care of my money and provide something that's actually valuable? Or do they not even care about their website enough to make sure that their, their pages work right? Um, and so if, if I click on anything and it doesn't work, then you know what, what, what happens here, right? Um, so anyway, bottom line is you have, um, you have to go through and make sure that everything works. That's a big part of the process. Once you've gotten there, now you have a product. Now you have something that you can sell to people that hopefully, you know, hopefully you haven't spent a lot of money building. Luckily, when I built this, it cost me for, you know, Replit charges was about $140. That's about how much it took me to build GenAIPI. So what, 4 times the monthly subscription? But whatever, I didn't even care, right? I've paid Chris hundreds of thousands of dollars, maybe a million dollars. Chris, I probably paid you a million dollars over the years, right? You know, Now, obviously, we were co-founders. We were paying— we were both paying those bills. But bottom line is, um, it costs millions of dollars to build stuff and to get things built. So for me to pay $140, I was like, this is amazing. Like, I'm stealing right now. I'm stealing from the world and building amazing stuff. So I was super happy to get that done. I then spent a little bit of money buying a domain, right? I had to buy genaipi.org. Um, I wanted it to just be aipi— aipi.org. That was my first thing, but it was already taken by something else, some other organization. But AIPI would have been super cool. But Gen worked because it was like general, and so it can apply to a lot of things, and it's kind of a play on words because of generative AI, and that's obviously big. So GenAIPI worked just fine, and it served me well. But bottom line is I had to do that. I also had to say, you know, what's another thing that we're looking for when we're, you know, when we're shopping? We're looking for social proof, right? Does the Facebook page work? Does this link go somewhere? Right. Can I— are there any posts? Right. Is there— are there people? Are there followers? Anything like that? Right. Does that link back to the main website? Can I click on the Instagram? Does that work? Yep. There's something there too. Fantastic. There's a few followers or something there. Right. And I don't really care that much if there's like tons of followers or whatever. I just want to know that it exists. Right. It's just another avoiding a paper cut type thing. Right? And so you have to build— and by the way, in order to do your Facebook page and your Instagram page, you need content, right? And so we're going to take some— we're going to take some time here now and just kind of do a little bit of— you guys, I'm sure, all know how to create images with AI, and that's great. I'm going to show you just a few of the tools I like to use for image generation, some of the processes I use to get really good things really fast, and then how I build social posts and things like that that I also that do. I usually use AI for business-related AI posts. If you see me post it under Jon Cheney, I wrote it. That's just, that's just my style, right? I don't even use AI to fix it hardly, right? Occasionally I'll be like, I'm going to test something with AI and see if it works. Like, I asked it to, um, one of my, one of my questions, so they said, hey, help me make a LinkedIn post that is guaranteed to go viral. And it did. It got like 40,000 views. It's awesome. Right? But I also tweaked it a lot. I spent like 30 minutes working on that post. And so even though I used AI to build that, it was still very much like, no, I want it to say this and let's tweak this and let's say this. And, you know, but the style and the, anyway, the flow was, was definitely guided by AI. But, but for GenAIPI, for my business, I don't care if AI writes that stuff. Totally fine. And there's a lot of really cool things to do. And then not only that, guys, my idea when I was launching this is I was like, well, I'm gonna launch this via a Facebook ad or a Meta ad. So it's Facebook and Instagram, whatever, right? I'm going to get all these people to click on the site. And so in order to build ads, it's basically the same thing as posts. You need text and you need images or video, right, to go along with that, to get people to do that. And so you can use AI to go from, you know, now I've got— we've built a solid product, right? Strategy, all this. And now we have a product. It all works. There's no paper cuts anymore on the product. It works. And now let's push it out to the world. Now let's market it and sell it, right? So that's kind of the last phase here that we're going to focus on for the next couple hours. And we'll show a couple other things, other things that will make it more powerful. We'll do some voice AI. We'll show some video AI. So I'm just going to bounce through a few platforms here. This next section, you don't necessarily have to, you know, be following along and doing things with me. You can, you certainly can. But maybe just watch and learn a few things, a few tricks. Okay. So one of the first things you have to be able to do when you are building an app or something, right, is one of the nice things to have is a logo, right? Help me build a logo based on everything you know about this project. HighWater is the name, but I don't want a— I don't want the name displaying in the logo image you create. Okay. Let's first look at this. So right now it actually has built this kind of 3, 3 things. Like, honestly, I could probably live with that. Is it amazing? No. But does it work? Yeah, totally works. People are not going to not buy from you because of your logo unless your logo is a paper cut. And that's just so bad that people are like, ugh, I don't, you know, I don't like that. Right? But this logo, while it's not like, hey, wow, a new Nike symbol, right? That's the new best logo ever. It's not amazing, but it actually probably works. But still, even then, I want to show, I want to show how to, how to do this, how to build these logos. So first of all, when I ask ChatGPT, it's going to just talk to me. Here's what we're trying to do. Here's the logo direction, visual style. Here's some of the concepts I'm considering. Maybe a surging wave or the river gauge, abstract H, droplet plus mountain, paddle blade hidden in a current swirl, high alert symbol. I don't know if I love any of these. I want— let's focus it more on the wave. And high water feel. Um, give me some more ideas. Okay, so we'll talk to it, we'll listen to it, we'll work back and forth, build our strategy with this. Cresting wave with a sharp peak, that's kind of like an ocean kind of feel. Stylized hydrograph turned wave, that could be kind of cool. Let's try this, let's just try that. Let's try this. Okay, so now it's going to build this for me. So this is cool. I, I like ChatGPT as an image generator. I think it's pretty good. There's some others that I like maybe a little bit better, although it just depends on kind of what I'm, what I'm trying to do, what I'm trying to accomplish. While this is working, I hate sitting around and waiting, so I'm not going to wait. When I was building GenAIPI, I decided to go to design.com. I just Googled, you know, free logo AI generator, right? And, and I landed on design.com and found some cool stuff here. But design.com is an AI generator and I was able to just pick this. My idea for the GenAI logo was like a brain, like a human brain with like circuits, right? And this is what kind of what it gave me, right? And the other thing I like is that it's a circle. I love circle logos because they stand out really nicely up in this favicon. The favicon is important, right? That's how people are going to see you on their tabs. If you don't install a favicon on your site and someone opens it, it's going to feel incomplete. It's a little paper cut. Okay, you have to install favicon. That, for those of you who don't know, I'll just drop it in the chat. Favicon. That's what that little tiny logo is that shows up on the browser. Okay. But anyway, yeah, so I went in here and I just, you know, I got the logo, I picked a color and, and just kind of, kind of ran with it, right? I said this color right here, green. Awesome. All right. I'm done. Color. All right. Let's get rid of that. And this doesn't even need to be here. And I'm done. Save. And then the cool thing I like about design.com, the reason why it was actually super useful is because when I wanted to download it, I'm able to say, okay, I want a transparent background, or I want, you know, the logo only, and I want a transparent background. And so this is what I would use then for my, you know, all of my different, you know, social profiles, right? If I go to, if I go to Facebook and go to GenAI API where it looks really good there, I've got my, my nice green thing here. I generated this with, I don't know what it was, maybe I used ChatGPT for that or Canva or something. I'm not sure what I did with that. And I could certainly always go back and prove these things, right? One of the things that you don't have to worry about is you don't have to get it right the first time. Okay. Okay. These, these logos are okay. I don't know if I love them. Again, this is not about logo design necessarily right now, or like whether or not the logo is good. What I want you to do is just kind of create. To me, this looks like a tsunami is attacking a city. That's what I see here. This, I don't know, man. I don't, I don't love, I don't love the bars. It feels like a wave is attacking a city. Let's make the wave look more like a river wave going through or cutting through. Let's actually say cutting through, you know, trees. Yeah, let's just say that. Okay, so we'll let it, we'll let it do that. And, and that's fine. Yeah, surf, surf the tsunami. Exactly. That's what's going on here. So, so anyway, guys, I think this is good. Another thing I meant to do this before I sent this new thing on. I, I, I like ChatGPT. I think it's good. I'm gonna show you some cool tricks with, with the, with images you can do inside of ChatGPT. But I actually really love a platform called Midjourney. Midjourney is, is really awesome. Some of you may have used it before. Um, anybody here ever used Midjourney? Yeah, Sharon, Jeff, a couple of you guys have. I think it's super fun. I was building some stuff the other day in Midjourney. This is, this is really cool. This was like supposed to be somebody who's like learning AI and they're starting to feel confident now. Kind of cool. There's a random dude that like runs behind her in this one with what almost looks like a rifle case. I think it's supposed to be just like a briefcase, but it kind Anyway, that's kind of fun. So anyway, you can turn these images— so I, I love, I love Midjourney because of the quality of, of the, you know, I might not use this picture, but, but just look at the detail, right? I mean, just look at how amazing that is and how, how like real. I mean, even this is like a cyborg, it doesn't look like a human, but it's a human cyborg and it just looks awesome, right? The quality of the images you get out of it, I feel, is way— this was a a post I did, you know, on how pioneers and entrepreneurs, you know, are both doing things for Pioneer Day post, right? So I'm just looking for something there. But I love, I love Midjourney. Use it to create all kinds of logos and designs. And, you know, you can even, you know, violate other people's IP if you want, which Midjourney is being sued right now. So, you know, what you do with this image, it's actually less Midjourney's fault. It's more your fault if you use this in a appropriately. So just user beware, right? Make sure you respect other people's IP in the right way. If you're making money with it, you might need to license it, okay? So just remember that. Most people don't. Most people don't pay attention to that, and you probably won't get prosecuted by Disney if you did that because you're a nobody and you're posting it and 70 people saw it. Nobody cares. But, you know, you guys can make that decision on your own.. But one of the cool new things you can do inside of Midjourney is you can have it create a logo or something like you can see here, and then you can actually have it animate it, right? So Midjourney is a little bit different. If you're not familiar with Midjourney, maybe creating one of these prompts might be a little bit difficult, right? And so one of the things I like to do is, yeah, see, this doesn't look like a river wave. You've kind of failed. You've failed at this. Write me a good prompt for this logo that I can take to Midjourney. Remember to focus on the river wave aspect. Okay, so now it's going to write a prompt for me that I can take over to Midjourney and do that. This, you might also use this style if you're this, this workflow, if you are making a video in like Veo 3, right? Veo 3 requires a specific type of prompt to actually work really well. And so, anyway, let's do this here. Let's copy this. And if you don't know how to do it, you just use AI to help you figure it out. So let's paste this. I don't know if this is gonna work. Yeah, these, yeah, all, all of these extra tags here are kind of, we're just gonna go like this. We'll just get rid of the tags. We'll just give it this thing here. Now we use V7. Wait, V7, is that new? As of like, when did it, when did V7 come out? I guess it's been out for a minute. Okay. All right, good. All right. Some of these are, some of these are a little bit cooler. This is kind of like a moon. I actually like, remember I said I like circular logos. This is also cool if I'm putting it on a hat or something, right? Like I've got my, I've got my GenAI hat. My Genii API hat. This is the only one in the world. Actually, there's two of them. I got, I got two of them. I got one for my wife, one for me. But anyway, um, it's cool if you throw it on a hat. It's nice to be able to have a logo that fits well, or maybe you want to put on your shirt right here. And so, um, I personally like simpler logos. Like, this is cool, but it's so complex. Try, try embroider— embroidering that. It's going to look terrible, right? It's not going to work well, right? And so you want to think about things What is this logo going to look like really when it's really small, right? What's it going to look like really big when you blow it up and you put it— this looks like a chicken to me. This is like a chicken to anyone else. That's what I see. I just see a rooster, you know. Anyway, so again, we could keep going on, on working through these things and, and figuring it out, but eventually you're going to land on something. And like I was saying with design.com, it, it can help you download, you know, your vector file or your you know, your EPS, your vector file, which means you can blow it up unlimited. Or maybe you need an SVG or a PNG for different use cases, right? And so having all this is nice. When you build with ChatGPT or with Midjourney, you're not going to have those file types, right? And so you're going to have to go through some sort of conversion. You might have to jump on Fiverr and ask somebody to build a good, you know, EPS or a vector file for you. And if you don't know what these things are, that's okay. Like, it, it's not, not a big deal. Just do your best. What you're gonna need to, what you're gonna need to do is when you're building a business by yourself, guys, you have to be a generalist. Now, if you don't know things about image generation, that's okay. That's where AI is gonna fill in the gap for you. But you still have to take care of the marketing and the images and the fonts and the buttons and the user experience and the cybersecurity and the email setup and the sequences. You have to do everything. That's part of the, the game of being an entrepreneur doing it by yourself. That's the price. Okay, luckily AI can really significantly help you in all of those things when you can't figure it out, right? Um, yeah, so Sharon, you did ask a good question. I can't work out how to take what I like and amend it accurately.
Um, so are you saying like if I want to adjust this, this image So for example, in Midjourney, I created a scene and I wanted some text in it. So I created like an avatar of a person. I wanted the name and it kept misspelling the name.
So it was almost like it's gaslighting me, you know. No, you know what? Great question. Let's scroll down here and see how Midjourney does with text. You actually just have to know that Midjourney sucks at text. I wrote a, I wrote a recent arrangement of a very popular Christian hymn, It Is Well with My Soul, and I asked it to create me a cover and I kind of gave it some ideas. And these are all pretty pictures, but if you zoom in, clearly you can see it has no idea what it's doing, right? It's really— it's with hilly soul, right? It's with— well, I've made a funny video where I read through all these. It can't do text. It can't. But what it can do, right? Let's, let's download this guy right here. Here, let's get one that doesn't have Let's just take this and we'll download. Where do I download? Where's my download button? Anybody see download? Oh, it's right up here. It's hiding underneath this. Ah, sorry. Okay, so we're gonna download this file. Looks like it wants me to save it and it'll just say image of ocean. Save. We're gonna go ChatGPT and we'll just open a new conversation here., and we'll give it my image. Say, add the words, it is well with my soul. So, and to this also, add the artist's name, Jon Cheney. This is going to be an album cover. Yeah, okay, perfect. So ChatGPT is quite good, Sharon, at adding text, right? It used to not be even just a few months ago. It couldn't do it well, but then in one of its updates, it nailed it, right? And it can, for the most part, get it right. Chris sent me something the other day that was, it was just like a summary of some idea we had, and it was made by AI in an image format, and it did not do well. Hey, that's weird. It just did a PNG. Or it just made— it's making me download this. Let's see if it worked. Um, oops, there we go. Why is it not opening? Hello? Okay, click here. It is well with my soul. So it did okay there, right? I think that's— I think that's all right. Um, make it more dramatic, and instead of asking me to download it, just display it here in the chat. It should be able to just display in the chat. I don't know why it's sending me away. Maybe it's because there's something weird going on with the— because I'm in the ChatGPT app, and so if I was on the website right now, I think it would actually do this differently. But hey, it's actually showing it now. Here's the more dramatic version. Show it to me. Or maybe it's not going to show it to me. Let's check this. I'm interested. Uh, we have so many things open here. Let's get out of here. Let's just go back to album cover design. I see no image. I still see nothing. You're not providing that image. Okay, you know what, um, it's possible, it's possible that because it, uh, let's just try this. I'm curious. Let's just go new chat and we'll drag the image in again. I think that the app is used— is not completely updated to use ChatGPT-5 yet properly. Okay, let's try this. Let's do a new chat. Sometimes, guys, AI just isn't— it's not that the AI can't do it, it's that somehow the interface is confused. And so instead of continuing to argue with it, obviously that wasn't going anywhere, just start a new chat and give it some new context. You're hiring that new employee, right? And it's just gonna, it's just gonna start over on the task. That other one was bad. Fire fast. Fire fast with AI. But yeah, it— I, I love ChatGPT, guys, specifically for combining things with images. Like, let's, let's do another example here. Let's say, let's look for this image of woman holding pie. Okay, let's see if we can find something here. Alright, I like this one. We're gonna use this one. Save image. Got a nice lady holding a pie for me here. Okay, now I need an image of a bowl of ice cream, and we're gonna take this one because that looks just awesome. Except I think this has a Getty Images thing on it. I don't want that. So let's do this one. Looks nice. Save image. Save. Okay, now we'll go back to— I have too many things open, guys. I'm just going to start closing stuff till I find what I need. Oh, it's still getting started and it's just going so slow. I don't have time to wait. All right, I'm not gonna wait. We'll come back to that, see if it works later. Um, what I am going to do though is I'm going to combine this right here with this. Okay, so let's say, um, please switch out the ice cream— or sorry, the pie woman is holding— with this ice cream. Maintain everything out, or keep everything else the same. Okay, so if you are selling a product, guys, if you have, you know, a side business or a home business, some, you know, whatever it is you're doing, right, and you need a product image, right? Product imagery used to take a lot of money, a lot of time, um, to, to be able to, you know, go and get a photo shoot done and get this just right. Um, ChatGPT is now very, very capable of combining these images in a way that is extremely, extremely efficient. You can have it create an ad, you know, after we're done with this, we can have it add text to it, have it create an ad, make it a '70s style ad, make it whatever. I don't know, like there's so many different things you can do. And ChatGPT is very capable when it works. It's just sitting in this getting started loop and I don't know why. I don't know why it's doing that. You're absolutely right. Yeah, I— this ChatGPT is broken, guys. I broke it. Yeah, I'm not sure, I'm not sure what it's struggling with there. And this is still just getting started, so I'm not going to sit and wait for this. I'm going to come back to it right now. We're going to move back over to Midjourney a little bit and explore. Um, so let's say, let's say I did like this one and I get over the fact that it's a chicken. Let's just animate it Um, yep, that's fine. Um, so guys, Midjourney costs something like, I don't know, $10 a month. For more advanced, I think I'm on a $30 a month plan or something like that. There's so many different image generators out there. Another one that I like a lot is Runway ML. Um, Runway ML, many of you probably haven't heard of. I recently reset or cleared my cache, and so that's why I have to log back into all this stuff. Um, man, Aleph is so crazy. I haven't even tried Aleph yet, man. I'm excited to try this. And I actually pay $99 a month to have like unlimited stuff on this plan. But anyway, if you go to, um, go to dashboard, all tools. So Runway ML guys can generate videos, images, audio, lip sync, right? So I know it's showing like an animated character here, but you could put your own face here and have it, you know, have you say something, right? You upload some audio and it does it, and you can create the audio, you know, here, or you can create it in ElevenLabs and somewhere else and bring it all in. So many different things that AI can do, stuff that used to take a lot of time and a lot of expertise now just doesn't take that anymore. We're gonna keep closing these things out until we get back to a good working position here. Those animations are still going. All right, it did okay with this one. This is actually one of the worst image or results that I've seen. The angle of the ice cream just looks weird. Maybe there's a second one here. The ice cream is too big and the angle of it is weird. Also remove the cloth so that she is just holding the bowl of ice cream directly. Make it feel natural. Okay, so we'll see if we get a better result there. But we're gonna, we're gonna play with, with some video generation stuff as well, guys, here in a minute. Let's, let's generate a video. Let's generate a video. And we're going to start— one of the things that you can do in here, first of all, you can, you can drop in videos or text. One of the things that you can do with this is you can actually upload video. I wonder, I'm not prepared for this. I wish I had a video. What is a simple video I have potentially that could be where we could change this? Oh, you know what we could do? What if we just use the video that we were watching yesterday of Eric Schmidt sitting there talking? Let's see if we can change what he's doing here. Let's see here, MPEG movie. Do Gen 4 screen recording. Is this it? No, it's not it. Um, maybe, maybe I did it here. Oh my gosh, I had this just yesterday. I don't think that's it. No, that was something else I was trying there. Tom Cruise. Man, where is this video? Okay. Yeah, that was a VO3.1. We could potentially, could potentially use that, but where is this? I don't know where my video is. Come on. No, that's too long. Yeah, and that's 4 gigabytes. I want something small that we can use to just switch something out here. Nope, that's too big. That was a Joe Rogan thing. That's also Joe Rogan. Oh, you know what we could do? Maybe we could just use this. That— how long is that though? It can only do 8 seconds. You know what, we're just going to use one of these ones that I, that I created that was like 8 seconds long, this news anchor one. And we'll just change the— we'll just change the background on it. I think this one will actually work pretty well. So let's just drag this in here and we'll just say Yeah, Gen 4. Change the, uh, background to a, um, volcano erupting. I'm not sure, I've never actually tried this, so I'm experimenting live here with you guys on this. Um, but I love, I love doing that. This is how I learn, guys. You just you just try it. So now we've created these chicken logos. They're kind of weird. I don't think I would use any of these, but it's all right. Anyway, they're kind of fun to bring to life. This is, this is kind of interesting. I kind of like how it feels like you're moving, but, but the wave doesn't feel right. It just feels too much like a chicken to me. So anyway, let's animate this one. I want to see what that one looks like. Um, all right. So this is going to take a second. So guys, I, I never sit around and just wait for this stuff while we're, while we're waiting for this, we're going to go over to Veo 3 as well. I want you guys to just see these different systems and what some of what is possible. How many people have tried Veo 3? Anybody here? Yeah. Veo 3 is super fun, but it is very expensive. It's $250 a month minimum to be on a package with Google that allows you to even use this for now. Hey, Google AI Ultra subscribers get double the credits. That's cool. I got more credits. Let's do a new project here. There's different models here. The thing that makes VEO-3 really awesome is, we'll just use the fast model, I think that's fine, is it can generate audio. I don't know, let's think of, let's think of something. Think of something fun. Let's go with the volcano. A volcano erupting and a plane flying quickly to escape the— what's that? Paraclastic cloud? What's that called? I don't know. Am I saying that wrong? That's the pyroclastic cloud. Pyroclastic cloud. That's what it is. That's the big, huge explosion that happens. Um, volcano erupting, um, epic cinematic music in the background, uh, realistic, um, filmed with what do you call that, IMAX camera or video cameras in super high quality. Yeah. Sounds of the volcano and the plane as well. Okay, let's just see how that happens. So obviously, I could make this a lot better and very, very specific. There was a video I uploaded to a social post I saw IKEA. Let's just look this up. IKEA Veo 3 video. This was actually not done by IKEA, but look at this. It's pretty cool, right? That was created by Veo 3. And think about what that would take to create with, you know, just a standard, you know, video agency, special effects, all the setup, right? I mean, it looks as real as you can imagine, right? It's unbelievable. But this prompt was done in a really very specific way where it says, okay, we're going to say, all right, first second, do this. Second second, do this. Between 2 and 6, do this. Between 6 and 8, do this, right? The yellow throw blanket gracefully lands, right? All chaotic sound stop. A single soft boom as the blanket lands, right? Sound of a furnished quiet room, right? So Really cool stuff here. A lot to create this kind of prompt though is hard, right? None of— I don't know how to do this by myself. And so, but if you know that this exists, guys, this is how I learn. If you— because a lot of people ask me, okay, Jon, you know, we're done with the class now. How do I, how do I stay on top of things, right? I watch for stuff like this. I follow some social profiles. I'll show you guys some of the profiles that I follow so you can see that. But you can, you can take this stuff and then say, okay, ChatGPT, help me create, you know, I'm doing this volcano scene in VEO 3, but I want you to help me set up the prompt so that it's this specific, right? And then we hand that. In fact, we'll do it here in a second and see what happens. I'm going to see what VEO 3 comes up with. Let's see what, what we get here. Okay, it didn't get the pyroclastic cloud correct. Just kind of smoky. But it did have some music. It has the sound of a plane. Interesting. So it did okay. It did okay. So now let's take, let's take this and let's go over to— come on, copy. Please. And let's go to ChatGPT here and go like this. Looks like Stephanie's back. Okay, please take this prompt. Use, use the prompt style below. And create a detailed prompt that I can hand Bayou 3. Okay, so let's go back to here. We'll copy the prompt and jump back into here, and we don't need this stuff. And let's see if we get a better prompt. Okay, this should be super easy for it to do, right? It's basically doing a, um, a JSON-style prompt here, or YAML prompt. Um, okay, shot opens, the breathtaking passenger plane bursting to be flying across the frame as the camera tracks alongside it. Change it, actually. Yes, make it a single tracking shot and make the plane fly away from the volcano like it just took off and is barely escaping the explosion. Okay, so we'll let it adjust that a little bit. Aspect ratio, scene description. I wonder if it can do the aspect ratio. It's interesting. Aspect ratio 16 by 9. I don't know if I can do 21 by 9, but that's okay. All right, so let's copy this. So now we've got this nice big fancy prompt that again, we don't know how to make, but we found somebody else that made something. Cool that we liked and said, hey, I want, I want a result that good. Okay, great. Well, let's see if we can, let's see if we can get this. Let's see what happens here. I'm excited. I'm excited to see this one. Okay. All right. Meanwhile, we were generating something over here. This is the most intense. That's, that's pretty crazy. Now, I only had to do 5 seconds, I guess. I guess I can only do that with Aleph. Aleph is this you know, change the background type thing. Um, examples of, uh, Runway A11y background changes. There's going to be some videos here. Uh, yeah, you can change anything. What if you have this video and let's get rid of it. Okay, so we've got this video. See if we can get to where we actually can see some examples. Play, please. Dust. That's exactly what I had written here. Let's see the result. The two people in the video disintegrate into dust. Yeah, see, that's crazy, guys. Do you know how hard that was for, for Endgame to do that? And how much, how much money they paid those VFX guys to make everybody disintegrate when Thanos snapped his thumb? Now anyone can do that. Anybody. My kids can do that. All you have to be able to do is upload the video and write this thing: the two people in the video disintegrate into dust. So crazy. I really, really love Runway. You can change— you can change things that happen in this. I've seen a lot of really cool videos of, of like, you know, people will take Hollywood movies and, and they'll just break it up in like 5-second clips and they'll change stuff. They'll like— like I saw one with the, um, what is it, um, Stayin' Alive, um, John Travolta dancing, right? But they made him just— it kept switching the scene of where he was dancing and stuff. It was really cool, really creative. And, and all they're doing is just taking these little clips and then putting them in here and then splicing them back together, right? Super, super cool. Very creative, fun way to use AI. All right, let's see if this is anywhere close to— that's actually a pretty cool scene. I actually kind of like that one. That one is way better than the other ones. I want it flying. I want the plane flying towards It's flying towards the volcano. I don't know why I can't get it to do that, but that's okay. But anyway, I already think like this one was pretty good. I'm a fan of, I'm a fan of this one. And what you can do now is you can say add to scene and now you're starting to make a movie, right? And now you can say, okay, let's do the next scene, right? Let's extend it, right? The plane explodes. Yeah. The plane explodes. Perfect. And now I've, you know, now I'm creating this next piece of the puzzle here. This pitch. Okay.
Fast.
Go. Please go.
Next.
The plane explodes. Extend. Plane explodes as it gets too close to volcano. Go. Whoops. Come on, work for me. Next. Extend. Here. Go. I don't know why that didn't work the first time. Refresh. Refresh the user, the user interface. Um, any questions about any of this image or video generation stuff, guys? I know that there's— I'm bouncing between Midjourney, Veo 3 runway, you know, all this stuff. But I'm trying to just show you how I can create really cool stuff quickly, even though I'm not an expert at it, by using AI to help guide me. I see other stuff, I use AI to then use that same style to generate something similar. Does that all make sense? Any questions about this flow, guys? One of the things we haven't done is pull some of the things back into Replit. Like, so for example, let's say we do want to use one of these logos that was created by, um, by Midjourney. Let's say we want to use— wow, that is not what you want to see in a kayak because that means you're rolling and the river is just having its way with you. Yeah, this one's just kind of crazy, and this one's kind of crazy. I just don't like any of these. So let's just say I'm going to use this, this image here. Let's just download it. We'll just save the image. Um, verifying you're human. Perfect. Save the image. Now I'm not going to do this completely properly, guys. I really should go into like Photoshop and, um, and, you know, delete the background, make it, you know, whatever. I am just going to crop it so that we just kind of get this I'll just try to go right on the edges here and go as close to square as possible. We actually, we can do 505, 506, 505. Hey, that's a square. All right. And then tools, crop and save. Okay. So now I have a WAVE logo and now I wanna put this into my project with, with Replit. Super easy. I'm just gonna grab this. I'm gonna drag it over here. I'm gonna say, please install this as the logo. Everywhere throughout the site and install it as the favicon as well. Okay. That's all it takes, guys. If you have an image that's gonna be your hero image, it's gonna be right here when you land. That's cool. Like, I don't have any imagery on this. This is very just like, it's a tool, right? Here's, you know, here's the flows, here's your stuff. It's, it's kind of no fluff. Right? And so maybe that's the style I'm going with here. But, but ultimately, you know, if I want to add an image in, that's super easy. I get my image that I've— some awesome image of the ocean. I just say, hey, install this as the hero image on the homepage. Send, right? I'm not going to do that because I don't want to do that, but that's all it takes to add these images or these videos. Maybe you want a video in there, right? That's great. You can upload it to here and have it stored in here, or you maybe you could upload it to YouTube and then embed a YouTube video in here. How do you embed a YouTube? Ask your AI to do it. That's all it takes. Okay, and this is actually where I want to go back to. I was talking, I was showing you guys Cougar IQ earlier. This is so cool. So I, I'm a BYU fan, big BYU fan. I was wearing the BYU shirt yesterday. Meeting went well, doing really cool stuff. GenAIPI— I haven't announced this yet, and I'll announce it once I, once I get the final document from them— but GenAIPI is going to be the official AI partner of BYU Athletics. And so I'll be working with about 700 of their athletes and teaching them AI, teaching them these types of things. And I'll get all kinds of cool, cool stuff from that. Lots of free tickets and fun things to do there. But really excited that I get to continue to engage with them. But I built this kind of dashboard. It's like a fan dashboard. And I didn't actually like really finish a lot of this stuff. I kind of worked on CosmAI, as you saw, and it needs some updating.. But the place that I did work on a lot, it's kind of just weird, but I built this past game so I can look at past games for football or basketball. And, you know, it automatically shows, you know, all this stuff and it pulled all this data from actually a free ESPN API, right? All of this is free. Okay. And so let's say I want to look at the Buffalo game last year. That was our kind of, that was our bowl game. And when I land on this page, I've got, you know, it shows the score. I can look at the box score and, you know, I went through and I had it, you know, update a bunch of things here. The team stats, you know, who won on the different things, all these cool animations. So I cleaned it up a little bit. I probably spent 2 hours on this, right? Categories won, right? So BYU just kind of dominated across there. Here are the top players and different things going on here. Here are the top rushing players, all that stuff, right? But then this was so cool. It also— the ESPN API pulls in video highlights automatically. And so it built all these things. And so if I want to watch, hey, you know, here's an onside kick stage, right? I can just click this and watch my cool— my Cougars do this surprise onside kick and recover it. And amazing, right? But I have access to this. I can also, I can also download the video. Maybe I want to go share that on, you know, on my social profile and be like, hey, remember this? This was so awesome, right? Whatever. I can easily quickly download that. I just think it's so cool that ESPN— all of this was created with free data, with a free API. And it's actually— there actually isn't an ESPN API. So this is an interesting thing. I'll share this. ESPN does not offer an officially documented public API. However, there's a hidden unofficial ESPN API the developers have reverse engineered. And so there's just a GitHub, you know, page on here that just says here's how it works, here's all the stuff you can get. I just handed this, I just copied this, handed it to Replit and I said, build something cool. I gave it more instructions than that, but I basically said, yeah, I want to be able to see the last few seasons of football and basketball and all the stats, and I just kept asking it to pull data until I had this CougarIQ thing. Done. Pretty awesome. Now we've installed our new crappy logo. I actually like the old logo better, So I might revert and I'll show you guys how to do that too, because that's an important piece of what we're doing. I also theoretically installed the favicon. So let's open in a new tab and see if the favicon shows up. Yep, there it is. There's my logo. You can— it's hard to see because it's small and it's kind of a weird thing. And I didn't— again, I didn't get rid of— I didn't get rid of the white, right, and make it a transparent background. And so it just kind of looks like that. A lot of, a lot of logos do that, right? You can see a bunch of other ones over here that you know, it's kind of whatever. This is my CougarIQ one. And so I also did a quick one with that just to have something there, but it looks better to have one than to not have one. That's critical. That's really important. Right? So make sure you get your favicon installed. Super easy. Um, I'm actually going to have it roll back. So let's say, you know what? I didn't really love that. Um, so this is the, this is the new, new logo. Um, it did that, but I want to go before that. So we're just going to roll back to here. We're going to say, hey, you know what? Let's not, Let's, let's get rid of that logo. Let's just go back. And so what's it going to do? This is what it's going to do right here. Do I want to roll back the database? I don't need to because, you know, you can, you can roll back the database. Actually, no, it doesn't matter because there's nothing there. We'll just roll back to this checkpoint. And so it's going to just kind of start over and say, all right, what do you want to do differently now? And so it should have now deleted that favicon. And reverted back to the old logo. So we've eliminated that. There it is right there. Let's check the favicon, see what it looks like. Okay. Oh, that favicon is still there. That's interesting. I don't, I wonder if that was just cached. It's very possible that was just cached. Let's try this. We'll open it in an incognito tab and we'll go to it. No, somehow, somehow it stayed. I don't know. Maybe the cache held into the incognito. I don't know. But I want to make sure that my, I'm actually going to just tell it use— why can I not grab that? I want to just click on that thing. Use the little WAVE logo here and update the favicon to use that as well. All right. So let's see if we can get the favicon to be running properly. At this point, guys, I'm just kind of doing stuff so that you can just kind of watch and feel. And when you go back later to, you know, figure all this out, you'll, you'll be able to go back and watch and, and see these different parts here. There's a question here. Can you create a Replit app and then duplicate it exactly, but change look and feel? Yeah. Lisanne, love this question as a white label marketing suite for service-based clients also. So there are two ways I would do this, right? Number one, I don't love the idea of just cloning an app and then giving it to different people for different use cases just because you're changing that. Here's why. When I need to then update a feature, what if I have 10 clients and 10 cloned apps? I have to update that feature all over the place, right? But if I want to just update, update the look and feel because it's for different clients, let's say I build something that's for ecommerce brands, and so I have a Walmart brand and I have a Costco brand and I have a Smith's brand or whatever, I've got all these different retailers, then I could just add a section in their profile that says, hey, update the, um, you know, update, you know, the color of the background and update the logo that displays based on their company settings, right? And so they can just set up like a profile or company settings. Now you can, if you want to, just you know, create a Replit app and then duplicate it, then, you know, you can, you can definitely click right here and you can click Remix. Okay. So again, right here in the top left, right where it shows the name of your app, right? Riverflow, which isn't actually the name of my app, but anyway, you click on this little more options right here and you click Remix, right? So maybe you work on something for a client, right? And, or, and you wanna share it with someone and say, hey, here's a good starting point. You take it from here. You just click Remix and then they can take that and do whatever they want with it, right? So you can make it public in order for that to, to work. So super easy to make that happen. You can also ask Replit how to do that. Okay, let's see if we got the favicon updated now. No, it's still not updated.
I'll work on that in a second. Sharon, question? I just meant it really from my own use because I totally get it. It's not multi-tenant. I totally get what you're saying about it. I wanted to be able to kind of create something and then like we're doing messing around with the logo because I've tried this sometimes and then I've not been able to, I don't know if I'm going to roll back too. So I'm a bit reluctant to do it if I only want to roll back one piece. So what I'm thinking is, can I literally take it, replicate it, and then just change the look and feel of that solution, my solution, so I can kind of test it with different audiences rather than a flat visual? I can literally say which one do you prefer and test it, you know, the MVP with different people based on the way it looks and feels, even though it's exactly the same solution?
Yeah. If you really are just trying to test look and feel, I would go live with one. You could, you could have the Replit create two, like, test A and test B, like CSS files, basically. Right? And just say, all right, I want to test two designs. And then you could say, right now I want you to activate test one. And then what you could do, what I like to do, Sharon, whenever I need something like that to be like an easy change, I can create an admin function, right? Maybe I wanted to change my GenAI API to a red color instead of this green theme that I have. And I'd say, you know what, I just want to try red and see if people do it better, or blue. Maybe blue will be more inviting, right? So I would just create an admin function that says, hey, give me the option to change the platform for everyone in admin and I can just activate blue theme, activate green theme, right? And I can just toggle back and forth, but I as the admin and the only one that changes it, changes it. And then if I change it, it changes for everyone. Anytime someone comes to the site, it's just changed permanently, right? So that's one way I would, I would do it. Maybe A/B testing colors is not super easy, or even full designs is not super easy unless you you literally build both versions and then have an easy way for the system to toggle back and forth between it. Just be careful. I mean, some— to me, that feels really complex. You know, you could also just do some screenshots and just have people look at screenshots, right? Like, hey, do you like this landing page better or this landing page better, right? And just have the screenshots. You can have AI, you could have like ChatGPT say, hey, here's my landing page. Please change the theme to, you know, all, all the primary colors to, to blue and fuchsia or whatever it is, right? And let it go do that, right? And then, and then it can, it can generate that preview for you. And so instead of maybe doing it on the full live system, you might be able to do much quicker if you just say, hey, here's a couple screenshots, here's a couple options. What do you like better? Like this? Hey, 70% of people preferred this color. Okay, great. I'm just gonna roll with that. And anyway, but if you really wanted to do something live where you're changing between two different designs and you wanna leave it live for a week one way and live for a week another way and, or be able to switch. Or maybe, maybe Sharon, you just give the user the option, green theme or blue theme, right?
Sometimes people like— I wanted to be a little bit more design-led on engagement, on, you know, gamification, things like that really. And even in adverts. So I want to, you know, if you're showing the experience, I kind of want it to be a flow-through so that, that's why I just thought if I had two, I'd turn one off. But it's a case of optimizing that Which one draws people in more quickly?
Yeah, yeah, I would, I would just kind of A/B test it. There's a lot of ways technically you can do it. Keep it simple. Keep it simple, right? Is what I'm going to just— that's all I'll say on that. It's very easy to overcomplicate and focus on things, you know, as part of the, you know, how to build a business, right? We really want to focus on getting customers right. And so I know some of what you're talking about, there is optimization and trying to do that through color and through feel. Um, but ultimately maybe worry about that after you have $100,000 of revenue, you know, cause it just doesn't matter that much. I don't think you're going to be successful or not successful because it's green or blue or because one button is in another spot. Right. Can you optimize those things? Totally. But most people spend way too much time on the little details when they've got this huge, this huge problem. That's just there, right? Which is, I don't have any customers. You don't have customers because you have a bad— or you don't, you don't not have customers because, um, because of a logo or because of a color or because of a button. It's because of your core offering. Fix the core offering, fix the pricing, get people to pay for it. And then when you've got tons of money, go nuts, right? Go nuts and take time to work on those little things. Or even better yet, pay someone to do that. Right? Pay someone that can, you know, study that for you, or just even work inside of Replit. Hire an assistant that can then go work on that project, and they're getting paid $15 an hour while you're out getting customers, right? So, all right guys, let's, um, let's dive into the social media side just a little bit here. Some of the social posts, how to generate all kinds of cool stuff there. We'll go into the General AI Proficiency Institute, and I'll go back here a little bit see if I can find, um, if I can find some. So I wish there was an easy way to search these, these conversations. That's probably something that will come in the future. Um, AI guide, logo generation, slogan creation. Man, I swear I didn't see viral LinkedIn post idea. There was the post idea I was talking about, right?, but I want to actually find— I have one that actually has a lot of social posts. Maybe I, maybe I used it. Oh, it's right there. It's right there. Um, okay, so there's, there's a lot of cool ways, right? So, so one of the things that I wanted to do on launch as I was launching my business is I wanted a bunch of social posts, right? And so I just said, okay, hey, let's make some social posts. Make 10 posts with images for social media to talk about the power of what GenAI is doing, right? So it already knows it's in the project, it knows what I'm doing, and so it says, great, Here we go. Future belongs to AI proficient, future-proofing. Great. Okay, so it creates all these 10, 10 posts. They're pretty good, right? They're fine. And hey, would you like me to generate the images? Yes. Right. Go for that. Right. Then it created all these images and I was like, I don't know if I love these images. Right. And so I was like, hey, also, these are all dudes, right? Give me some, give me some females as well. And then it gave me a bunch of, again, in this kind of style. And I don't know, that just didn't feel very inviting. And so I said, okay, let's, let's forget that for now. Let's create a carousel post, right? Carousel posts are great. You can get people to scroll through and they stay on your profile a little longer. They share it if it's useful. And so I had it create a bunch of text for some slides and things like that. And then guys, what I did is I actually just pulled it all together in Canva, right? You know, you don't have to get fancy and use AI for everything. Now Canva does have some really beautiful AI features now that are starting to get a a lot better. Again, I'm gonna have to log back in because I logged out of everything with my cache clearing activities, um, and installing the new OS. That actually might have done it for me already. But inside of here, guys, you can just say, hey, you know what, I want to create a— I want to create, you know, Facebook, uh, carousel, or Instagram carousel. Yeah, carousel Facebook ad right there, right? Um, LinkedIn carousel Maybe you want that, right? And so it just— okay, here's, here's the style. Maybe I want to do this one right here. And okay, great, apply. Perfect. Now go into ChatGPT and pull all your text and drag it into here, right? Replace your image with something of yours and you're done. You've created something that's super professional looking in seconds, right? And really, it's not seconds. To really do this, you're going to take I don't know, it might take you 15 minutes, right? But instead of having to come up with all these designs and all these things, use stuff that's out there that's already ready and built and beautiful. Let's go back to projects. So yeah, this was one of my, this was one of my ads that I used that I, that I made go live on, on Meta, you know, Meta ads. Um, how many people here have never put together a Meta ad and paid for an ad? Yeah, a couple of you. Most of you have, it seems like. Seems like you have done that. It's not that hard. Um, I can spend time in there, um, and show you some of the AI features, but it's really very self-explanatory when you get in there and do it. You have to have a Facebook page and an Instagram page and maybe a Threads page if you want to do that. Threads is cool now, but, um, There's a lot of people using a lot more than you think. And then just go into the Meta Ads Manager and just follow their thing. Use the AI features, use the AI features that are in there. This is actually a video. Anybody see how to make this video play? Anybody see a play button? Maybe it's just right there. There it is. Okay. So yeah, this is literally just a plane, right? AI is in a distant future. And I had the word distance. I thought of this travel thing anyway, but just like it just provides something that's going to catch your mind. This was actually one of the best performing ads. That I, that I did. Very simple, right? This was written by ChatGPT. What is your AIQ? This little thing right here, a stupid little plane background, but it was just clean. It had the logo there, it had this, it just, it worked, right? And so this is one of the best performing ads. People clicked on it a lot, and I sold a lot of tests and, and certifications and courses just with this thing right here, right? I've run some ads on that. Really fun. Does this look hard to anybody? Like, it's not. It's not hard. It's just a matter of saying, okay, am I using the right platform for the right thing at the right time and then just putting it all together? You build with Replit, you build with ChatGPT, you research with Grok, you build your images with Midjourney and maybe videos with Runway or with Veo or whatever you're using, or maybe just using stock videos like this. And getting some content from ChatGPT, but it's just this workflow of saying, all right, I need to do this now. I'm not great at this. Help me with this, right? Remember what I said earlier? I don't know if you were here, Stephanie, but I think it's an important one and maybe something to think about for a book you're writing, right? If you're trying to take on big things like starting your own business, you have to be a generalist, and AI can fill in the gaps, right? AI can fill in the gaps where you're not strong. Maybe you're great at sales. Awesome, right? It's gonna help you with the images and the graphic stuff that you're not good at. Or maybe vice versa. Maybe you're awesome at Canva and you're like an expert. Ooh, I love Canva. This is fun. But then you just, you don't know how to sell it, right? And so that's okay. Let's talk to ChatGPT. What if we jumped into ChatGPT right now and said, okay, I need you to help me build a sales strategy so that I can get my first 100 users of this app. Interview me one question at a time about this so that we can build a plan together. Right? You don't have to just do it for your business strategy or your business plan. Use it for your marketing plan. Use it for your sales plan. Okay. Who is your ideal first user? Not just kayakers, be specific. Is it a Class 4 weekend warrior in Colorado, a raft guide in California, right? Newbie in Southeast? Probably, yeah, probably a Class 4 weekend warrior, right? And so I'm going to go through and it's going to just interview me and pull information out. What do I need to know? Okay. And then when we're done, it's going to come up with a plan. Follow the plan, do the plan, right? Reach out to those people. Right? Where do they hang out? Are they in subreddits? Are they on Facebook groups? Are they on Slack? Where are they, right? Lots of Facebook groups. That's where they're hanging out. Perfect. Now we have a direct path. Which Facebook groups, right? So again, it's just being an amazing coach. It's not doing the work for you. It's just saying, okay, let's— you know where these people are, you know this stuff. Now let's do this, right? Um, There are different ones based on location. I just want to kind of get to the end here. Okay. Which regions? Let's start with Utah since that's where I am and other paddlers know me. Perfect. Utah credibility. It tells you when blank, so you don't blank. Um, um, fill in the blanks for me. It knows my project. All right, you got it. Best kayaker in the world. That's so funny. All right, so at least it gave me a few options I can choose. It tells you when to grab your boat so you don't stop showing later missing the goods. It says your favorite Utah runs are in. Rivers here run well. Option 2, let's try that. Okay, so it's literally like writing the things that it's going to ask me to then post in a Facebook group, probably, right? So I might, I just might not be strong at sales or marketing. That's okay. I happen to be very strong at that, right? That's what I'm good at. But I'm not great at visual design. That's a place I'm weak. And so I'm going to lean on AI to do that, right? Let's call it, let's call it Eye Water and Be Excited. About it. Okay, um, show me the plan now. Okay, so it could keep asking me questions forever. Sometimes you have to be like, all right, that's enough, let's go. So here's my target user. I need 100 users, paddlers who are already checking gauges, they want alerts on homework. So here we go. I'm going to target Utah Whitewater, Ogden Area Boaters, Wasatch Kayak Club. These are real Facebook groups, guys. I am a member of most of these groups here, and it did not know that. It just, it just knows stuff. I am so impressed at that. Anyway, steps: create a personal launch post. Here's what I'm doing. This is my post. I can obviously be like, yeah, this doesn't feel like me. Let's change it. Let's make it stronger. Replying comments, get people, send DMs to 10 to 15 people. Boom. Hey, that's my first, you know, 25 users. Then, all right, phase 2. By the way, I can do all of this in, I don't know, an hour. Probably maybe 30 minutes, and I've got my first users. Okay, the thing that so many people do— there's two things that people do that, that help them just guarantee to fail at their startup. One, they don't tell anyone about it. They build it and they don't tell anyone about it, right? How many of you guys have posted about something you've built on your personal Facebook profile? Personal Facebook profile, not LinkedIn, not some business page, but on your personal Facebook profile. I'm not seeing a lot of hands here. All right, okay, all right, there's a couple. All right, most people haven't, and, and I understand why. Maybe somehow you don't want people to know that you actually have a job. I don't know. The reality is everyone has a job. Maybe you're a mom and you don't have a job, right? Maybe you do have a job or maybe you're a stay-at-home mom and you have a side gig. How many times have you posted about your side gig on Facebook? Have you done it 10 times this week? If you're broke, you have no reason to not be posting 5 times a day. You just don't have a reason. There's no valid reason that you could give me why you wouldn't be posting about it. Okay? You have to, you have to not be afraid to post about it. There's so many people who have reached out to me from like, I haven't talked to them in 20 years. They were high school, just acquaintance and like, hey, I've seen all your posts on AI and your GenAIPI thing. That's awesome. I own a company. Will you come train our people? I've made $100,000 this year from messages like that on Facebook easily. All right. Don't be afraid to talk about it. Okay. AI can walk you through. The structure of how to do this. And then you just got to talk about it, guys. Talk about it. You can do it in free ways. You can, you can talk about it on Facebook and Instagram and Twitter and, and LinkedIn and, uh, what is the— Blue Sky? You know, wherever you're hanging out, there's other people hanging out, right? Talk about it. Talk about it all the time. Okay. And then the other thing that people don't do is they wait too long to launch. They're embarrassed because it's not perfect, right? And they want to wait until it's perfect. You do not have to wait. Now, there's probably— there's probably a threshold somewhere, and it might be a little bit different for everybody. But honestly, guys, it's probably way earlier. The quote that I love, I don't remember who said it, but if you're not embarrassed when you launch, you waited too long. Okay. If it can do what you want it to do, get it out there, put it out there. It's okay if the formatting isn't exactly right. Now, if it's like glaringly terrible and there's huge font, it's awkward and you're like paper cuts everywhere. Okay, fine. Get rid of the paper cuts and then just push it out there, right? Even if it's free, push it out there, get people to click on it and ask for it. Say, hey guys, I have a new product. I'm really excited about it. Tell me what you think. I've made it free for the next 3 days. You're getting users, you're getting real feedback, right? Even if it's not ready at all. I have— I had somebody who paid me, paid me. Well, he reached out to me to literally pay me $100,000 to create an app for him. Okay. And again, I don't know how to create an app. This is actually before I discovered Replit and all this stuff, right? And so I was just like, okay, I'm going to use my consulting company, right? And, you know, my team, Chris and Taylan and a few other guys that, you know, we build stuff together. And, and I quote him $100 grand. He's like, okay, let's do it. He's like, I already spent $100 grand with another company and what they gave me was terrible. I was like, okay, well, if you want to start over, like, I'm happy to work. He's like, yeah, I've seen your work. It's great. I want to work with you guys. And I was like, okay, fantastic. Like, let's do it. And like the day we were supposed to sign that agreement, I got on a call with him. I was like, hey man, I don't think you should do this. Because he was literally— he told me, he revealed to me, I didn't know this at first, he said, yeah, I'm borrowing money from my parents and they're taking a second mortgage out on their house. And I was like, okay, like, this is a bad idea, dude. Like, I— you need to prove this. I said— I asked him this question, I said, have you ever, ever talked to one person about what they think about what you've already built? He said, no, I haven't shown it to anybody. He spent $100,000, hasn't shown it to one person. That's insane, guys. That is insane. You have to get it out there. It doesn't matter if the green weird design isn't perfect, right? You can put it in front of your friends and say, hey guys, ignore the design, we're still working on that. I want you to just focus on the content. And this idea. What do you guys think? Would you pay for this? What would you pay for this? Is this worth $10 a month, $20 a month, $5 a month? Should it be free? What is this? Would you use it at all if it was completely free and I handed it to you? Would you use this, right? Get it in front of people. Don't be afraid. Don't be afraid to post about it. Don't be afraid to post personal things on LinkedIn and professional things on Facebook. You will stand out. You will stand out, guys. I really— this is so critical. Let's talk about this for a second. Now that anybody can create an app, is there value in creating an app? Sharon's saying yes. I think Jeff was like, maybe not. Yes. Okay. Yes. Hey, we got some like, I don't know, what are those? Got some explosions behind him. It's exciting. Some fireworks. I'm going to argue probably no. Why? Because anyone can do it. My 11-year-old can recreate your app, Jeff, literally, because he's taken my course. He's, he's sat in on one of my first courses I did here. He knows how to use Replit very efficiently. He can make amazing websites. He's super smart. He's 11. Okay. And he can create an app as good as any of you guys, as well as any of you guys. Now that might not be 100% true because Jeff, you know, your specifics to your industry, Sharon, you know, your specifics to your industry way better than he does. He's just a kid, right? But the value has moved. It is no longer in the software creation. It is now in the marketing. Stephanie, this is critical. This is going to be a critical piece that we need to cover and we need to discuss with the team because if we teach a bunch of people how to make cool stuff with AI, they're not just going to magically have customers. It's not build it and they will come. That does not work. Build it and they will come does not work when everyone can build it. If you're in the small 0.1% of the, of the, you know, population or 1% of the population that knows how to build stuff, well, maybe build it and they will come applies. But now if 90% of the population can build it or 40% of the population can build it, now you just have anyone can build anything. Right. And so you, you have to, you have to really pay attention to where is the value created. Right? The value now is going to be created in distribution. So the people that have big followings, they're going to have huge advantages, right? I'll share with you guys one thing that Stephanie and I are working on, and we can't share a ton of details, but at a very, very high level, I built GenAIPI starting in February. I've been working with AI for years, but really kind of officially kicked it off 6 months ago. And about a month, maybe 2 months ago now, I was reached out to— by, by Dean Graziosi and Tony Robbins. And they said, hey, we're building some AI stuff, we're going to go help a bunch of entrepreneurs, you want to come join us and help us, help us guide to make sure we're teaching the right things? And I said, you guys have the audience, right? I understood this already, right? You guys have the audience, I've got great content for sure, but my audience is a thousandth the size of yours, maybe a ten thousandth the size of yours, right? And so Absolutely. Let's go. How can I help? Right. And Stephanie is helping as well. And so, Dean and Tony, they'll announce it'll be more official. There will be some great, you know, free sessions and lots of stuff. And there'll be books and training courses and things like that and all that stuff that they do, right? Those guys go out and they find the best in the world, right? The best experts in the world, right? Way more expert than me, right? About everything they do, whether it's money or whether it's health or whatever it is., and then they go and they share that with the world and they do a great job at doing that. And so they're, they're now doing that for AI and, uh, it's really exciting, really exciting to put that together. But is their product gonna be successful? Absolutely it will. Why? Because they have the distribution, right? They can absolutely demolish GenAI API because they can make as much money as I've made so far in an hour, right? Because they can get on stage in front of 600,000 people. And show them in, in very genuine ways, just with a huge audience, right? How they can use AI to dominate, right? And how they can do that. But, but it's very important. It's very important, guys, that you realize that now that you know how to do all this stuff, now you still are ahead. I still put you guys way ahead. Most people don't know that you can build software like this. They don't know that you can build a whole business in a weekend. They just don't know it. And that's fine. That's totally fine. You guys are ahead now. You have this advantage. You've been given all these tips and tricks and tools, and there's so many more that we don't have time to cover in this, right? I might show a couple more things, a couple more tools. I encourage you guys to follow me on social media. I'm posting a lot of content on LinkedIn, Facebook, Instagram, whatever. I usually post the same stuff across all my profiles, so use whatever you like best. Follow me there. You'll see some of the things I've taught in this course, um, in some of the videos I've been releasing. I'm doing a series of 20, 20 days in a row. They're like 5 to 10 minute videos of just like, hey, check out this tool, here's how I use it, right? I've already introduced Replit, I've introduced, uh, Midjourney, I've introduced ElevenLabs. I'll show you guys ElevenLabs in a minute. But, um, but anyway, guys, um, keep learning, keep following, keep, keep trying to build this, build this out. But what I hope you've gotten out of course is a new way to think about things. I really hope you believe Denise and Emily and Sharon and Stephanie, Jeff, everybody. I hope you guys feel like, okay, I know how to ask questions, and because I know how to ask questions, I know how to find answers. AI is the most powerful answer giver ever other than God, in my opinion. That's a different type of answer and for different purposes. But if you need to know how to do something, whether it's sales, marketing, building a software, connecting to a new system, maybe you want to create, maybe you want to create and do something like, like this guy does. This guy is the best at social media, bar none. Nobody even comes close to touching him. There's a lot of great people out there, but Gary Vee has it figured out. If you want to learn how to do social media, well, buy this book. I won't make any money off of it. It's $20. Buy the audiobook, whatever, whatever you want. Day Trading Attention by Gary Vaynerchuk. You know how many times he posts on social every day? 60. That's crazy. And he did that early on and he did it by brute force, right, to start, because it was just— that's a ton of work, obviously. Now he has a team that helps him do it. But I could show you guys, and I guarantee you could jump on ChatGPT and say, hey, let's put together a system that will allow me to post 60 times per day. Now that doesn't mean it's 60 unique posts. Maybe you're posting 6 times on, on LinkedIn and, you know, 20 times on Twitter and 4 times on Facebook and 3 times on Instagram and whatever. We add it all up, 60 posts across all platforms. Okay? You could create an automated marketing engine that just does this for you. You could use N8N. That's a great tool that we kind of introduced just a little bit today, right? You could use Replit to build your own custom one for your business that automatically uses an API to ChatGPT, uses an API over to Midjourney or to Runway or to Veo or whatever your, you know, whatever image generator you want to use. Um, use an API to then take those posts and throw them into a Google Drive, right? API them into the Google Drive. From the Google Drive, API them into a, a system like, like Buffer. All right. Anybody here heard of Buffer? There's a lot of systems out there like that, right? But anyway, this one allows you to basically from one place publish to, you know, in this case Threads, Bluesky, or is that, I think that's Bluesky, Instagram, and TikTok all at once. Right? And so you can schedule the post. And so you can API right into this. So you have AI create a bunch of stuff based on prompts and all of your information and your business. And every day at whatever time, it creates all the posts. The AI creates all the posts, sends them to a Google Drive. Buffer says, hey, I see stuff that— actually, you would use something like Zapier, right? You'd use Zapier or N8N, right? You can use whatever you want, right? But, if we Export Apps, and then you go to like Buffer. Um, you can then create a Zap with Google Drive, right? And say whenever something happens in Google Drive, set up posts in Buffer. Boom. Now you've built yourself an automated system that can post 60 times per day about your business. That's pretty cool. Now I don't need to walk through all those steps with you guys because I'm just trying to help you think maybe a little bit differently now, right? With AI and with all of these automation tools. A lot of these automation tools, for me even, even though I'm kind of technical and I kind of get it, right? Before AI came along, it just felt so out of my reach. I was like, man, I just don't know how to do that. Like, there's, there's some things that are just a little complex. I don't know how to build an API. I don't know how to set that stuff up. But now that I have AI, I can just say, here's what I'm trying to do. I'm trying to build a system that will allow me to post 60 times per day. All right. And here's the, here's the count. You know, maybe it's 30 times a day or 10 times a day or 1 time a day for that matter. Right. How many of you guys are posting 1 time a day on social? Nice, Sharon. That's pretty good. Most people aren't right. I like— it's hard. It's hard to post 1 time a day. Right. And I do it. I do it religiously. I do not miss it. Because it generates a lot of revenue for me, right? I make a lot of money from just being active and talking about things, and then people ask me questions because I'm an expert on it, because I talk about it a lot. In order to talk about a lot, I gotta learn a lot about it, right? And I don't actually use AI to do this, right? But you could, and I could— I should do this for GenAIPI because I'm terrible at posting to my, my, my business profiles, because it's already hard for me to just do my personal profiles every day. But maybe I should set this set up an automation to be able to post and maybe it doesn't have to be 60 times a day, but maybe it's 12 times a day across my different profiles and just talking about different things that are happening and different features and different things and different thoughts, whatever. So that people say, hey, this guy gets it. This guy is trying to help employees learn AI, right? Or entrepreneurs learn AI, whatever. So are you starting to think a little bit differently, guys?
Stephanie, go ahead. So I've been in marketing my entire career, so obviously this question of how do you get visibility, how do you grow an audience, that's like age-old question. That's not like a new thing, right? I mean, we all know that. But now when you're saying that anybody can do all these things, so what makes it more probable that you will be successful is your audience? How can you use AI to do that when everybody also has the same tools and the ability to use AI to build an audience?
What is the thing that sets people apart to be able to sell your thing to more people? You and me both asking the same questions. Like, that is, that is the question. That is absolutely the question, right? When everybody can do it, how do you stand out? And the reality is, let me just respond with this. How many people today can create a video and post it online? Basically everyone. As long as they have one of these, they have all the tools they need to do it. I know I'm not saying like a fancy edited video, just like, here's a video, put it on LinkedIn. Here's a video, put it on Instagram, done. Like super, super easy. And if you want to get a little fancier, you can open up one of 1,000 editing apps and make it a little bit better, add some subtitles, add a transition, add a logo, whatever you want to do, right? And any of us, all of us are capable of doing that. Again, my 11-year-old is capable of doing that. Well then, why do some video creators stand out, right? Better content. It's just better. They've been able to deliver. And, and I would also argue that the better content and consistency is how most people win. Consistency is actually one of the most undervalued, um, elements in people's success, right? My content might not be amazing that I post on everything, but I do it every day. And so because I do it every day, I'm naturally just stacking my days. I'm getting smarter. I'm getting better. I'm seeing what works. I'm subconscious. Even if I'm not, you know, going to, okay, that was 21,000 impressions. That was 13,000. That was 100,000, right? Like even if I'm not actually measuring those things, I'm subconsciously like, oh, hey, that one actually kind of worked well, right? And then my brain is just going to get better at it over time because I do it a lot. And then if I want to put out good content, I'm going to do it. So anyway, That— I'm not done answering that question, but Jeff, what's your question?
Well, I mean, just to, to kind of go off of that, and, and similar to this, hey, if everyone can do it, what's, you know, what value does it still possess? Um, I've been a videographer for the last 15 years and worked in film, um, since, you know, before the 2000s. I'm trying to remember, my first gig was on a film crew, 1999. Um, it's amazing to me to see the technology has increased so rapidly and it's now ubiquitous. Anyone can have a cinema quality camera on their phone. Um, in fact, the, the new movie 28 Years Later, which was a follow-up from a Danny Boyle film, was filmed all on an iPhone. Yeah, a major Hollywood film. Um, however, it's interesting as I still follow all these video subreddits and things and people are posting their work of, hey, look at this job I did for this tattoo parlor, look at this job I did for this wedding, or whatever. And the shots look beautiful, and it's— there's perfect zooms and sound and all these other things. But the problem is, is that the basics, the fundamental basics of human communication and storytelling, are missing. And they don't know how to tell a story. They don't know how to connect to another human. They don't know those things. So where It's easy, and I've found myself falling into this trap during this course of looking at all the technology and getting so excited about all the things it can do for me, and I'm forgetting sometimes, all right, well, what's the, at the end of the day, what am I actually trying to do? I'm trying to simplify this, I'm trying to make this better, I'm trying to connect with this person and have them have a good experience. And so when I lose sight of that, it's important to remember, hey, the, the, you know, the most important currency is trust, right, between me and these people. The most important thing about my video work is storytelling. Am I telling a story? Am I getting action to be taken? So anyway, that to me has been a comforting thought as I go through this, of not necessarily like, hey, I've got this skill set, so I'm okay, but it's a good check to put yourself into of, yeah, look at all these AI tools, look at this thing that I've built, but am I, am I still doing what got me here in the first place, which is the ability to be able to communicate and touch someone's heart person to person versus the cold steel of technology.
So spot on, Jeff. And I'm thinking back to what we were talking about yesterday about creation, right? Creation is still something that we can do in a way that AI can't. Right? And that's why I love, for example, as one example, that interview, like interview me about this, right? Help me, help me design a plan here because, because your knowledge, Jeff, on how to create that story is critical. And so if you're like, hey, help me, help me put together a quick, you know, you know, commercial shoot, right? Ask me all these things. Ask me this. Ask me this. Ask me this. And then you're going to be like, okay, now this is what we want to do here. Yeah, we're going to do this and here's how we're going to connect here. This is what we're going to do here. And then all AI is doing is just organizing and, and, you know, it's speeding up. You're moving faster because of AI for sure. But you're still the boss. You're— it's still going to be a successful commercial shoot because it has Jeff's fingerprints on it, right? All over it, right? It's not just this AI thing. And so to use that to go back to what you said, Stephanie, right? Or you were asking Stephanie, how do you stick out? Well, you still got to put in the work. You still got to work with experts. You still got to learn. Now, again, as an entrepreneur like me where I'm trying to sometimes do a lot of things by myself, I can use what AI knows that all the experts know, right? I can say, hey, you are an expert marketer, right? Or you are excellent at connecting with people with video or whatever it is, right? And I can then, you know, try to have it coach me and guide me through that. But I'm still going to have to— two things are going to happen. One, I'm still going to have to do a lot of work. Work to follow its guide, right? None of these outcomes are just simple and automated and done for you, right? You still have to put in the work. But then too, I'm actually going to get smarter about that thing. I'm going to learn as I'm, as I'm going through and asking about what would the experts say here? How do I, how do I create a video that connects better? How do I make this shot really like shine here, right? And, and AI knows that stuff. Now the reality is, guys, And I know this is, this is kind of, you know, going against what I just said now, what we're talking about, but AI is going to get better. Right now is its worst, is the worst it's ever going to be, right? And so the stuff we're seeing right now is like, oh my gosh, this is crazy, right? In the future, you won't have to babysit Replit to get through all these steps, right? 5 years from now, just be like, hey, build me an app that does this, and it's just like, done, first shot, it works. And you don't, you know, you don't have to go and fix all these things and ask it, you know, what model it it's doing and whatever. It's going to get better. It's going to be able to create videos better than humans. It's going to be able to create music better than humans, to be able to write better than humans. It's going to be able to do so many things better than humans. Just is, right? And even if you don't agree with me, let's just roll with that assumption. If that is in fact the case and we get to ASI in 6 years like Eric Schmidt said, now how do we stand out? Now, how do we stand out? How do we beat AI? How do we provide value as humans? I want to talk about this for a few minutes. What do you, what do you think? How do we create value as humans? I'm not saying I have all the answers. I'm asking the question. That's it, right?
I have some ideas, and I really, really want to hear what you guys have to say. Stephanie, can I add to that, like, also posing a question, kind of going off what Jeff said as well? We are creators in this group. We care. Like, there's a contingent of us around the world. We care about quality. We care about storytelling. What happens when a majority do not anymore? Because, and I'm going to that shiny object syndrome, this is amazing. I just saw Adele do a duo with Eminem. Well, that's not real, but that's amazing. And so what do we do when somebody no longer cares about the original intention and purpose of art? Because I do believe that things are magnifying to the, the sense that people are like caught up in the glow and glitter of this, and maybe that will shift. But as a creator, I am concerned that the people who consume what I'm creating may not care as much as they once did because they can, you know, have access to the explosions and all the other things rather than the core of something. And this is what I'm afraid of as a creator, as a human, because I believe, Jeff, everything you said is correct. The human element is crucial.
But what happens when that gets dissipated by things. Let me give an example, and then Sharon, we can have you comment. Um, and I'm not saying this is the answer for everything, but it's just an example that's coming to my mind. Um, McDonald's is generally a, you know, it's fast food, it's supposed to be cheaper, it's supposed to be more accessible. And, you know, maybe for, for some people it's, um, you know, maybe that is their nice meal out, right? And just depends on who you are, and it's going to play its role. But bottom line is it's not trying to be this like white white glove service, right? So I want you to think about the future of McDonald's as compared to the future of Ruth's Chris, right? Ruth's Chris is going to be white glove humans all the way through the process, right? Waiter comes to your table, asks you, takes care of everything, it's done, you're sitting down in a sit-down restaurant, that's awesome. And then you're gonna have the McDonald's, which are going to And I think Ruth's Chris is going to lean in on that. That real human experience is going to matter more. It might even be in their marketing more, right? Whereas you're going to have McDonald's, which is going to switch to either a fully automated or maybe a one-man shop, right? Where everything else is done by AI and robots to cook your burgers and to take your orders and to deliver everything. And the voice you talk to in the drive-thru is AI and everything, right? And so what will that do to McDonald's? Theoretically, it could make food cheaper. Now they don't have to pay employees. They're just paying machines and maybe one employee to, you know, feed the machines or whatever it is they have to do, or maybe completely automated. You only have a human coming and checking on it every once in a while. And it'll take a few years to get there and to perfect that model. But I know, I know for a fact right now that Burger King is working on that. I know that McDonald's is working on that. I've read articles about it. Okay. And so you're going to have this automated world that's cheaper and fake a little bit., right? And it's not, it's not that the, the Big Mac is fake. It's actually the same output, but it's just you get rid of the human element and therefore it becomes cheaper. Humans are the most expensive thing. And then you have Ruth's Chris, which is going to say, hey, no, we are the real thing all the way through. And so there are— I think Ruth's Chris is pretty well, well positioned already to deal with this AI change, right? Um, will people pay more for a real Ed Sheeran concert than, you know, in person, live, than a fake AI one that's just as good. The music is just as good. You put an android on stage that looks like him and dances like him and sings like him or whatever, and maybe people will pay for that. I don't know, maybe they won't. Um, but, but there's definitely going to be fake and real. It's going to be, it's going to be, um, something that continues to to expand, and it's going to get more obvious in some ways, and then maybe not in some ways, right, Stephanie? Like you said, like that example of, hey, did you see that Taylor Swift with whoever, you know, Eminem video? Oh, well, that's not even real. I don't care, I love it. Where can I download the song?
You know, like, I get that. Sharon, go ahead. I am— I haven't got the— this, you know, to come out in one go. I'm sort of saying on the fly, but What I'm trying to express is that I think, you know, the world is moving more and more towards authenticity and being authentic. So whilst I totally agree that the synthetic or, you know, the fabricated, all those examples are still great, they're entertaining, it's no different to digital media and things like that, it's just loads more of it and loads more personalization of it. I think it just means that there's a lot more before people still want then the authentic. And they still want the organic and they want the real experience. And I think it's just more noise, you know, it's just another collection of content before we then crave where we personally really crave those things. And that's different for everyone, what that is. So I don't think that in many ways it's changed massively. In many ways it's not changed at all. So that's my thinking. But obviously the displacement of workers and what they're going to do and whether— I think I said yesterday in terms of are they gonna sit there and are we gonna all learn around and be a bit like the philosophers and play more? And we're just gonna create and we're all gonna be walking around on our togas. I don't know. 'Cause it's like, there's gonna be an awful lot of people that, you know, are low skilled that have not got jobs or the classic jobs. They won't be replaced enough. You think about that pyramid of McDonald's fast food, vast majority, you know, all of a sudden that's one to many or one to millions. It's a, you know, how does that work economically?
Don't know, but I know we'll figure it out one way or another because we always do. Yeah, we all— we do. The thing that, that I worry about, guys, is the systems. Our systems, our societal systems, right? Just tax, right? And governments and welfare and, you know, those types of things that are set up right now to help a certain group. They're really built in a way that only works if you have a group supporting them. Right. And if those ratios change by 20%, like, society falls apart. The systems don't exist today. We don't have a society on planet Earth that can absorb that kind of a change right now. And that's the thing that's just a little bit scary for a lot of people. And then for me, just personally, like, again, back to the Spotify example, example yesterday, like, there's a million songs uploaded every day and something like 40% of them are AI. How do I compete with that? It's deflating to me a little bit, other than the fact that I'm pretty resilient and I'm a go-getter and I'm going to make it happen, right? Like, I'm someone who's just going to figure it out. But I feel those emotions and I empathize in a very real way with people that are like, I can't compete anymore. Right? Like, I— how am I supposed to keep up with that, right? And so, so yes, we're going to have to figure out how do people eat, how do people have a roof over their heads. Um, the reality is AI theoretically should be able to produce more output as a whole for humanity. Humanity should be able to produce more because of AI. But when you then break it down to the individual, the family, the individual the individual households. How, how do we actually navigate that? I don't have the answer. I don't have the answer. Um, but, but when AI can do all these things better than everybody else, where does the value come from? I think, I think we've already hit it. I think we've already hit it, and it's, it's humanity. It's the humanity that, that, uh, the human connection, the real human connection whether it's being able to thank the chef personally at Ruth's Chris or talk to the waiter and have that personal connection there or sitting close to a live performer at Bluebird Cafe in Nashville or wherever it is that you're having that personal connection, therapy, travel, adventure experiences. There's a lot of places where there's where AI can't touch it. AI can't touch a zipline, right? I still am going to go want to do that, right? And there's going to be people that have to operate that zipline, and there's going to have to be people that sell and market that zipline, and they're going to have to be— so it's not that there's going to be no jobs left, right? But will we move more towards less repetitive, boring work that robots and AI can be trained to do, and more towards meaningful connection with other humans? Across every industry. And to me, that's exciting, guys. Like I said, the transitions— I don't know how it's gonna happen. Like, it's gonna be weird, and there might be, there might be a lot of struggle to get there. But I think the end, the next phase of humanity, is we really get to focus on what matters most. And that's exciting. So in the meantime, you know, you guys are now the latest graduates of this program. And, um, and hopefully you can take it and you can— I think there's an opportunity for the next 5 to 10 years right now, maybe 3 years. I don't know the timeline. I could argue it a few different ways. I almost sneezed. Anyway, income. Anyway, so I can argue a few different ways, but I think there's an opportunity for you guys to make a lot of money right now. Jeff, the fact that you could whip up that thing right now and now it's live, like Nobody knows you can do that. The vast majority of people in the paint world have no idea. The paint world is already behind technologically. You know that already. And you guys are at the forefront and now you're going to be pushing things out so fast, your competitors can be like, how are you doing this? Right? I think there's opportunity there. I think you make a ton of money right now. Save up during the years of plenty. Right? Because who knows what happens once everybody figures out you can do that. Okay. But, but, but honestly, guys, Take, take what you've learned here and go and do awesome things. Now I want you to continue to learn. Um, there's a couple ways. Obviously you can follow me on social media. I'm only one little, one little blip, right? There's so many people that I learned from. Um, if you go and you follow Gen AIP— I think it's Gen AIP Institute— um, on Instagram and other places, whatever. Um, specifically on Instagram, go to the people that Gen AIP is following. Those are the people I follow to learn from. Okay, there's like 50 accounts on there. You don't have to follow all of them if you don't want, but go in there. There's a few really good ones in there. If you want to know specifically, just ask me. Um, I'm going to send a follow-up email after this, after this class, sometime today after the YouTube video is uploaded and everything. Um, and I'll, I'll include both YouTube videos, one from yesterday and today. I'll include a link to that Google Drive folder that has all these amazing tips and tricks and things and workflows and all kinds of fun stuff you can, you can look through and use and refer to later. There's also a Facebook group that's called Build a Business with AI. There's only about 100 people in it right now. Not everybody joins it because not everyone has Facebook, or maybe they don't want to join it for whatever reason, but most people join it and, and there's good conversation in there, right? Lots, lots of really smart people who have also been through this course, through this training, and they're really smart. Smart, right? If you have a question about anything, you're building in Replit, you're working with Midjourney, you're doing VEO 3, you're trying new platforms, you're looking at stuff, they're, they're talking about all the new stuff, they're trying stuff, they've, they've experimented a ton for months and months and months. And many of them have gone on to make 5 and 6-figure businesses already just in the last few months, right? From taking this course, they've gone out, they've launched, and now they're live and doing really, really cool stuff. So you can learn from that. I'll also include my cell phone number in the, in the follow-up email, right? You guys are part of my, part of my circle now. Feel free to text me if you have a question. If you need me to work like for a long time, I'm like, hey, help me build this platform and I need you to work with me for 2 months. I'm, you know, that's that. Now we're moving into consulting work and that's fine. I'm happy to, I'm happy to do that in under the right circumstances. I've said no to a lot of people, but the ones I like to work with, I'll say yes to probably. And then if you can't work with me, there's other people in that group in the Build a Business that are doing stuff like that too, doing some consulting. And so, yeah, anyway, I'll also ask you right now and in the email again for any referrals, right? You bring anybody to this course or, you know, refer people when they sign up, I will, I will just Venmo you $100, right? So just, you know, send it my way, send a group text. Really don't, don't feel bad using my phone number. And really, if you have a question about something like, hey, Jon, I'm working on something, I can't figure it out, you know, what would you do here? Don't feel bad to send me a message on social media or on or via text or email. You know, anyway, really excited for all of you guys to take this stuff. I hope this was helpful. I hope it was valuable. I feel like you guys now know how to do stuff that, you know, the $1,000 or $1,250 price point or right, of this is supposed to be like a steal of a deal in my mind, right? The types of things, the amount of money you can save now by being able to use AI. Now you don't have to hire people to do a ton of things. You can actually launch a business for a few hundred dollars and, and turn it into money. You just got to get out and you got to ask for it. I also, guys, am working right now on a self-guided, a self-paced version of this course that is, you know, it's probably going to be more like 30 hours long. It's going to be really long because I'll just go into depth on each topic and it's going to— I've started recording some of it, but I'll give you guys access to that for free. You guys will all, since you've done this course, you've been in the inaugural, inaugural year of, you know, launching some of this stuff. You know, I'll make, I'll make sure you guys have access to it and, you know, I'll sell that I'll sell the self-paced course for less probably and let people, let people go and do that. But, but also it might all change, Stephanie, because I might, I might end up shifting a bunch of that effort over to, over to AI Advantage. So that's, that's part of what's, what's exciting now. But anyway, guys, you're going to keep seeing a lot of AI stuff from me. The world's going to keep putting out cool AI stuff. The progress of this— I want to share one more thing that I think is pretty interesting. Anybody here familiar with Moore's Law. Yeah, so Moore's Law basically states that, that like processing power and storage capacity and all these things, every 18 months doubles in its capabilities and therefore becomes, you know, if it was $100, now it's only $50, right? You want to buy a 1 terabyte hard drive, you know, approximately every 18 months those prices cut in half and therefore become more available. And so So the power of computing just continues to expand at about a rate of 2x every 18 months. AI increases right now 10x every 12 months. Okay, so let's do a little bit of math here. In 3 years, right, Moore's Law, 18 months, 18 months, 2 times 2. There's a 4x increase. In 3 years with AI, 10 times 10 times 10. It increases or becomes more available by a factor of 1,000. It becomes 1,000 times more accessible to the public. AI is progressing faster than anything we've ever seen. It, it's, it's, it's hard for us to even fathom the difference between a 1,000x improvement every 3 years versus 4x, right? It's moving 250 times faster than CPU power and storage power and all these other things. Now, we are limited a little bit by electricity and some things like that, and there's lots of problems to solve still, but it's moving very fast. And hopefully you guys have felt that. And, um, anyway, I think we're done here. I'm happy to stick around for a few minutes if people have questions, um, and then we'll, uh, you know, we'll finish up. So you feel free to drop off if anybody has any questions, I'm happy to stick around for a few minutes if you want. Thank you so much. I'm going to drop off, but thank you. Okay.
All right.
Thanks, guys. I don't see any questions, so we'll just see you guys. I'll send these recordings out and we'll go from there. Thank you very much.
I have a question. Jump in, speak up, or hold your peace. Sorry, I, I put them in the chat. So my questions for you are, do you use— can you use Replit to build funnel pages and flows as well, or is it mostly just like, like a website?
Yeah, tell it, say, hey, um, say build me a Russell Brunson style, um, ClickFunnels. Boom, done. $900 ClickFunnels. You just saved $1,000, right? So yeah, totally. You can absolutely, you can absolutely do, you know, funnel pages and flows. And one of the things we didn't go over, guys, and it it's a little bit tricky, um, is connecting like Stripe, connecting payment to your system. It's not that hard, right? It just requires some testing, right? So set your products to $1 or $0.20 or $0.50, whatever. Charge yourself, buy your product as a normal customer, go through it, test it until it works. And it might take you $30 of testing, but the good news is you're paying yourself, and so you'll get deposited minus a few fees, right? But, uh, bottom line, it only costs you a few dollars to do your own testing. But, but when you're building funnels, you need two things working well: payment and email. Okay? And I recommend for email, I recommend a system called SendGrid, right? Which is owned by a company called Twilio. Twilio does texting, SendGrid does the email, right? But those are my, those are my things. So like when you get an email from GenAIPI, like a password reset email or a purchase confirmation or whatever, that's all coming from my SendGrid API. And, and it will— and by the way, if you didn't remember that, you could just say, hey, Replit, I need, I need to set up a forgot password and some purchase confirmation emails. What can we use? And it might say SendGrid, it might say something else. It'll say, go here, get your API, I'll set it up for you.
Just, just work with it, it'll help you. Yeah.
What, what CRM do you hook into, like, for your database? Um, well, a database is different than a CRM, just to be clear. So CRM is going to be like, okay, here's my companies and my contacts and my deals and all that stuff, right? Like a HubSpot or Salesforce, right? That's really kind of more of a CRM, client relationship manager. A database, like we're talking about in Replit, those are all the pieces of data that your, your, your website needs to run. So for example, for my website, my river one, I'm gonna have river name, river flow, river location, river, whatever, right? You're not keeping that in a CRM. That's just data that helps my app run well. Okay, so, um, so anyway, feel free to, feel free to dive into, into those and play with the different, different things. But you could create your own CRM if you wanted, um, that's personal to your business. I haven't done that for GenAIPI. I just use Google Sheets. I just, you know, it's just something I haven't done yet, and, um, and it's working right now. So, you know, if it ain't broke, don't fix it. But, um, but at some point I might say, okay, hey, I want to automate this flow so I don't have to manually put these things here. And I might do that. The other thing, guys, that we didn't talk about a whole lot, and I'll just finish it up here. Maybe some people will watch the recording of it at the end, but the whole concept of surf, don't swim is important, not just because you need to figure out how to surf on things, but it's because sometimes it's better to surf than to swim, right? Like for example, how many of you guys think you could rebuild Calendly, right? Do you know what Calendly is? Calendly is just a basic service where you just, you send someone a link and they can schedule a meeting with you. Do you think you could build that with Replit? Piece of cake, right? It would take you 30 minutes. Okay. You would just say, hey, Replit, build me a Calendly clone. And it would build it. But then you have to build all the features that Calendly has. And when Google changes their calendar permissions, then you have to update it and you have to keep Now you have to be responsible for it, right? So instead, why don't you just surf on it? It's $10 a month. Just pay Calendly and now you don't have to worry about it anymore, right? That, that's a really critical kind of the build versus buy, surf, don't swim, um, approach you have to figure out. And so yeah, sometimes you want to build your own CRM and maybe sometimes you don't, right? A CRM is actually one of the easiest things to build. It's just a basic, you know, relationship database and Repliq can build it.
In no time flat. Do you have any guiding principles that dictates when you build something versus when you just hook in another software that's already designed?
Yeah. It's going to be a conversation that I'm probably going to have with AI about it. It's a cost consideration. Like I was talking about that AI SEO thing, trying to figure out, hey, how do I get LLMs to suggest GenAIPI is a great training program? And so I found this solution that was $700 a month, and I'm like, that's crazy. There's no way I'm going to pay $10,000 a year for this thing. And I went to Grok and I was like, Grok, there's no way they're doing anything crazy. Like, no, they're doing this, this, this, and this. You could totally do it. I was like, great, replicate and build it for me. So there was my, you know, surf, don't swim. I was like, yeah, I'm going to— I'm just going to swim on that one. It's going to take me a little bit more work because now I'm responsible for it instead of outsourcing it to to somebody else. But that one was easy enough. It's really just kind of a calculation of cost versus time. And it's going to be different for everybody. If I'm an exited entrepreneur, maybe I'm okay dropping $10,000 on something and you might be like, "You know what? I'm willing to put in some work. It might take me a week, but I'm going to save $10,000." And I'm not saying I just blow money. I'm still probably at $10,000 $1,000, gonna build it on— or I'm gonna at least investigate what it's like to build it on my own. But everything you build, you got to maintain, right? So that's something that you just have to remember.
If you build it, you're now responsible for it. Yeah.
And then you mentioned the, like, like the voice calling service, and you didn't touch on that. I know, I wanted to get to it, and we just kind of got into some conversations. ElevenLabs— I'll share, I'll share my screen for just a second. I did a video on it just literally last night.. And so I want you to go watch it. It will be— it'll be— just go to my LinkedIn, Jon Cheney, follow me on there, whatever, um, or don't follow me, I don't really care about that, but go to this video right here, Day 6 of 20, okay? And, um, and watch this video. But this one talks about, you know, um, that I use ElevenLabs and what it can do. And I go through— it's an 11-minute video or 12-minute video almost, um, and, and I The state of the art. It's awesome. But ElevenLabs basically, guys, is a voice platform that is incredibly easy to use. You can create your own voice, you can clone your own voice, you can use other voices, you can build agents, you can do creative stuff. What I mean by agents is this, let me just show you. Here's a demo. Let's do Amira here. Amira is an agent that I built. And by built, I mean it took me like 20 minutes to set this up, guys. Okay? And I built it for a, a, uh, a company or a, uh, an Airbnb, kind of a destination place called Outpost X. And it's kind of like a Star Wars feeling, you know, amazing kind of glamping experience, sci-fi glamping experience. There you go, right there. Kind of near Zion National Park. So if you're in Vegas, you can get to it pretty easily. Um, it's amazing. Like, I took my family there and it was so cool. They had like speeders and stuff. It feels like you're in the middle of Star Wars, really cool. And he said, hey, you know, I know the founder, and he's like, hey, could you build me like an assistant that can help people answer questions like, hey, how do I rent the speeders, what's the Wi-Fi password, where do I get food, where's the bathrooms, those types of things? And also like make it feel really cool. I'm like, sure, let's do that, right? So I built this ElevenLabs assistant here, right?
And so you can talk to it.
I'll just give an example. Welcome, traveler. I am Amira. How shall I guide you?
Hi Amira, I am hungry. Where can I eat here? The cantina kitchen is always open and available to all guests.
It is stocked with cookware, stovetops, and a dishwashing station. Tent guests will find labeled— Anyway, I can talk to it about whatever I want. So it's now an assistant that can help people at the, you know, at the resort. And I could also ask it questions about, hey, tell me a story about, you know, the history of Outpost X. And it has all this lore and it's about, It's like you're on another planet and they have all these people and characters and stuff. And all I did was take all— they just— they, um, Outpost X just gave me a Google Doc that was like 20 pages long of all this like content they'd created. And I just took it, I handed it to this ElevenLabs agent right here. I'll show you exactly how I did it. I uploaded it to right here to the knowledge base, right? I just clicked add document. I added some files and boom, it's done. Here's the basics, Wi-Fi passwords, all this stuff, done. Now I've created a custom agent that feels like you're talking to a normal human being. It's really pretty amazing. Jump in and play with it. You can create your own agent that has your own voice. This is my voice. Hello, how can I help you today? Hey, I just want to learn about AI. Can you teach me something? I am designed to help English speakers with translations between Chinese and Spanish. Yeah, so I actually did give it those instructions, so I'm glad it listened to me. But, um, but anyway, I created my voice, a clone of my voice, um, in 20 seconds, right? I just recorded 20 seconds of audio. And, and by the way, I do that right here. I come in— sorry, I come over here go to Create a Platform, go to Voices. I click Add. I say, okay, I'm going to clone my voice, and I record audio. Start. Hey, this is Jon. I'm just recording my voice so that you can create something that sounds like me. And I'm going to talk really fast for a second here, and then I'm going to slow way down, and I'm going to say Z and X and Q and P and give you a few few different sounds, and we're done. Okay, and I've now done all the work that it takes to create the agent. I click next, and time well managed is the secret to creativity. It's done. That's how hard it is to create a voice, right? It's amazing.
By the way, this platform costs $5 a month. Wow.
Yeah.
Can— is it possible to like clone your voice and clone your body? Like if you— if If like you have like a call that you wanted it to like show up on Zoom with, it's like very repetitious sales call.
Like, can it do that or is that not a thing yet? It can. The hardest thing there is connecting it into Zoom, right? Is getting your video to display properly and getting that all set up. I didn't go over some of the Agentek stuff systems that sometimes I do on these calls. Um, again, watch some of those videos on LinkedIn. LinkedIn. You'll get a lot of cool stuff from some of what I share on there. Um, but, but it is—
the answer is yes, but it's not super easy yet. Okay. Yeah. And again, by agentic, you're—
that, that is connecting different systems. What's agentic again? Yeah, an agentic system is going to be like Replit's an agentic system, right? You guys have used already one, right? But here's another one called Manus, right? Manus can— I can ask it to do any— Replit builds software. Manus can do anything, right? So for example, I could have it, um, I, I had it do this. I had it extract from a website. So this is a huge chat, wait for it. Okay, so right here is the Utah Homeowners Association registry. It's a public site, public information, right? I can click A right here and then search and all of these different, you know, places pop up here. So I can look at this Farmington Station Owners and it's got, you know, the email and the telephone and address and name and all these different people, primary contact, all this stuff. Great. Lots of, lots of information here. But somebody reached out to me and said, hey, I want to be able to scrape this and put it into my own Excel document so that I can, you know, go through it and, you know, do a bunch of sales activities around it. I was like, okay. I asked Manus to scrape it and put it onto a spreadsheet for me. I went through, it did this whole first step and then the context actually reached its limit and so you just click inherit and continue. It can actually just say, okay, I'll take everything I've done and I'll keep going. I did that, I did that again and then after about 2 hours it was done and it had created— oops, here we go. Let's click this and let it load. It's kind of a bigger file. It had created this Excel document for me. HOA name, number, address, city, state, zip, primary contact, contact email, phone, email. Pretty dang cool. Yeah, yeah. So AgentX for marketing, reaching out, you can do it. It can do so many things, right? There's a lot of other systems. RunnerH, um, there's there's Operator now that's inside of ChatGPT. There's— this is the next kind of phase of AI that's coming, is right now a lot of what we did was like human plus AI, right? The next phase is AI just does stuff for you. You say, I need an outcome, and it just gets to work, and it opens systems on its— it runs its own computers, opens browsers. You can hand it your usernames and passwords. It can log into stuff and do things and log stuff, and they can do crazy stuff for you.
Wow.
Really fun.
Wow. Kind of the next step. So like in my business right now, like I'm kind of getting owned by a lot of processes and like an idea that I had from this course was to just like have a conversation with AI about like, this is what I'm doing. Like what pieces can slash should I automate?
Like, is that— Do that. Like a use case for— That is the perfect use case for what you were trying to do. Or for what AI can do today, you want to go in and find the mundane tasks. You're just like, I hate this, right? Like, for example, maybe you're moving data from one sheet to another system, right? You're inputting data, automate that, right? And if you don't know how to do it, that's okay. AI will guide you through it and you'll be able to build it out. You can build a little Replit thing, you can build an N8n thing, or you could just say, hey, you know, Manas, do this for for me, right? And it'll just do it, right? You can even create scheduled tasks for Manus, right? So it could go every day and scrape the internet, find some things and send a report and put together a nice system. I had Manus just a couple days ago. Again, we're working with, you know, Dean and Tony on some fun stuff. And he did a Facebook, he just did a Facebook Live for 10 minutes. And he said, hey, Jon, would you go through and look at the comments and see what people are interested about? What are they asking? What do they want to learn? What are they afraid of with AI? This stuff, right? I said, great. And so I went in and I just kind of copied a bunch of, of, of, um, comments from that, right, from the live video. And I actually did that manually. I just kind of said— I just like scrolled down and copied a bunch, put them into a Word document, and I had like an 80-page document. And then I said, put together a website, um, that, that kind of summarizes the data. So it did. Sentiment, 335 comments, sentiment, concerns raised. And I just, you know, here's the learning interest. This is what people want to learn. Prompting, personal development, content creation. Here's their current AI usage levels, right? Most people are moderate, right? There's a few little none, but most people are kind of dabbling in it. Here's fears and excitement, right? Things that they're worried about and they're excited about. Here's a few key quotes from different people. And I sent this link over to Dean. He's like, what the heck? How did you do this? Like, anyway, so, and, and it, it took me, it took me as long as it took me to write the prompt, right? Minutes, minutes to do that type of a thing. So, um, man, this is pretty cool. Agentic systems are really cool. Can take that busy work of going and having to like filter through and do all these searches and try to analyze data, man, it can just do it for you. And Manus created that site for you, that you didn't dump that into Replit to say automatically? Yeah, now I certainly could take— I could, I could, I could tell Manus, say, hey, I just need you to give me all the data, and then you could take all the data, drop it into Replit, and have it build it out. But honestly, I probably wouldn't do that.
Manus is capable of building sites and presentations and things like that. Wow. Well, how is it doing with like, like if you're wanting to create like a dashboard for a client or something, but like there isn't— I guess, I don't know, I—
the pulling the data, I think, is sometimes challenging. Yeah, you have to have API access. You have to have some, some way. So like, maybe, maybe all of your stuff is in HubSpot or Salesforce, right? Those systems have APIs, right? And so if it has an API, you can, you can just pull, pull from that data. You also might be able to automate exports or something like that and have it automatically email somewhere every day. There's a few ways you can kind of get around it. You also might be able to use Manus with a scheduled task to have it go in, because what Manus can actually do is like open up a site, click on things, fill in forms, quick stuff. Like, you can do stuff like a human, right? And so, um, yeah, it's just a matter of, you know, to get back to your, your question of like, well, how do I do this in my business? Well, yeah, just identify those tasks and talk to AI and say, hey, is there a good way that I I spend 9 hours a week doing this thing. Can I give this to an agentic system? Can we build a system?
Can I spend 9 hours this week, build that system, and never have to do it again? Yeah, no, that's awesome. Yeah.
Are there open APIs for like social media so you can pull like comments and reactions on posts and things like that? There are some, right? Some of them require some approvals and things like that, but, um, there's a lot of ways around it. And, and sometimes you can just publish publicly, you know, you can give it a, um, you know, a username and password and have it get in and figure things out. There's a lot of ways that it can get around APIs or API restrictions. Um, but, but yeah, there's thousands and thousands and thousands of APIs out there. Tens, maybe, maybe millions. It's probably actually millions of APIs out there. Every system you can imagine has an API. Some of them close it off, you know, for, for obvious reasons sometimes and sometimes they're just like, yeah, we don't do it. Like ESPN, ESPN doesn't provide an API, but smart developers figured out, oh hey, there's these endpoints here and if I ask it this question, it sends me an answer and someone figured it out, put it all together. Super cool. Google is still your friend. Reddit is your friend. I have learned so much on Reddit when I'm trying to figure something out like that. I'll just ask the question. Someone has asked it on Reddit.
Can I use Replit to do this? Reddit will tell me, you know, anyway. That's awesome. Well, thank you so much. Honestly, I think the biggest thing this course has done for me is like opened up my mind to like what's possible. And even like, I think this might be simple, but like the most valuable thing to me is like what different platforms are good for, how you're like, use ChatGPT as your project manager and Groq is amazing because it's pulling from all this user data. So use that as your market research. Like, even like that 30 seconds, I was just like, holy cow. Like, that alone has just been so illuminating for me because like, I see the potential of AI, but like, you know, and I've joined tons of little courses like that are just dumb, you know, that are basically just like how to write a prompt, but they're not like about like actually contextual. Yeah. And like, and contextually, like, like why does this one exist versus this one and how do you use them together? And it was so— it was—
it really was brilliant. Like, thank you so much. It was so good. Really, really appreciate that, Lisanne. Thank you.
Please write that in review, which I will also ask for in the email that I send out. So absolutely.
Well, I said it here because you have it recorded. So yeah, and there it is. I love it. Michael, did you have a question? You'd stuck around. I didn't know. I was just soaking everything in. Sorry, guys. I'm kind of just— I love it. He was dropping on you. But no, yeah, this is awesome, dude. Thank you so much. Yeah, loved all of it. Been great to have you. Uh, keep learning, stay connected, keep asking questions, dive in. Like, the, the real learning, guys, happens from doing, right? Like, people ask me all the time, like, I'm a, I'm a, again, a class 5 kayaker. I've taught kayaking my whole life. I've taught hundreds of people how to do that. And they're like, man, how do you just, how do you just like go across this huge rapid and it just looks so easy? And there's only one answer: time in boat. That's it. You just gotta do it. You just gotta get in the boat and do it, and you get to feel the water, and you're gonna mess up, and then you're gonna fix it, and you're gonna be like, okay, now I can do it. And eventually you can do it right. And so that's, that's how this is, right? Just like spend 100 hours in Replit, you'll be a Replit expert. Easy. You just will, because it will. And the cool thing that I love about it is that it teaches you as you go. It talks to you. It tells you what it's doing. Ask it, why did that work? Why did that not work? Here's what's happening here. Oh hey, now I learned something, right? Anyway, just be active learners, guys. Thanks so much, guys.
Lisanne, I think we're talking tomorrow, or is it tomorrow? Yeah, I'll have you on the podcast, so I'm excited.
This is a good preview for sure. I love it. I love it. Okay. All right, I'm going to end this call and we'll send the follow-up stuff soon. Thanks, guys. Awesome. Thank you. Thank you. See you. See you. Bye.










