Watch Me Start an Online Business With AI in 85 Minutes

I'm going to vibe code an app and try to find paying customers live without leveraging my audience. So anyone in the world that has access to a $20 a month vibe coding app should and could be able to do exactly what I'm about to do and show you no holds barred. So if you don't know what vibe coding means, it means using natural language like I'm using as I'm speaking right now to code an app without any coding knowledge. I don't have a computer science degree. I don't know what HTML stands for or means. I can't make a website if it's not using something like Wix or Squarespace. I have no coding experience, but you don't need it anymore, right? You can do whatever you want. So thank you, AI. Here's my idea for the app. I have a leadership position in my church. I'm a member of the Church of Jesus Christ of Latter-day Saints. And every now and then we do fun activities with the youth. We have like 60 or 70 youth in our, in our congregation, also called a ward. And when we do an activity, we need everyone to sign liability waivers. We are not liable if in case something bad happens. So this is a pain because we're sending out PDFs, they have to fill it out, sometimes they have to print it, sign it, bring it to the activity. It gets lost, you know, on the floor of your truck. And if a kid doesn't bring it to an activity, we usually don't turn them away because we don't want to. So either we take a risk or we put a bunch of people through a bunch of friction, and it's just annoying. So do solutions for this exist yet? Kind of. There are waiver tools, liability waiver tools, and DocuSign and e-signature tools, but they're all cumbersome and none of them are niched down enough for this use case. So I'm going to build a solution to a problem that I actually have today, and I will be my own customer. But that's not good enough for this video. I want to find a paying customer by the end of this video, if at all possible, not from one of the people watching this, although that'd be great. I would love your money. But I specifically don't want to use or leverage my audience to find a paying customer because I want this video to be as realistic and approachable as possible to as many people as possible. So I'm gonna use some other tactics that I know to try to find customers, and I'm gonna share my screen. You're gonna watch my prompts, you're gonna watch everything I do, Facebook ads, Reddit, whatever it takes. I'm just gonna build this live and hopefully it's entertaining and engaging. So, so I told you the idea, the problem that I'm solving. Okay, I need to think of some names. So I was thinking of names with the word waiver in it. Okay, I like domain names that are .coms preferably, not .ai or .net. .online is cool because it's kind of a newer TLD ending, but I prefer .coms. So waiver this, waiver that. I don't want more than like 3 or 4 syllables, simple, whatever. And so I'm going to share my screen, sign into my Namecheap account real quickly because that is what I use to buy domain names. And I'm going to buy the domain name for this business right here in front of you. I'm on Namecheap. This is what I'm buying. onlywavers.com. Done. onlywavers.com. It's a $12 domain name, 4 syllables, easy to spell, easy to remember. Does it pigeonhole me into only one thing? Yeah, kind of, but it is what it is. So that's what we're going to go with. onlywavers.com. There we are. I'm going to go here and this is my Replit account. So Replit is one of several vibe coding tools, an AI tool built on top of an LLM, large language model like OpenAI's ChatGPT or Groq or Gemini or Claude. Those are the big 4, and it enables you to more easily build websites and apps and whatever. So they just came out with a new product called Replit Agent 4, and it's really, really good, and I'm gonna use that to build this app. But before I do that, I want to get a killer prompt to give to Replit so I can save myself some headache down the line. So I've got my Claude account pulled up right here. The only thing in this chat is I want a vibe-coded tool, help me think through it. It didn't even give me a good name. And then I said, what about onlywavers.com? And it's like, that's too similar to OnlyFans. And I'm like, you're overthinking it. That's a non-issue. So I bought it. I'm using Fable 5, which is their strongest model. It's also their most expensive model, but I'm a token maxer at my core, AKA I don't care about like being efficient with token usage. This is the most innovative product the world has ever seen, as in artificial intelligence, and I wanna get the most out of it, even if it means like the equivalent of driving a Ferrari to the grocery store. Okay, that's what I'm gonna do here. So I'm gonna use Fable 5. Okay, here's my prompt to create a prompt. Now, don't overthink your prompt. A lot of people are like, you gotta garbage in, garbage out. You gotta have the best prompt ever. Don't overthink it. Any prompt is better than no prompt. And if coming up with the perfect prompt is enough friction to prevent you from actually doing the thing, then just tell it whatever. Okay. I don't actually usually do this for Replit. I usually just tell Replit something like this, but this is just one step further to ensure that I have less back and forth with Replit because the last mile in vibe coding is the hardest part. Picture building a house. We had this house built 10 years ago and we got all the way down to the drywall and the paint and everything. And then that the last little bit of making all the changes and stuff took just about as long as everything up until that point. Okay. The last mile is a real thing. And in logistics, delivery, electricity, whatever it is, the literal last mile is the hardest part, the hardest part to figure out, right? That's why Amazon owns, you know, who knows how many thousands of delivery vehicles. It's hard. It's no different with VOD coding. It's hard to find that perfectly finished product. And so a good prompt will make it less hard, hopefully. So here's what I'm thinking. It needs to be a SaaS from day one. So I'm telling Replit, I'm not just building this for myself, although I am. I'm building this to resell. Someone tells the app the name and date of an event in mind. Okay. That person has the option if they want to upload a PDF, someone, of the actual waiver for the app to emulate. If certain fields have to be there as required by the organization the person represents, a link will generate that can be sent to any number of people. When people get the link, they will add the name of the person or people that need to be on the waiver, maybe an adult or maybe an adult signing for kids. And all the link recipient has to do is add a name and tap to sign. Dead simple. And then the link sender will have Signed digital copies from each person in their admin login. The recipient doesn't need a login, but the person in charge of the event does. If the person sending the link doesn't have a waiver PDF to upload, then the app will provide a generic but protective one. It will use the Stripe API for payment and only costs $13 per year to use. One plan, no free plan. Keep it simple. The confirmation emails will have links back to the product. For viral referral marketing, use SendGrid for the emails. Let people get all the way up to generating the link before they have to pay, but also don't hide the pricing page or anything. So I don't want it to be a bait and switch. I don't want people to think it's free and then have to pay to send it. But I also want to remove all the friction of getting them to make the link. And then, you know, it's the sunk cost fallacy. They've invested 5 minutes into making the link, typing in the name of the event, really more like 2 minutes. And so once I've done that, like, oh, it's $13 per year. Oh, I might make it like $20 lifetime as well, or $15 lifetime. I don't want to waste The time and the effort that I just put into this, even if it's only 2 minutes. So I'm gonna go ahead and pay, but I do like that idea to make it $17 for a lifetime. Okay. Another important thing. I'm not just telling it it's for Replit, but it's for Replit Agent 4. Okay. Because Claude knows how these tools talk to each other and Replit Agent 4 is built on top of Claude. So we're kind of using the same tool here. I could use Claude Code to build this, but as soon as you start talking about like a terminal and all that stuff, like you've lost me. Claude Code is not too technical, but for me, it's technical enough for me to not wanna use it. Okay. So tell your prompt the name of the app that you're using to vibe code. There's a pro tip right there. So let's go. Let's do this. I should have told it definitively that this is called Only Waivers, but I didn't. All right. Let's go to the output here. Replit agent prompt, waiver signing SaaS. Okay. Add in that it's called Only Waivers. Okay. Let's see what it says before the prompt. $13 a year versus $17 lifetime means everyone picks lifetime. You're selling it. Whatever. I don't care, dude. I know what I'm doing. Okay. I mean, in a sense I do, in a sense I don't, but that's fine. My bet is that people need this for one-time use. We have subscription fatigue, so I'd rather get $17 than $0 from them. And whatever, if a month from now I decide to do away with the $17 lifetime, I'll do away with it. And people who already got it, they'll be grandfathered in. I'm not gonna take it away from 'em, but they're gonna be rewarded for trusting me early. So tap to sign is worthless without an audit trail IP address. Okay. This is a whole product, so it's in there. Perfect. See, if I would've prompted Replit, Replit may not have added the actual legal stuff that makes this legally binding, but Claude knew to. Miner's language, that's good. PDF emulation, hardest to build. Okay. Signed PDF generation. Link expiration, so dead links don't pile up. When it finishes, test the definition of done checklist. Use Stripe test mode until the signer flow is bulletproof. Okay. Added a branding section. Okay. So we're going to pull this up in Google Drive. Here it is. I'm just going to Ctrl+All, Ctrl+Copy. Then let's go back over to Replit. So we're going to go to Replit. I'm going to paste in the prompt right there. I'm going to zoom in so you guys can see better, and then we're going to go CloudFable 5 default. Perfect. And I think it's using Agent 4 by default, but just to be sure. Okay, here we go. Use Agent 4 for this. Go. Here we go. So this is building my app right now, writing the API contract critical path to getting the frontend build started. While this is happening, I'm not going to rest on my laurels here, pun intended. If you know, you know, I'm going to start seeing if I can make ads for this before it's even done, because what do I always say? Sell it before you have it. I'm going to go back to Claude, a different Claude window, because just yesterday I installed the Facebook Ads MCP. Let me explain MCPs, Model Context Protocol. I don't know what that means at all, but I can explain what an MCP does. Okay. So imagine all the magic of ChatGPT or Claude inside any tool you use on a regular basis. For instance, I use Mercury for my banking and I can go into ChatGPT or Claude and connect Mercury, just link it together. That's it. That's an MCP, but really I'm just tying them together. And then I go, hey Claude, where is my spending out of control? And it's gonna go look in my Mercury account because I gave it that permission and I can give it read-only permissions, write permissions if I want it to start sending wires for me. I don't. And Claude would say, Your Chipotle habit's out of control. You're spending $16 a day at Chipotle. You need to get that fixed or whatever. Or your Anthropic addiction is out of control. You're giving me way too much money. And so MCPs are incredible. I got a quote to make an email automation for my Beehive newsletter. It was a $10,000 quote. It's a pretty complex automation. It's something that I could do myself, but I'm not great at it. And it would take a few hours and I didn't want to. I got a $10,000 quote and then I'm like, hmm, I wonder if the MCP could do it. So I connected Beehive. to Claude, and I said, make this email automation based on first principles of nurture sequences. And I just explained kind of what I wanted it to look like, and it was done, right? It's like vibe coding and automation in a completely different product, but within Claude or within ChatGPT. It is absolutely magical. So Facebook Ads has an official MCP, which is a game changer because Facebook Ads are kind of hard to set up if you've never done it before. There's like, what's a campaign, an ad set, an ad? What's the difference between that? Do I want Facebook to choose my audience? Do I wanna choose it? Do I want on the right side of the screen Facebook Reels? I didn't even know that existed. Instagram Reels? You don't need to know any of that stuff because the best expert in the world at Facebook ads, AKA AI, the most expensive person that you could ever hire is now $20 a month, right? You don't need to hire an agency. So I am gonna say, and you don't usually need to tell it to use X, Y, or Z MCP, but I do just to make sure that it does it. So if you look at connectors, these are the MCPs that I've connected to Claude. Beehive, COROS, which is my watch, it's like a Garmin. Fathom, which is an AI note-taker. Facebook Ads. Function Health, where my blood draws go. GoHighLevel. Gmail. Google Calendar. Google Drive. Gusto for payroll. Higgsfield for AI content. Kubera for net worth tracker. Mercury bank account. Plod for this pendant that I'm wearing under my shirt that records everything. Robinhood. We all know what Robinhood is. Stripe. Typeform. Zapier. Yada, yada, yada. Okay, so I'm gonna say, you know what? What am I doing? What am I doing? It would be better To go into here and do it in the same chat window. Why wouldn't I? Okay, let's say 25 to 60. Okay, let's see. So now I'm going to go to Google. I'm going to type in LDS liability waiver PDF, and it has the permission and medical release form 2017 in the title. So this is what I'm asking parents to fill out. So I'm going to download that, and I'm going to go back to Claude. Okay. So Facebook could find the people for this app, right? In fact, Facebook will do a better job at finding paying customers for this app than I will. But Facebook doesn't know that I want to test this audience first. It's very important to me. It doesn't— like, Facebook has AI. They know way more about this stuff than I'll ever know, but it's not in my head. It doesn't know what I want, right? And it doesn't know that, like, my knowledge base is based around this problem. I'm trying to skip the line a little bit. I can probably convert Other types of people like, you know, the owner of gymnastics gym or the trampoline park, I can probably convert them cheaper or easier. There might be more of those type of people, but it's not my problem. So I want to start with the audience that's closest in line with the problem that I'm experiencing and trying to solve with this product. So I've never used the Facebook ads MCP before. This is my first time. It might blow up in my face, but we'll see. Now, while that is generating, let's go back to Replit. Okay. Writing the API contract. Okay, I don't know what any of this stuff means, but that's okay. We're just gonna let it keep going. That's fine. Let's go back to Claude. Okay, copyrighted by intellectual reserve. You can't ship a clone of it in the product. I'm not trying to do that. Use it as a blueprint for what an LDS-friendly template needs to collect. Oh, interesting. Meta killed religion-based detailed targeting. There's no BYU LDS Living. Yep, let's do that. Cool. Okay, campaign will be created paused so nothing spends until you review. And I'm gonna go ahead and open up Meta Ads Library in another window just so I can check in on things myself. Meta Ads Manager, I meant. I'm going to share a different screen just because I don't, I don't know if it's going to show something that's like personally identifiable. So I'm just going to do that. All right. While this is generating, I'm looking at Claude right now and it might be like displaying API keys. So I don't want to show that, but it's creating campaign. It's creating an ad set right now. This is pretty dang cool. Let's look at Replit. Oh, Replit's ready. Okay. So it's been like, I don't know, 10 minutes since we started Replit. Sign waivers, zero friction, mobile-first waiver signing for events, gyms, and tours. No apps, no accounts, just a link and a signature. Start for free. View pricing. Everything is free until you generate a share link. Okay. You didn't need to say that. You don't need to advertise that. Create an event, share the link, collect signatures. I like that. Designed for the real world, insanely fast, legally binding, yada, yada, yada. We're going to go ahead and publish it. just because we can. I'm gonna go ahead and connect my domain name. So I just bought this domain name on this livestream. You guys saw me, it's like one of the first things I did. Click this, only waivers. I don't really like using Entry, and I'm gonna change my screen again because it might show like my username and stuff. But just know that this is like Plaid. It's like an easy integrator for domain names. So it wants my Namecheap account. I'm gonna go to manual setup because that is like, That's just what I'm used to. So I'm going to go to Namecheap here. I am not sharing all of my screens right now because there is sensitive stuff going on, but I'm just connecting my domain name right now. A record is going to be that text at that. This must be the most boring thing ever, but I'm almost done. Okay. I'm sharing my screen back to Replit. You're all set. Only Wavers will be successfully connected. Done. Okay. Only waivers now. Like, I really don't like this. We've got to make this public. Right now it's invite only. We gotta make it public. Okay. Thankfully it's anonymizing my keys right here. We're gonna hit publish because the default for Replit, even when you click publish, is for it to be a private URL. So we're fixing that right here. It says you gotta do it like twice. Kind of weird, kind of annoying, but we're working with what we got. So. Let's go back to Claude and see where we're at. And you know what, if you guys, I'm just remembering right here on my shared screen that I do have an affiliate link. So if you want to use Replit, I'm putting it in the chat. I don't know if it appears on all the platforms, but you're welcome to use that link and that would be great. I get $20. All right, let's go back here. Republish. It's now public. Let's see this. All right. The site isn't working for me yet. It takes a little time, but I'm going to go back to publishing. SEO rating healthy. It shows that it's public. Connect Stripe. I'm going to do that. And Replit already has my Stripe API. I'm actually quite paranoid that it's going to display that. So I'm going to share a different screen, share my Claude screen while I check that out. Replit is suggesting tasks, send waiver copies and organizer alerts by email, confirm the whole signing flow works on a real phone before launch, make sure paid customers can generate links after Stripe is connected. Those are the 3 action items it's getting me. giving me right now. And it's quite crazy to me that we're already to the point where we're like testing this. We're like live testing this. And I've been building this for about a good 20 minutes. I'm also going to need to add Google Analytics so I can see how many clicks are coming through here. Stripe is connected. That's it. It didn't display anything sensitive. I didn't need to change my screen, but it already had my API key. If you didn't already have that, you would literally just Google Stripe API key, create a free Stripe account, go into your settings, API key, generate, copy, paste. That's it. I don't even know what API stands for. Actually, it's like Application Program Interface or something, but I know that it's how technology or software tools talk to each other, right? And it's a really long key. Okay, Stripe is connected. I can now accept payments. The website is not live yet. I guess like the domain name, it's not live on the domain name yet, but on the Replit site, it is live. Now I'm going to send waiver copies and organizer alerts by email. I use SendGrid for this stuff. So You've got companies like Mailchimp or Beehive, which are like public-facing newsletter companies. And then you've got companies like Mailgun and SendGrid, which is like the plumbing or the infrastructure for emails, really transactional emails. So like order confirmation, your Amazon package was delivered, that's a transactional email. And companies like SendGrid power that. And I already have a SendGrid account and my API key is already connected to Replit. So it's just what I know. It's kind of a commodity. I don't know if it's good or bad, it just works. So, so it's waiting for input, send waiver copies and organizer alerts by email. It just needs to request the SendGrid credentials from me. So I'm going to share my Replit screen so you can see how simple this is. Boom. There's my API key. It's anonymized. It's already in Replit. I'm just going to click next. And then what email do I want this to come from? I want it to come from heychris@cofounders. I don't even need to tell you because it's anonymized, but whatever. H-E-Y. Okay, this is working really fast. Now text. If we want like text reminders, we would want a Twilio API key. Twilio is basically the SendGrid for phone stuff. I think I already have that connected, but I don't want to mess with that. That's like too much friction. We don't need email and text notifications. That would be good to have, but not at launch day. It's just too much friction, right? Remove friction, remove friction, remove friction. Put it on a shirt. Tattoo it on your forehead, right? In any business, you want to remove friction to getting the money and add friction to losing the money. Okay? That's why, and you know, you want to do this within some bounds so you're not a scummy marketer. That's why when you go to cancel a SaaS product, it's like, are you sure you want to cancel? Are you sure? Do you want 50% off? Like they're adding friction to losing money, right? And it's smart, but it's annoying and you gotta like, just don't be too annoying with it. Okay? Everything is wired up correctly. The signing route sends emails in the background. Okay, let's go back to Claude, 'cause remember Claude was working on Facebook ads this entire time, which is just magical. It's just magical. Here we go. It's done. It's already done. Like, like, guys, one prompt. Look at this. Okay. It's just like, it didn't need any input from me here. It was like, hey, FYI, don't wanna get sued. FYI, you can't target people based on religion as of 4 years ago. What ad account? I told it and it created it. It's down $20 a day. CBO is campaign budget optimization. It just means like from what level of the ads are you going to be spending money? And this is like the highest level from the campaign level. Traffic objective, hard capping targeted advantage plus audience expansion turned off. So your 25 to 60 male US settings are actual limits since the whole point is a narrow creative match audience. So basically Claude told Facebook ads, hey, I know you think your AI is really smart and it is. I'll admit it is, but we don't want to use it yet. Okay. Ad set optimizes for landing page views, so you pay for people who actually load the site. I stopped before creating the ad itself on purpose. The ad needs a destination URL that resolves and a Facebook page to run under. OnlyWavers.com isn't live yet, and I wasn't going to guess whether this runs under one of your existing pages or new Only Wavers page. When the site is up, tell me the page and I'll finish it. Ad copy. The creative is a targeting meta camp. You plan the youth conference, you book the camp, and now you're chasing 40 parents for permission slips the night before the trek. This is so, this is so good. This is so accurate. Only Waivers fixes this. Create your event, text one link to the ward group chat, and every parent signs from their phone in 30 seconds. You see every signature in one place. $13 a year. I'll tell you what, if you're selling to an audience of my religion, Church of Jesus Christ of Latter-day Saints, you kind of got to be cheap. We're cheap. We're just cheap. Okay. Hardest place in the world to start a business along I-15 in northern, central, and southern Utah and all of Utah. So $13 a year. That's what we're starting at. Words like ward, stake, track, mutual youth conference mean nothing to the general population, everything to your buyer. Wasted impressions on non-LDS men will click nothing, and Meta's delivery system will learn from who clicks and narrow itself. The form is barely a waiver. It's mostly a data collection form, allergies, medications, yada, yada, yada. Your V1 collects a name and a signature. Silas the Chase. Okay. If LDS organizers are the wedge, the Replit prompt needs one addition. Organizers can toggle on. Okay. And then I'm going to screenshot the homepage of Only Waivers, drag it in here. Okay. Make a new prompt for me to add to all this to Replit. Remember, I'm talking like an idiot because it doesn't matter. AI knows what it means. Make LDS-specific language on the homepage, add the medical stuff and make it clean and frictionless. And LOL, don't let Replit tell people it's free, but they have to get to the end first. The CTA should just be make and send a waiver in 4 clicks or whatever is true. Maybe even add a hero image of a youth conference or something to drive home the point. Remember, and this is going to be a thing, this is a real thing. Okay. I'm not just doing this for the purpose of this video. This app doesn't have to stay this only. Like I would like to sell to more people than, you know, 3% of the US population. We can change the app, the website, the branding with, with one or two prompts. It's not that hard. So this isn't forever. This isn't permanent. Make and send a waiver in 4 clicks, yada, yada, yada. I've added an instruction forcing the agent to count the real click path and adjust the number to whatever is actually true. Cool. I wonder if it did that because I said, or whatever is true, because I don't want to lie, but I also don't know how many clicks it's going to be. A CTA that lies by one click is worse than a boring one. It's making an illustration instead. Fine. All right, there's the reprompt. We're gonna go back to Replit and apply changes, SendGrid secrets, restart API, yada, yada, yada. Now Replit doesn't really like it. Really none of the vibe coding apps like it when you're like telling it to do multiple things at once. But if you see these boxes up here on the left, there was just 2 and it disappeared. That's Replit doing things concurrently, same time, right? SendGrid API over here, Stripe API over here, user interface over here, user experience over here. That saves time. Okay. Now I'm going to paste in this prompt from Claude, and then I'm going to open publishing. And it's on autoscale. That means if I go viral, then everything will be fine. It won't— the app won't shut down. Production database connected. You don't have to do anything with that. SEO rating healthy. Don't care about that right now. Adjust settings. Public. Back. We're going to hit republish. Solid V2 spec. Okay. Now I'm going to go back into the Meta Ads library. All right, here's Facebook Ads. Manager, only waivers, LDS event organizers, US. This was made in Claude. This is— I've never, I've never touched this. Okay. And it's not live yet, but so go to campaigns. We're going to go edit. We click edit so we can see kind of what it did. So as promised, it's got a nice name in there. $20 a day is what it said. Campaign budget, CBO, remember auction. We're looking for traffic. We could have it optimized for conversions, but we don't have a pixel for that yet. A pixel just tells Meta that, you know, it's doing what it's supposed to be doing. And when you set up Facebook ads, guys, Meta is just going to be screaming at you, use our AI, use our AI, use our AI. And yes, that's a sales pitch, but also they want your ads to work because if they don't work, then you'll stop spending money with Meta. So their AI does work, but it's not like it's a blanket statement. They work all the time for everything and always just choose their AI and never use their brain. That is absolutely not the case. So 25 to 60, location United States, men. We could do men and women because there's plenty of female leaders that send out these as well. So I might change that, but let's see the detailed targeting. It didn't target people by like BYU or anything, but what it is doing is it's doing that in the copy of the ad itself, right? Speaking to things that only they would know. Let's go back into Replit. How this is going. Database validated, approve and publish. Okay, we're publishing. Here we go. It's still hosted on the Replit domain, so I'm gonna open domains, type in my domain name again, checking, next. Namecheap, manage, 54 minutes ago. I just love how everything's working all at the same time here. Go back here, overview, apply changes. Talk about removing friction. Like that's kind of been the theme of this video. Replit is removing friction by hiding all this stuff. 15 actions. Like, you can read it if you want, but guess what? Most people using Replit don't know what any of these things mean, myself included. They don't want to confuse people. They don't want to give more information than they need to, so they hide it. They remove the friction. And a year from now, Replit will probably look even more simplified than this. So we go to domains, open domains, still loading. Okay, now I need to make the creative, aka the, you know, what the ad actually looks like to the general public. So I'm going to do that. With this bad boy right here. Got this microphone. It's a DJI Mic 3, I believe. Bought it on Amazon. I'm gonna show you how easy it is to actually make Facebook ads. How about that? I've got a receiver here and a transmitter here, and I've got an iPhone 17 right here. And so I'm just gonna— I'm plugging that in. Okay, you can see that little icon, which means it should be working. I'm gonna talk into this. Hello, hello, hello, hello, hello. You see the mic's bouncing. So then I'm just going to open the camera app right here. That's it. I'm going to go to video, front-facing. Okay, there we go. Look at this freaking guy right here. If I wanted to get Claude to make a script for me and then put it on a teleprompter app, I use this app. It's literally called Teleprompter. Okay, it's $20 a year. I would just take the script, paste it into Teleprompter app, and then it would look like this. And I would just hit record and then read the script as I'm filming. But this is stupid simple. It's only going to be 15 seconds or so, and I probably won't even need to edit it because the goal will be to take this in one take. So if you are 20 to 50 years old and an active member of the LDS Church here in the United States, as I am, you've probably experienced this problem. You've got a ward activity, youth conference, stake conference, lake day, and you need to send liability waivers. But those things are a pain and they're cumbersome. You got to print them, sign them, give them to the bishopric member. He's going to lose it. You don't bring it. Your kid doesn't swim. Maybe he does, but then The church is liable. I fixed all that. OnlyWaivers.com, $13 a year, a few clicks, you've got a waiver. Send it out to everyone you know and it's fully compliant. Use it right here, OnlyWaivers.com. All right, so I stumbled over my words. It was 30 seconds exactly, but we're going to YOLO it, okay? Because done is better than perfect. The LDS Church here in the United States, as I am— There it is. Okay, not faking this. I'm just going to hit AirPlay or AirDrop. I'm going to AirDrop it to my computer. I'm not gonna edit it. I'm not gonna add captions. I'm not gonna add a thumbnail. I'm not gonna remove pauses. I'm just gonna AirDrop it and we're gonna YOLO it. So let me share my screen again. We're going back to the ads. So I'm just gonna save some time and if it doesn't work, it doesn't work. But, um, I'm just gonna upload it right here and maybe it works. I don't know what API permissions it has through the Claude MCP, but we're about to find out. So I would've liked for that video to be shorter. I have better luck with 15 to 20 second videos, but at least I'm speaking directly to the audience. So What that means is if I'm a 72-year-old woman who's not an active member of the LDS Church, I'm going to swipe away from that so hard, which is perfect for me, right? And then YouTube's going to learn, oh, don't show it to 72-year-old women that aren't members of the LDS Church because all the 33-year-old men in Provo, Utah who are members of the LDS Church are clicking the heck out of this. We're going to show it to a lot more people like that. So Facebook's going to figure out pretty quickly the video works for the ad, 31 seconds, perfect for Reels. No public URL. I can't push it to them directly. Open Ads Manager, go to media library, upload to your phone. Okay. I can't read the file specs, but I can't watch it. I'm going to push back a little bit and say, I'm going to say, read the transcript. I'll add it to Meta myself. Oh, look at this. Look at this frigging thing right here. You see my URL, onlywavers.com. There it is. We're live, baby. We're live. It's working. Okay. Uh, don't like that. View pricing. Well, it hasn't made the changes that I told it to make. Okay, so I googled Meta Media Library and then I just clicked create a folder and it's loading that right now. I'm going to close my Namecheap tab because, okay, because I already fixed the domain. We're going to go to Ads Manager, top left, Media Library. I'm going to poke around in this while I share my Replit screen. I'm all over the place. Prevent Stripe webhook. Sure. Catch broken Stripe payments before they reach customers. Sure. Look at all these blue things on the left. They're running concurrently. Let organizers view their plan and manage billing from the dashboard. Sure. It's doing things that it just thinks I need and I do, which is really nice. So I go back here, shows it's not live on the domain, but we're going to refresh and it should there. So there it is. Boom. And let's go ahead and click republish because why not? Okay. Let me go find the Meta media library. Media, Control+Find, media. There it is. Okay. Go back to Meta Media Library, and we're going to create a folder called Only Waivers. Go into the folder, add images and videos right there. Okay, now I can add that video, or Claude can add that video to any campaign within that account. Speaking of Claude, it's still working to get that transcript from my video. Remember, if Claude says it can't do something, it doesn't mean it can't do it. It just probably means it doesn't want to spend credits doing that thing because it doesn't think it's necessary. It wants you to find the easier way. Because Anthropic, aka the owner of Claude, wants to save money. So push back, push back. Like, what do you have to, what do you have to lose by pushing back? People pleaser, not against AI. So it's still trying to get the audio from the video because I want Claude to tell me if that's a good video or not. If it's not, I'm going to push it anyway because I don't want to take another one. If it is, great. So website's live and Meta media library video is uploaded. It's right there. Cool. So now once Claude is done trying to transcribe the video, it can go Add it to my campaign. Now Stripe is just, or Replit is just pushing all these changes that needs to be pushed. All 3 services running, yada, yada, yada. And you can see, go back to Replit, that I've got 3 things running at once. Stop dev and production from fighting over the same webhook. Don't know what that means. Catch broken Stripe payments. Cool. Let organizers view their plan and manage billing. It's billing all those at the same time. Okay. And The website is publishing at the same time. You guys watched me buy this domain name a good hour ago or so. It's already live. It's so interesting what Claude is doing. It's installing tools to transcribe that video. He's like, oh gosh, he's right. I can, I can transcribe it. I just didn't want to. We're going on like 5 minutes from this. Definitely would've been easier for me to just tell it what I said, but whatever. Let's just go look at it. Look at the preview. Okay. There it is. Okay. Every permission slip signed before the activity starts. Built for ward and stake activity leaders. Create your event. Text one link to the group chat and every parent signs from their phone in seconds. No app, no account. It's for parents. No chasing. Make it send away around 4 clicks. Set up your first event before you pay $13 a year when you're ready to send. Look at that. Send link. Cool. Watch the list fill up. Youth conference, track girls camp, campout, service projects designed for the real world. Touch signatures. View pricing, lifetime, yearly. I love that. That looks great. Subscribe yearly. What happens? Okay. Checkout failed. Checkout failed. That's one of the, what one of the commenters said. So what happens when an error happens, which will always happen? I just screenshot it and I'm gonna drag it over to Replit. Seems the checkout links aren't working. I need to be able to accept payments for this ASAP. Get on it, please. Always say please and thank you to your AI. Okay. Because once they're embodied, We don't want them to come kill us. That's just the fact of the matter. Okay. I should say we want them to be a little less likely that they'll come kill us because they probably will anyway. Okay. So Claude was not able to read it. I pulled the audio, but the transcription model has to download from a server, yada, yada, yada. All right. I'm just going to tell Claude what I said. Okay. I just told Claude what I said. Okay. Replit is working on 3 different things at once. And here's what I'm gonna do. This will be good. This is good learning. I'm going to go to analytics.google.com and I'm going to go to admin. I'm going to go up here. I'm going to go create account and I'm going to call it Only Waivers. Okay. Go to next. I'm going to go Central Time. You got to have Google Analytics. Okay. Chicago time. Cool. Industry, business, small. Next. Yes. Yes. Yes. Yes. Why not check the boxes? Yes. And then we're going to call it web. This is not an app. And then we're going to put in the domain onlywavers.com. And then stream name is the same. Enhanced measurement, absolutely. Create and continue. Now it's going to give us this. We're going to copy that. Okay. And then we're going to go back to Replit. Please add Google Analytics. Give it my tag. And it's got that queued up, but it's in the queue because it doesn't want to confuse itself. So checkout fix is live. Publishing will put it on your real site. Let's try it. Okay. I click pay and it's making me, it's making me sign in. I don't like that. Let's just go through the flow. Make and send a waiver in 4 clicks. Okay. I got to fix that. Okay. We're swapping from Stripe sandbox mode to Stripe real mode so we can actually accept payments. All right, so I'm going to take this prompt, which is, I don't know, probably good enough, but I'm going to give it to Claude. I'm going to tattle on it. Okay, well, Claude's already working on this, so scratch that. I'm going to let Claude stay in the mindset of making the ads, and I'm going to go back to Replit and just give it, give it this prompt. Okay, okay, finish Stripe Sandbox. Okay. We'll click republish. I click republish like it's going out of style. I don't know if Replit like charges you extra when you hit publish or not. Probably does. Whatever. I'm just excited to make all the updates go live. So this is a very key update here though. I don't want people to have to create an account to generate the link. That's, that's important. Now I'm very excited to see if the checkout link actually works. Because it should, and it's not in sandbox mode anymore. So theoretically I could get my first customer. All right, we've got our Google Analytics stuff. I asked Replit to do 2 things at once. Google Analytics is in, now it's on every page on the site. Pay first, account second. Let me check what our login system supports. Republish. And then let's go to the website and then let's just see if I can check out like as a, as a customer. So refresh, refresh. Make and send a waiver. Okay. That hasn't updated yet. I should probably do some FAQs, but I'm not gonna concern myself with that now. When I click this, it doesn't take me back to the homepage, so I gotta fix that. Sign in, get started. Okay, it is really like pounding that home. Okay, let's go see where Claude's at with its ads. Redundant video and copy is a real problem. Your video carries the pitch. The text carries the specifics the video skips. New copy. Okay, I kinda like that. I don't wanna take another video, but that is better. Oh, that's the primary text. That's not the script. So I guess it's just gonna switch the primary text the headline and the description in the ad to better match what I say in the video, which means I don't have to take another video. That's great. Okay, ad set, upload the video, tell me the Facebook page it runs under. Okay, video is now in Meta. Make the daily budget $200 instead of $20 because I want results as soon as possible. I can pause it or turn it off at any time. Push it live ASAP. Turn on closed captions. Okay, let's go check in on Replit, pushing some changes. There's an issue publishing your artifact. So now, thanks to the fact that this is an agent, an AI agent, AKA an AI that will go out and do stuff for you, Replit Agent 4 in this case, it found an issue and it's going to fix it for me with an agent. And I clicked that twice. So let's see what we've got going over here. Over here, it's fixing webhooks or something over here. It's generating some images based on my prompt. And then over here, this is like the main build. So this is already going to be fixed. I click republish, approve and publish. Okay. Claude is basically asking for permissions to do some heavy lifting in my Meta Ads Manager account. I said yes. Technical constraints, ads update entity. Okay. Conversation's getting a little long, so it's compacting it. Okay. The full guest flow works end to end. That's what Replit's telling me. Draft event, Stripe checkout webhook activates the link, stores the pending account, and the clerk invitation email went out. Now checking the user interface. It looks right. Verifying the account claim step that a user signing up with the payment email inherits the plan and event. Okay. Let's look at the preview. Okay. Look at that. It fixed it. And look, even that is fixed. I didn't even ask that. It fixed itself. I, when I click the logo, it goes back to the homepage. Make and send a waiver in 4 clicks. Trek. I went on Trek as a 16-year-old. It changed my life. It was awesome. When is it? Let's look at the calendar. Ooh. Okay. October 14th. Cool. Generate my link. Oh, you gotta pay. Let's do the yearly plan. Oh, my screen went away. I went through the signup flow. I went through the signup flow and it worked great until I clicked on the $13 link to pay. Stripe screen got stuck loading here. Okay. Now, okay. Okay. If I go to the public website, which anyone in the world can see now, and I go through the signup flow, there it is. Pick a date. Then I'm going to stop sharing my screen in case it shows my address, generate my link, pay $13 a year. I just want to see if the same glitch happened and it did not. And it's not showing my address either. So I'm going to— this is what happened. It worked fine in the real world. Now it says sandbox, but this is on the internet and it looks like it's going to let me pay, but it also says it's a sandbox. So I don't want my address to autofill. So I'm going to change back over to Replit and try to actually pay myself so I can see if this works. Even has Apple Pay. How cool is that? Okay. It's accepting my card, put in my zip. Okay. Subscribe. Your card was declined. Your request was in test mode, but used on a non-test card. Okay. That's fine. This is very fixable. It's still in test mode for some reason. So it kind of makes me want to cancel this request to fix the UI on the signup flow, because I think in the real world it works. So remove Stripe test mode so I can actually accept payments. Test/sandbox mode. Okay. It says my video is missing from the ad account. Okay. The API has no caption toggle, so I have to do that myself. A pixel with a purchase event on the Stripe success page is a 15-minute Replit prompt. Okay. I do need a pixel. Now a pixel is how my Facebook ad can optimize for conversions, right? Because the person that clicks on ads is completely different from the person that buys things from ads, or the person that adds stuff to their cart is not necessarily the person that buys from ads. So with the pixel, you basically get your website to talk back to Facebook and say, hey, that person that you sent, they actually purchased, right? Can you find more people just like that? And it will. So it's important. All right. It can't find it. Yada, yada, yada. Wrong library. Open the video in that folder. Copy the URL. I can do that. Facebook page is Koerner Office. Give me the prompt to give to Replit to get what I need for the pixel. And then we're going to go back to Meta Ads Manager. And then over here, we're going to go to here, media library, only waivers, video. Here we go. Where is it? Okay. So I got to change my screen again because I don't know what it's about to show. So I'm going to go back to Replit. Okay, this is taking a bit, taking a good minute. We'll hit publish. I'm gonna go back to Meta. I think I figured it out. I think I uploaded the video to the wrong place. It looks like Facebook glitched out on me because I added it to the right place, but it wasn't the right place. So I'm gonna add media, add video, do the same thing all over again. Image 3879. All right, I really think Meta just glitched out and it— I uploaded the video to the wrong ad account because you can see all these other videos here. Uploaded within, you know, this year. This is my active Meta Ads Manager account. So it's actually uploading to the right place. Finished. Cool. It's going to be there right at the top, I imagine. And then I'm going to go tell Claude, go find it now. Oh, these are images. There it is. Cool. Okay. So how cool is this? What Claude spit out. So I could just say Replit, you know, give me a pixel for Meta, but I had Claude, you know, make an overkill of a prompt for me. And there's a lot going on here. I don't know what it means, but it looks cool and it looks like a lot. So let's go give it to Replit. Copy, paste, boom. Okay. And it looks like all of my tasks are done. Let's go ahead and hit republish. Nothing's broken. Okay. One publish attempt caught my work mid-change. So me over here saying hit publish, hit publish, hit publish. Actually don't, don't do that because I broke my own thing. Uh, okay. So I need to go generate different API keys. I thought it was just gonna use the one I already have in Replit, but it's not. So I'm gonna open that in a different page cuz I definitely don't want y'all to see that. I'm gonna create secret key. I'm gonna use my Touch ID to validate my identity, and I'm gonna stop sharing my screen for a minute so I can put this API key in here. And I just told, uh, Replit, give me the field to enter my real Stripe API key. Okay, so to do that, you're going to go to Settings and then Secrets and then New Secret. I'm gonna go ahead and put in my Stripe API key. Let's go check on Claude. It's looking for my video. So right now Claude is asking for permissions to fetch videos in my Facebook Ads account or to create an ad. And I think it found it cuz now it's making an ad. This is so cool, guys. This is so freaking cool. Like just, I'm using like PhD-level experts that would cost thousands of dollars an hour, multiple at the same time, a coding expert, an ads expert, copywriting expert for like, well, $200 a month, right? $250 a month. Most of that is Claude, but all right, guys, it's live. It's freaking live. All 3 levels active. Campaign $200 a day. Ad set active. Ad active. Yourself a video under the Koerner Office page. Learn more. Okay. It's going into the Meta review queue first. So we've got like a live ad and the business isn't even live yet. That's how you should do it. Sell it before you have it. Guys, I've been saying this, come on. Wonky costs, yada, yada, yada. Captions. Okay. I can do that myself. Pixel. Ooh, I still need that. At $200 a day, you'll roughly get 100 to 200 landing page views a day. You need about 15 sales a day just to break even on ad spend, which almost no culture. That's okay. That's okay. Lose money on year 1 revenue. Year 1, dude, come on. Day 1, week 1. Your real bet is each buyer's ward has 20+ other units watching. That is my bet. Referral footer in every signed waiver email and lifetime value past year 1. If you're judging this on week 1 ROAS, you'll kill a campaign that might be working. Okay. We still have to give Replit agent or our pixel prompt. Okay. I'm going to swap back over to the Replit screen. That's cool. You toggle between design and build. I've never done this before. Okay. We're going to go back to build. Uh, okay. Finish Stripe sandbox setup. Open publishing. Are my Stripe keys good now? Probably not. Look at this. A publishable key, a secret key. I don't know what the difference is. I guess it needs both of them, but I only saw a place to put in one. So we're just, we're working through it. That's what you got to do. So right here, Stripe API key, but there's only one field. So don't want to click that. Confirm the live site is accepting real cards after republish. So I'm going to go to my site in a different window and I'm going to do this. I'm just filling it out, generate my link, $13 a year. I don't think it's going to work yet, but it's telling me to check it. So I'm going to, I'm going to check it. Card. Cool. Password. I know you can't see this, but that's okay. Subscribe. Okay. Still in sandbox mode. Still in sandbox mode. Tell me what Stripe key you still need from me based on what you see in my settings. I'm gonna have Replit get to work on the pixel stuff, and then I'm gonna ask Claude how to fix the Stripe API key stuff. Copy that. Okay, I'm gonna work on Stripe for now. Do this. Okay, we're gonna go back to Claude. Okay, so I asked Claude, having issues in Replit with the Stripe API key. It wants both secret and publishable, but I only see a field for one of those. Still in sandbox mode. What and where to add which key in Replit to make this finally work? The secret key is the only one that's truly required. The publishable key only matters if the code uses stripe.js on the front end. Okay, so I need the secret key. I've been using the publishable one. All right, so I'm in Stripe, create secret key, and I have it. Okay, so now I'm gonna go back into Replit. I'm gonna add it, but not, I'm not gonna show you. I'm not gonna show you for obvious reasons. So I'm gonna click edit and I need to know if I leave this checked or not. This is the last mile I was talking about, guys. Leave it on. So I'm saying the one I generated starts with SK Live. Put that here. Then I'm going to screenshot over this and use Replit to check. Yes. Paste it there. Okay. I got a Stripe secret key. Okay. So go back here. I'm going to go back to Replit, paste that in. Okay. So now I'm going to share my Replit screen. Endpoints behave correctly with no pixel configured. Site works identically. Nothing fires. And I'll ask you for the pixel ID. Okay. Check my Stripe key now. We good? So Replit still needs my pixel ID. So I'm going to get that from Claude. Now keep in mind, you can use one Facebook, like Meta Ads pixel across multiple different campaigns, right? You got this kind of app, a tree trimming business. Like it doesn't need to be one pixel for one website. Let's see what Claude says. Okay, Claude gave me my pixel ID. We're going to paste that in Meta pixel ID secret in Replit. I'm going to stop sharing my screen so you can't see that. So I put in the Stripe key, I put in the Meta pixel ID, and then I hit publish. So surely we're not done. There's going to be something else that breaks, but we should be pretty dang close. Meanwhile, Meta is approving my ad. It might even be approved by now. I don't even know. All right, we're back in Replit. We're publishing. Make ad tracking survive ad blockers by reporting purchases from the server. That's cool. Wouldn't have thought to ask you that. Okay. Look at this. Look at all these checkmarks. Pixel is set up and live in the preview. Base pixel code, view content, initiate checkout, purchase on the success page, event ID, green, green, green, green, green checkboxes. Republish done. After it's live, say the word and I'll confirm live card mode is on. I've also added an optional task, sending purchase events from the server so ad blockers can't hide your sales from Meta. Worth doing before you spend real ad budget. Genius. Genius. I don't want to click that while it's publishing because I don't want to screw up again. But yeah, let's wait for this to publish and then we're going to go to the website and see if we can like check out. Okay. Publishing, publishing. Okay. Done publishing your app. While we go test it, we're going to click start on this. All right. So we're published. So now I'm going to onlywavers.com. And let me just share this. I'm going to click on Make and Send a Waiver in 4 Clicks, type in anything, go to the calendar, type in any date, generate my link. And then I'm not going to click on this yet. I'm going to stop sharing my screen because I don't want it to autofill my address. So I'll just walk you through what I'm doing. Pay $13 a year, clicking that, loading, Stripe page is open. So far so good. Putting in my email, putting in my card information. Fingerprint ID, putting in my zip code. Don't want you to save my information for a faster checkout. Clicking subscribe. Gosh dang it. Still doesn't work. Why doesn't it work? All right. Back to Replit to give them a piece of my mind. Live card mode on, still not working. And I know the key is good. If I do all caps, it's going to know that I'm serious. All right. So I need to ask Claude what my meta CAPI access token is. Maybe it knows, maybe it doesn't. Okay, asking Claude that. All right, so I got to do a couple more steps to get the CAPI access token. Ignore that task for now. We're going to push that through. Okay, I found the problem. It's not your key. I checked the project settings. The keys you entered never got saved. That's not true. There's no live Stripe connection on file, which is why the site— that's not true. Open the publishing page, look for the Stripe payment section, and choose connect. Okay, got to go integrations, Look for Stripe. Click Enable. Oh my goodness! Cancel. Cancel task. Click Publish. Okay. This is the Publish pane. In that pane, find the Stripe/Payment section. Click the option to connect your LiveStripe account. Okay. Thank you for the URL. Stripe App Marketplace gave me this page. Install App. Okay. Live mode. That's my Stripe account. That's what we're talking about. Gosh dang it. Come on, man. I've never had this issue. I've got a literal headache. Okay. Okay. I should, I should be good. I better freaking be good, man. What a rigmarole. Okay. Done. Check. Oh, you better work. Don't. Ooh. Don't tell me it works again and then have it not actually work. Okay. I still have to freaking publish. Okay, I'm 91% sure that it works now. Okay, it's publishing, it's publishing, we're waiting. I'm gonna ask Claude, is my ad approved and live yet? Okay, now I guess I could check myself. Boom! Approved and live, active at every level, delivery active. That means review passed and Meta is serving it right now. Learning phase runs 3 to 7 days. It doesn't take that long, guys. It, you know, it It's always it depends. But now I'm going to go to the Meta Ad Library. Just Google it, Meta Ad Library. And I'm going to go see if the world can see my ad because sure, it's approved, but does it appear there yet? And maybe there's a delay. There shouldn't be. So this is one of my favorite websites in the world. You can see anyone's ad. We're going to go all ads. We're going to go the Koerner Office. Okay. There it is. And it shows nothing is active, but I bet you anything that is outdated. We're going to check by going to my actual Meta Business Suite, and we're going to go to Ads Manager campaigns. I'm going to share my screen. You guys are going to like this. You're going to like this. I hate having to switch my screen. I am such a complainer today. Look at this. Only waivers right there. Ad sets right there. Learning ads right there. There it is, guys. Instead of last 30 days, we're going to do today so we can see if it has any impressions yet, AKA has anyone seen it yet? August 18th. Loading, loading. Impressions. No impressions yet, but that will change within minutes. I can almost guarantee it. We're going to look. We're going to click edit. Koerner Office Instagram page. There it is. And let's do exactly what Claude suggested and let's go to display link. So that's where it's sending people. Yep, there it is. We could say, hey, I want to A/B test 5 different descriptions, headlines, and primary text. I think I'll do that actually. Learn more. Pixel languages, creative testing, 25 to 60, attribution model standard. $200 a day. Now I think I want that website events, right? Yeah. Why would that not be checked? Definitely want that. So here's the problem with making changes to your ad once it's live. Sometimes it has to get reapproved and it could take minutes to hours for that to happen. So you want to avoid that as much as possible. I'm going to go up here to my ad and we're going to go here, edit media, edit video. There we go. Okay, captions, boom, generate automatically. That's it. I could add a call to action sticker, review captions. There they are. Save the video. You could edit them because if it, you know what, actually Claude told me to make sure LDS right there, it says it right. Save the video. Okay, save. So we've made 2 changes. We've made sure that the pixel is tracking website events. We added captions. Now we hit publish. Okay, so we're going to go back to Claude and say, I just want to make sure that checking the box was the right thing. Okay, Events Manager, check your pixel. I'm going to say you go do that and we'll see if it can. Man, we're— yeah, we're getting hot and heavy in here. All right, let's go check on Replit. Replit was published, so let's go to the website. I'm going to stop sharing my screen because I'm going to go to the website and refresh it just in case. Make and send a waiver, yada, yada, yada. Generate my link. Opens $13 per year. Click on that. Opening the Stripe page successfully. It still shows it's in the AgentGo sandbox, so I don't think it's going to work. I'm going to drag that over to Replit. I just yelled at Replit because I'm an ungrateful little brat, and I said, still in sandbox. I published, then refreshed my site and no dice. Fix it ASAP. That's really annoying. But what's not annoying is that my ad is live. Okay. I'm very, very happy about that. And if we go back into the ads manager, it shows my new ad is in review. Oh, but I've got landing page views. Holy crap, guys. Oh my gosh. I spent $44 in the first like 10 minutes. Oh man. And people can't even buy. Yikes. I'm spending $1.37 per landing page view. Hey, that's okay. This is— hey, sell it before you have it. That's the trade-off. That's the trade-off you make. I wouldn't have it any other way. Hey, at least I'm failing live, right? So for now, I'm going to pause my ad. Okay. I did the hard part. I built it. I'm going to pause it. Okay. For now. So I don't burn money. This is Replit's fault. I checked again after you published. This is what Replit is saying. Site really is on the sandbox. Never got attached. In your Replit workspace, click on the Integrations tab. Gosh, dude, I was just trying to do this. View all integrations. Let me finish reading first. Stripe, click Manage. You'll see 2 sections, Stripe API keys. Expand the live. Okay, I'm going to duplicate this window so I could still do that. Integrations, view all integrations. Stripe, enable, configure, manage connection. Okay, test connection healthy. Live account is not connected. Okay, so I'm gonna change my screen. Okay, here's where I'm at. I clicked Manage Connection under Stripe. Claim Sandbox. It opened a new window into Stripe. Okay, the Koerner office. I'm claiming my sandbox. Okay, I'm clicking Continue. Okay, now I'm looking at a fake Stripe account. Okay, activate your live Stripe account. You can see this, right? Okay, got it done. Install Repli app on your live Stripe account. So it's showing me like the approval page that you guys saw earlier taking. It's just one click. Okay. Okay. This is what it showed me when I clicked on that last step. You got to click this. There's all my Stripe accounts. Here's the one I want on refreshes the page automatically already installed live mode continue. Oh, this better work. App installed successfully. That is the first time I've seen that. Now I got to publish publishing. Republish. All right, this should work. All right, we're published. I'm going to ask Replit at the same time that I go check myself. Okay, blah, blah, blah. Generate my link. $13 a year. Oh my gosh, it didn't freaking work for me. I'm so annoyed. All right, I'm just waiting for Replit to figure this out. I think I'm guessing Replit just needs to fix something on their side. This should be very simple. Like this is not rocket science here. Okay, here we go. Bug is fixed. I'm going to hit republish again. In the meantime, for redundancy, I'm going to go into my Stripe account, product catalog. We're going to go, here's my Stripe account. I'm zoomed way in. 2 new Stripe products created by the API by Replit today, updated today, $13 per year, $17. So those have been created. So let's go back here. Once this is published, we're going to go try to check out ourselves. All right. So now I'm going to go try to check out myself. Your store is fully open for business, live payments, live products, and ad tracking all running on onlywavers.com. When sales start coming in, you'll see them in your dashboard. We'll see about that. Okay. I'm going to test this for myself and I'm going to even share my screen as I do so. So I'm going to wait for this to finish publishing. Oh, I also need to add like a contact us form to the homepage because if people aren't able to check out, they got to reach me somehow, right? So, okay, it's published. Refresh, refresh, refresh. Make and send a waiver. Generate my link. $13 per year. It no longer says sandbox. Subscribe. Here we go. $13 per year. Come on. Are we in business? Come on, baby. Come on. Come on. Why is it taking so long? Oh, so it still needs to make my link. Of course, it's confirming my payment and it's supposedly making the whole thing we're actually doing here, a link for this form. We don't even know if the product works yet. So this shows like it's broken. It's stuck. But when I go in my Stripe account, look at transactions. $13. Chris K succeeded. I just got paid. How lame would it be if I counted that as my first customer? Don't worry, I'm not that lame, but it works. It works. My website works. Does the product work? I don't know yet. Believe it or not, that's actually less important because we can fix that. But if we can't accept money, we can't fix that. So I'm going to screenshot that. We're going to go back to Replit. Okay. I kind of asked it to do a few things at once, which I shouldn't do, but I'm impatient. So I told it to fix this. I told it to make sure that link generation actually works because it might, but we just didn't get that far. And to add a contact us form. I just told Claude that I've had 60+ landing page visits from the Facebook ad. Go see if the pixel is good and everything's working. I also paid the $13 myself and it worked. Do you see that conversion in the pixel? Okay, so it's working on everything. It's doing a bunch of stuff. Just waiting here. I might need to come back here and finish this up because my kids are coming home from school. And I tried to log off when that happens. You can see here, the cloud is just taking forever. Replit's taking forever. Last week I spent 3 hours building this, onlywaivers.com. I'm using replit.com to build it. I'm trying to solve a problem that I actually have for myself. And that is getting parents to sign liability waivers for youth. That's actually quite hard to do because it's a cumbersome form and there's like, you know, what allergies do they have? And you have to like handwrite it and it's just outdated. So I'm trying to fix that. So we can all be honest with each other here. I worked on the app a little bit, so I'm gonna get you caught up to speed on everything I've done. It's not done yet. I think the product is basically done. Surely there are some bugs that I'll find later, but I have not started Facebook ads. I haven't done any marketing yet except for one little thing that has zero views, so I won't even count that, which I'll talk about in a minute. So I'm redesigning the homepage a little bit to add more trust signals. I want people to know that a real human is behind this, that I'm behind this, and that I stand behind These $13 and $17 fees that I'm charging and that I have a 100% satisfaction guarantee, lifetime refund policy. I don't care. I'm just trying to get a customer. I'm trying to get a foot in the door. So I'm willing to do whatever it takes to make that happen. I'm going to do some new marketing methods. I'm going to post to Reddit, a couple targeted subreddits, and I'm going to post to some Facebook groups. And then I'm going to run my Facebook ads because those are already set up. Hard part's been done. I just need to hit Reactivate. Okay, so here's what I've done over the last hour. I just did one prompt and here is what the prompt was. I said, add more trust signals. If you remember, I took a, a selfie video for my Facebook ad. I uploaded that video to Google Drive. I gave the Google Drive a link to Replit and I said, embed this video on my homepage. That way people coming from Facebook ads can see the video that they've already seen. And you might think, well, that's redundant. Why would you do that? Because it's a trust signal. It signals to them that the video that they just clicked on is also on this website. It's not a bait and switch. I'm getting what I'm expecting. I also said, go to chrisjkerner.com, extract info from my bio there, because I didn't want to have to type out my bio, and put like a headshot and some bullet points in my bio on the homepage as well. So they can see that a real human is behind this. And then here's my Replit page. Beautiful. Start, start, start. And so if you can see these 4 things, we've got an SEO scan happening. That's the green one. And then these 3 blue ones are the tasks that I just had it start. So let's go back to my homepage and see if those updates are there. I told Replit to make it a little tongue-in-cheek, add some like niche references, some deep cuts. There it is. Yes, I built this. Watch it work. No guy in a headset, no stock footage of people shaking hands, just the actual product doing the actual thing. If you are 20 to 50 years old— Designed for the real world, looks clean, full transparency. This app is new, 100% lifetime money-back guarantee. Not, we've been around since 2019, I have 50,000 users. Actually new, built in 2025. That's not the year that we're in, so I'll have to fix that. Because I kept watching leaders chase paper permission slips around the cultural hall the Saturday before every activity. That's actually true. The waiver process it replaces is about 40 years old. I like our odds. Don't trust me yet? Google my name. I have a Wikipedia page. I built 4 companies, yada, yada, yada. I'm not going to risk it for $13. That would be embarrassing. If you hate it, I'll refund you, no questions asked. I'm the ward member who got tired of a clipboard. There it is. It pulled all this from my website. Boo, Wikipedia. Okay, Koerner Office. Awesome. Get in touch. That works. I tested it earlier. 2026 there, but it says 2025 in the blurb. Got to fix that. Now, this is another thing I did. Resources page with one prompt. I added a blog that's SEO friendly. How to send liability waivers. I want to get ChatGPT to start recommending me. Still has my pricing. But what I'm most excited about is this. I now have like a live preview page so I can see how the form filling out process actually looks and feels without having to pay $13 or $17 as a phony customer and trying it myself. The version of the form that was live a few days ago when I last filmed was not thorough enough. It didn't ask for allergies or emergency contact. It just said, what's your kid's name? Sign here. That's not going to fly. So now it's like super, super intuitive. And I used Claude. I'll show you the prompt that I used later, but I used Claude to like describe in a technical way how this prompt should look. Address, emergency name, medical info. Look at this. None of these apply is the default option because most people don't write these in. If you check yes on any of them, then none of these apply reverts back. That way, if no one checks anything, then It will fill out the actual form with the answers submitted on this page. Okay. And then you can add other kids and it's super cool. And most of it's going to be done on the phone and you can just use your finger and sign. So pretty dang cool. This is what my app is filling out. So in the database of the church leader collecting all these forms, they're not going to have something that looks different than this. They're going to have a bunch of this exact form. That's filled out with a digital signature carried over from my app. It's gonna work. So all that to be said, the homepage is live. Let's start marketing this thing. This is gonna be the fun part of this. So I'm going to start with Reddit. I'm using Fable 5 to, to type a Reddit post for me that I can post in the Latter-day Saints subreddit about what I've built. I think I have a whole other video about marketing on Reddit. There are ways to profitably launch products on Reddit. But you have to go through the backdoor's backdoor. In my experience, the best way to promote on Reddit is to just write a true factual story about what you've done and why you did it and not link to it, not even name what it is, just tell your story. And then as it starts to do well and you see the views climb and you see people are commenting, A, answer comments telling people what it is, because that's usually allowed. B, you can edit the post and add a link Or the name of your thing near the bottom of the post. Sometimes if moderators are really strict and grumpy, they'll delete your post when you add a link. Usually in my experience, they won't. They'll leave it up because you've de-risked the post by then. The moderators already see that their own community loves your post and they love you and they can see what the upvote rate is. It's 93% of people are upvoting it versus downvoting it. They can see all the comments, engagements, they can see the sentiment and they're like, probably going to do more harm than good if we delete this this post. So we don't want to do that. Okay, but Claude wrote a story here that is not true. Okay, so I basically just told Claude, don't lie. Meanwhile, we're going to go over to Facebook. We are going to see if I was accepted into any of these groups. Okay, this one has 500,000 members. It's a public group. 300 likes. Okay, 200 likes. Okay, this one looks pretty active. There's almost too many people in here and there's not a ton of engagement compared to how many people are in here, but it looks good enough for me to post. So I'm not gonna use AI to post this 'cause sometimes it's faster to just write it yourself. So I'm just gonna say, okay, so I'm speaking to the pain point and posts that have a picture or a video of some sort always perform better, always. So I'm just gonna screenshot the waiver itself so they know what I'm talking about and I'm gonna post it. And should get approved. You know, different groups have different rules. Some groups go through like the moderators first. Yep, this one's going to the moderators. I don't know how long it will take to be approved, but it should be approved. And then in the comments is where the magic will happen. I will talk about my solution and basically finish what I was saying in that post. Let's go to another group. Let's do the same thing. Surprised there hasn't been a better solution for this. Okay. And it is actually interesting. Sometimes I see groups that don't have much engagement. But if you post like a thought-provoking question that is designed to generate replies, answers, comments, Facebook will start showing it to a lot more people in the group than it usually shows posts to because it sees that this post is an outlier. It's getting a lot more engagement than normal because you're soliciting comments. So it will show it to more people. And sometimes it's not that hard to be the best performing post in a group, even if you're brand new. Okay, let's just try it. Same thing. And we're gonna screenshot the Waiver again. All right, post. Okay, submitted for approval. Now let's go to our Facebook ads. So right before this livestream started, I found this post in the Latter-day Saints subreddit that's pinned and it says self-promotion time, 28 days ago, and they're encouraging people to promote themselves. So I posted right here. You can see 10 minutes ago it was posted. I have one view and it's my own. So I didn't expect it to do much, but it's an asymmetric bet. If I wanted, I could send the link directly to my comment to a bunch of friends that are on Reddit and say, hey, can you upvote this? And that would kind of send a signal to Reddit that maybe we should surface this to more people because it can, it can sort comments however it sees fit. So let's go in here. I want to see how active this community is. 3,000 weekly contributors out of 53,000 people. Now let's see how that compares. 3,000 out of 53,000. This one has 400 out of 9,800. So that's 4% versus like 8% on the other one, give or take. So I win in 2 different ways. This one's 5 times bigger and it's twice as active. We're gonna go to hot posts. I love doing this. You can totally see what type of post resonates with people. Top post this week: Anyone else just not care about church anymore? All right, that was 6 days ago. Opening chapters of Job, which we've been learning about this week. Some drama. So we'll go to create post and we're gonna paste that in there. And I'm gonna do kind of an intriguing title. That looks too perfect. So I'm gonna take it back to Claude and say, I said, add in some typos and sloppy grammar. Just a few not perfect sentences needs to look more real. All right. It's a little better. It still doesn't look super realistic. And then let's go ahead and, so we're speaking to the pain points. We're gonna screenshot it. And we're going to add the image. Actually, we're not gonna add an image 'cause in some subreddits it can put it in like a different section. So I'm gonna leave this image free. Okay, so it's telling me to repost in other communities. Okay, so I will post it in one more subreddit. Okay, start running my Facebook ad campaigns again. Posted twice to Reddit and it's already removed by the moderators of the second subreddit. So that's okay. I have 49 views on the original subreddit. And right there. Okay, so we'll go back. Look at this. It was removed by the moderators, which means they probably don't allow cross-posting. Like it was an automated thing, not manual. So 58 views. It's kind of crazy. Like 58 eyeballs on this within the first 3 minutes, literally 58 eyeballs for free. And I don't have like a paid Reddit account. Like this is just, I don't have a bunch of followers. The whole premise of this video is I can't use my audience, which is one reason why I'm posting to Reddit because Reddit's not a place where you really build an audience. So 75 views in the first 4 minutes. I'm enthused by that. Now no one knows what it is yet. No one could go to my website, but theoretically people will start commenting and I can reply to those comments. And it's tedious, but it's worth it. So let's go over to Ads Manager. There it is. Okay, check the box. There it is. It's learning right now. LDS selfie video right there. We're going to change the date to today. Update. We do not have any landing page views yet, but that's okay. We're going to A/B test. We're going to add 4 more to the primary text, the headline and the description. Super easy. We'll go back to Claude. My prompt to Claude: take the one ad and A/B test 5 different headlines, text, and description on it to accelerate my learning and results. Also do an audit of what we can maybe fix in the ad. Very impatient. Don't want to make a new ad or get it approved again. Okay, let's look what Claude is saying. Any copy change is technically new creative, so a re-review is physically unavoidable. There's a version. Okay, we're going to duplicate it. Multiple text options. Dude, no, you do this for me. You add the UTMs and those 5 variations for me is what I said. I don't know that it can. It only has so many permissions. A UTM parameter is basically adding something like that to a link so you can track where that link came from. It says this will reset the learning phase, but our learning phase has only been like an hour. We ran this, this ad on Tuesday for like an hour. So I'm not really worried about resetting anything. Another thing we can do is duplicate the ad. And A/B test it. That way we don't have to reset anything. I'm actually going to stop it and tell it to do that. Let's see if anyone has read it. 192 views, 2 upvotes, 1 comment. Let's look at how much Claude is doing for me here. Look at that. It's creating ads. Whew. That's so cool. Like if you've never run Facebook ads before, you couldn't know how magical this is. Facebook ads are either thousands a month paid to an agency or Hours per month banging your head against the wall. Tedious, tedious. A/B test. Try this, try this, test that, watch this, babysit that, fix that, that broke. Make more creative, more videos, more ads, more, more, more. Like it is a literal full-time job. So to just type a prompt into Claude, it's insane. Look, done. I don't even need to reprompt it. Duplicate it. Done. Okay, so now we have 6 ads running, but I'm not sure if what it did, which is make 5 separate ads under one ad set, is better than just making one ad that has 5 variations of those 3 things. Is making 5 ads in an ad set a cleaner A/B test than 1 ad with 5 different variations of text, headline, description? No, it's messier. Then why did you do it? All I wanted you to do— should we still do that or roll with this? Said we should roll with it. All right, it is— it does give me per-message UTMs, as in I can see specifically in Stripe where the paying customers are coming from, which ad, whereas If we did this all within one ad, I couldn't. Let's go check on Ads Manager ourselves. Wow, 2 of the 5 are already active and these 3 are in review per landing page view. So remember, Meta is optimizing, AKA working hard to try to find people that are more likely to click and actually get to the landing page, which is not the same thing as merely clicking. Some people click and then they bounce. They need to land on the landing page and Meta knows that they're landing there because I have a pixel there. If you don't know what that means or how to do it, you just say, Claude, how do I add a pixel to my website? And it will tell you. We're gonna check on Reddit. 319 views in 16 minutes. This is one of your top 100 posts of all time. 4 upvotes, 5 comments. Gonna go back to Facebook Ads Manager, refresh it. All right, we got some views. This doesn't take long. It's been like 10 minutes since I made that live. Here we go. The original one that I posted 3 days ago has 11 clicks, 11 landing page views. It's reached 3,200 people. The frequency is 1.12, which means of these 3,200 people, they have all seen it on average 1.12 times. So 10% of people have seen it twice. I think I'm doing that right. And I'm paying $1.49 per landing page view. Sheesh. That is a lot. It can only go down from there. Hopefully I'm paying $4.56 per 1,000 views, 17 link clicks, $0.96 cost per click. Okay, so I'm spending $1 cost per click, $1.50 per landing page view. So I think that's like 2/3 of clickers are ending up on the landing page. Click-through rate is only 1%. That's not good. All right. Now let me show you how to do this. So you got to go to columns, performance and clicks, and go to customize columns. And we can look at all of the columns that only we want to see. So I already know they're delivering results. I don't need to see that frequency. I don't care. I don't care when it ends, what the schedule is. I don't care about the quality ranking, engagement rate. I don't care about any of these. I don't care about CPM, shop clicks. I do want cost per landing page view. And then you can just go down here and see what else do I want to see? How many 3-second video plays? How many through plays, which means 15 seconds. Remember my video, my ad is 31 seconds. A through play is them watching at least 15 seconds. I wanna see that. I wanna see cost per unique click. I don't need to see those. Some of these are redundant. I'll see that. And then I'm going to hit the dropdown, save as a new preset, only waivers. So now when I have this preset selected, I will only see the columns that I wanna see. I don't need to see actions or reach. Let's look at Google Analytics. Only waivers. View real time. 49 active users in the last 2 minutes. I'm going to make sure that my Contact Us form still works. And it doesn't. My Contact Us form does not work. Oh, another thing I did over the last few days that I switched the email sending service from SendGrid to Resend. I got them confused. I always use Resend. I already had Resend connected to my Replit account. So I just, I use that again. All right. Hopefully it fixes the contact us form. Go check on ads manager. So my cost per landing page view was, what was it? $1.50. If you look at this, it has already dropped over the course of 10 minutes to $0.65 from $1.50. So that's great. I'm spending $0.47 per unique click and $0.65 per unique landing page view. Click-through rate, what was it, 1% before? So that's gone up 14%. We have 158 3-second video plays and 23 through plays. Ooh, a lot of traffic on the website at this point. 22 users in the last 5 minutes. It was just 5. So people are using it. You know what? I kind of just want to like only show this ad to people in Utah. Or Utah County, the highest concentration of members of the church in the world. I would bet that some of my new ads are active, which is why we went from 5 to 28 users Over the last 5 or 10 minutes. Okay. Yep. There we go. The crayon story video has 6 impressions and 1 click. Click-through rate just went up a little bit more. We'll refresh Ads Manager. Click-through rate went up again from 1% to 1.14% to 1.3%. That's a good sign. On this ad, it's already 4.5%, but it's a small sample size. Just as a way of zooming out, like I used to have a lot of heartburn with Facebook ads. How much they cost to really learn. But then I thought, man, how much is tuition? How much is college tuition? It costs $100,000 a year to go to USC. And I know that's a random stat in a random school, but I just— my best friend Nick went to USC and he told me that yesterday. It's like twice as much as what it was when he went there. $100,000 a year, $400,000 for a 4-year degree. You can buy a lot of Facebook ads for $400,000. You can do a lot of learning on an offer with $400,000. This one's got a 20% click-through rate. I should keep waiting. So opportunity score, 96. That's a grade out of 100. I don't know if you remember this, but when we started the ads, it was very low. It was under 50. But as Meta starts to see, oh man, maybe this guy kind of knows what he's doing. Maybe the fact that he rebuffed all of our sales pitches was a good decision for him. Hmm. Who would have thought? And the Contact Us form is now fixed. One prompt. You heard the ding in the background. I just sent myself a message on the Contact Us form and it is fixed. Also refreshing Stripe. Ooh, okay, we got a lot of impressions here now. See, this is where all those Google Analytics visitors were coming from. We got even more now, close to 30. Look at the click-through rate. The more impressions, the more accurate this number, right? Because these are very small sample sizes, but this one has 8% of the impressions as this one, 15 times better click-through rate. I like what I see here so far. I wish I could see who these ads were being shown to. Maybe Claude knows. All right, we're going over to Claude. We're checking on the MCP. I said, how can I see where these ads are being shown to? My API access can't split stats by location, but Ads Manager does in 2 clicks. Campaign breakdown menu by delivery region. Okay, let's go do that breakdown. Okay, here we go. Geography region. California, Florida. Where's Utah? 41? Come on, dude. I mean, either it's magically finding members of my faith in all of these states, which it could be, but I don't, I don't like this. Most of these states have 3% members of my church. I don't think it's that good that fast at finding them. All right, I'm gonna tell Claude to fix it. So I'm gonna screenshot this. Big problem. Those states do not dominate. Show in Utah only, or even Utah County only. Make the change ASAP, and it might need to like make a new ad. I don't know. Okay. Wow. It fixed it. All right. That's so cool. That is so cool. Utah only men 20 to 60 hard age cap kept all six ads still active. No re-review takes effect. You're right to call it Meta's optimizing for cheap clicks anywhere. I would push back on Utah county only county level isn't a Meta option anyway. Good to know. Okay. Cool. Let's based on the screenshots. I sent— I want to see that if there's any correlation to success in those LDS-heavy states or not. That's interesting. I had 7.3% click-through rate on 41 impressions in Utah, the absolute highest, which makes sense, but it's small sample size. But that's what my results could look like, right? I was getting 1%. That's 7 times better early on. So that has me excited, but it doesn't seem to be correlated with States that have a lot of members of the church. Okay, now I'm really excited to see how this does. Still no approval on the Facebook groups. Reddit seems to be dying down. We're at 1,400 views, but this is the number 8 post on the Latter-day Saints subreddit today. So there's something. All right. So if we go to Google Analytics in real time, we'll see Utah is dark blue. That's looking great. Pricing. 3 people were on the pricing page. Create a waiver. 6 people were on that. So people are going through the funnel. Form start. This is looking good. All right, I'm gonna refresh Stripe. Oh, okay, okay. I see a 13 in here. I haven't clicked on it yet. I'm gonna change my screen over. So this is my real Stripe page. You can see the timestamps, the day, these $19 and $38. These are business plans that I sell for my YouTube videos. Okay, they come in right here, but right here, invoice subscription creation. So this one right here, this is me. This is my test transaction that I did this morning. Right. And I can't click on it because it's going to show my home address, my number and everything. This looks exactly like this, but this is not me. All right. This is a different card number. And this happened minutes ago. I think this is a customer. I think this is our first customer, but let me click into it. I'm going to change over to Google Analytics so I can see actually who it is. Dude, the first dollar for any business is the best dollar in the world. It will never get old. I don't care if it's literally $1, $1,000. Doesn't matter. We got people in Utah, Idaho, Spanish Fork, Tooele. Okay, let me click into Stripe. Let's go. Boom. My dog just jumped. Sorry, bud. He was asleep. We got a customer. Only Waivers yearly plan, $0.68 in Stripe fees. Idaho. It's not even Utah. It's in Idaho. It's probably that person that's been browsing in Boise who's still on my website in Boise for a while. So There it is, folks. We did it. I clearly lost money. I've spent probably around $100 in ads so far, including the time I spent offline, 6 hours building this app to make $13. But I feel amazing. And I'm going to reach out to this person and wonder like, what are they doing? What are they using it for? Yes. Hell yes, Dr Pepper. It is $13 in my Stripe account. This is real. It's confirmed. Card ending in 2525. I think that's, that's what it was. Beautiful. Guys, thank you for watching. Thanks for hanging out in the Koerner office. Uh, this has been fun. If you want me to do this again, then let me know and I will. Love you. Bye.