Intro to VIBEcoding with the I ❤️ AI Community

Don't say anything, it won't automatically put you on camera, but if you say hello or ask a question, someone on the internet might see you. So welcome everyone. So glad to have you here. It is our Fab Friday. We are here to talk all about, uh, of course this is our time for our, our I Love AI community to come together and get to know each other and update each other on what's been working and not working. Uh, we also of course have our community chat where we do a lot of that as well, but this is our chance to come together. Together live. But today we have a special, special, uh, opportunity to share a little bit about a very specific topic. And so you might have seen some of the posts today around vibe coding. And the reason I wanted to talk about vibe coding is because Lovable has launched a really cool campaign called She Builds, um, on Lovable. And of course, there's many of us that identify as she's that are building on Lovable. And I was like, what a great campaign. Uh, so on Sunday from midnight, meaning Sunday, and it's kind of cool because I don't know if you know, but do we do this globally? But at least in the US we do this time change thing. Do we do that everywhere? So we actually get an extra hour of usage. Uh, but Lovable is, um, at midnight is going to be, uh, opening the doors, allowing you to get on the platform and Vibe Code credit unlimited for the day. Now, granted, you don't get access to all the features that you would need to productize, like cloud and certain security capabilities, but you can do a lot of ideating and a lot of practice. And of course, I am all about learning by doing, so we are taking advantage of that. Now, because I do think it's fun to build together and build in public, We are also hosting a live 2-hour just vibe session. There is no curriculum per se. Um, of course I'll give you my tips, but we have lots of builders in our community. So if they are on, um, Lovable, uh, because they are running this campaign, there's so many cool things that are going on. For those of you who are in the school community, you can actually see it on our calendar. And when you click on the calendar in the school community, you'll also see all the cool things that they are doing right now. Oop, I forgot to approve note taker. So sorry about that. But that is in place now. So you could just head out to— I think I just Googled like, or perplexity or whatever, She Builds and She Builds on Lovable. I guess I did.

Maybe.

Anyway, let's see here. Oh, that was the She Builds website. Where I volunteer. Just kidding. So I'm going to go back and— She Builds Lovable. Let's see. There we go. Yeah, that's what I searched for. She Builds Lovable International Women's Day. It's on a Sunday, which is amazing. So I hope all of the women who are listening to the sound of this take some time for you, get an opportunity to do something that you truly enjoy. It is just a day to celebrate. Um, all of the things that we have done. I also think it's a great day to read, uh, you know, maybe work out more than you normally do. Like I always say, self-care isn't always a bubble bath. Um, sometimes it is, but sometimes it's doing the thing we've kept, you know, ourselves from doing that we wanted to. And that's why I love this, uh, campaign. 'Cause one of the things that I truly do love to do is just build to build and see where my mind takes me. And it's never been easier than to go from idea to execution than it is today. And so this is, an opportunity to just carve out a little bit of yourself, 24 hours. Now, in addition to being able to build on the platform, Loveable is offering completely free 24 hours powered by Anthropic. You know, I'm a fan, love Anthropic. Um, but participants are also, you are gonna get $100 in free Anthropic credits. That means that you can at least maybe MCP yourself to a cool feature that you've built, another tool maybe that you've created, or just give you an opportunity to play in that environment, but also give you $250 in Stripe fee credits. Like, so I don't know about you, but I'm like, wait, wait, wait. So if I build something I can monetize, that seems like a challenge. I feel like, I feel like that's a challenge. I feel like we should do something where we can build something we can sell. And ultimately the best thing you can sell is your experience, right? What you know. So that's what I'm gonna encourage you to do from today until the day we get this little free window to leverage Stripe and Anthropic. Like, I don't know. I mean, some of you know Claude, right? On Claude, they're pretty good at artifacts these days. Like imagine building something that codifies something, you know, doesn't have to be everything, you know, but something, you know, that would be helpful to a community of people and put it up for free. I just think it's gonna be pretty cool. Uh, it also serves as a really great opportunity for us to kind of kickstart our march with building, but more importantly, building for like as a business, right? Building something that'll help you generate revenue as a company, solving problems, uh, identifying a problem you can solve. And so I think it's a really good opportunity maybe to play around with some of these ideas that you have. I have so many. I have a Moleskine notebook full of things that I'm like, I wanna build this. Noelle, focus. You have a business to run.

No.

Oh, and this. Focus. Do the thing, right? Some of you can relate. Um, but now I can spend all day Sunday, just build, build, build. Now I will tell you, for those of you who know me, Um, we are Girl Scouts, uh, in our house. I grew up as a Girl Scout, my daughters are Girl Scouts, and so this weekend, of course, uh, Saturday night is what we call Mall Madness. And there's a mall here in South Florida that opens it up to all the Girl Scouts in the area, and they get to spend the night in the mall, which is madness. But anyway, um, it's the first time my daughter ever did an all-nighter, and it was very fun, and we get to shop and they have a bath— they, they have a Bath and Body Works. I mean, that's all you need, really. So I will be doing that until about 3:30 AM and then it's on. So I, like I said, we will— I will be vibe coding in an open room on school.

I'm—

it's not, not a class. It's just for fun. But if you don't want to do it alone, if you want to share your success as you go along throughout the day, don't hesitate to join us. Like, it won't be open the entire day, but I'll do one morning session, another one at midday, and then we have a dedicated session on the calendar at 4:00 on Sunday. So that's what this is about. But before— right now you might still be thinking like, that sounds fun. I don't even know what it is and it sounds fun. But let's talk a little bit about vibe coding and understand the concept and what it is and what it does. And the reason I think it's a really, you know, What, whatever mechanism inspires us to learn something new, I'm a fan of. And so this campaign giving us credits, giving us the chance to build something more than just an idea, but actually build something we could monetize is super exciting to me. One of the biggest challenges though is that when you start a new platform, if you're not really clear about what that platform is good at, what that tool you're using, why, why you would use it, what that tool is good for, you may end up bringing to the table a problem that's not suited for that solution. So I'm really big about being kind of, well, not kind of, very intentional about my tool selection and problem identification to make sure I'm using the right tool at the right time for the right problem. Specifically, I do this for my customers, but today, um, I'm doing it with you all. So hopefully by, and, and it's gonna be really exciting. I've opened up the registration for this vibe coding session we're doing on Sunday to everyone. It is also a bit of a taste of what you're gonna see in my next big event that's happening, which is gonna be at Mindvalley. I don't know if you have seen the Mindvalley invite, um, but I'll throw my specific link into the chat so that way you can come, you can join. I will be trying to find you all. We're gonna have this like sea of, uh, uh, little pictures of everyone who's on the the session, I think you have to be VIP for me to see you. But if you're a VIP, I'm gonna find you. If you're in my community, I'm gonna call you out in front of 150. Well, maybe that's gonna make you nervous. I'm just gonna wave at you and, and make you feel welcome. Um, but I won't ask you questions or anything, but I do want you to know you're seen. And I'm super excited that we're in a family of AI leaders together. Uh, so let's talk a little bit about what vibe coding is and why it's important. So this concept came up and I loved the day they came up with this topic. I think there's a few of us here that the second it went, like the go-to-market strategy launched, we were like, that's, that was like made for me. Like Vibe Coding is like made for Noelle, made for AI lead. Like it's just what we do. But what it was solving was a challenge, right? We've now seen AI technologies get very, very good at engineering. Like very good to the— so good in fact that companies like Salesforce, Amazon, Microsoft are declaring how many lines of code these AI systems are creating. But not only that, how much value and time, like value they're providing, but time they're giving back to engineers to worry about the more complex use cases. Most importantly, I think the thing I, I ended up going to, uh, speaking at the RSA conference last year, and Anthropic was there, of course, Claude Code, um, really big, uh, go-to-market efforts last year to get everyone thinking. And one of the things that, that they said as part of their strategy was that when you have an idea to do something that solves a problem, so that was the first, first positioning, identify your problems, clearly understand what you're struggling with, what is blocking you, what is keeping you from doing the thing you wanna do. Um, and in your case, like let's say you're a small business owner, the thing you wanna do maybe is make more money. What is stopping you from doing that? Like, where's the friction? Is it you're not getting out on social? Is it that you are not generating enough advertising that's converting? Is it you don't have enough, uh, or, or your sales aren't good? Like, where in the process as a business owner is it breaking down? And if you understood it well enough, could you just ask an AI system to solve that problem for you? Well, that is what we saw en masse in 2024, 2025. And in 2025, as we started really honing in on this concept of Vibe Coding, we had an opportunity to, to see what was possible. Uh, Claude Code ended up saying as part of their— and they did a developer day, and in the keynote they positioned Vibe Coding as indexing on the importance of your idea and clarity of your understanding of how you want that idea to come to life. And the fact that the infrastructure, the code, the scaffolding, the language you choose to build in, that all of that is relatively— well, it's irrelevant actually. That's what they said.

It—

the code you build over time will become irrelevant. And I didn't see— maybe I saw it a little bit back then, but I always like to tell you all this, like this, it dates me a little bit. When I first got into engineering and coding, it was right at Y2K. I was learning, I was a Fortran developer. I had become like crossed over to the dark side, uh, to client-server. I learned a language called Visual Age or Smalltalk. I became an expert in Visual Age Smalltalk, an IBM product, um, that we were selling at the time I was working for IBM. And then Java came out, Enterprise Java. Do y'all remember these days? Anyway, so Enterprise Java comes out and I'm like, this is it, I know it. And here is the Like what happened then? Today I'm like, oh my gosh, how interesting these cycles go. So this again, 2000 and, you know, we're talking 2001, 2002. I'm teaching COBOL programmers, Fortran developers, people who are building, you know, very, like functional, procedural, you know, these huge blocks of code, monolithic apps. I'm teaching them the art of Java, and soon, not too, not too far away from that, did we move into web services and breaking up code. And I remember, uh, being on the team that was teaching about 10,000 people that there was a new way to write code, that it wasn't in Notepad or Notepad++ or Borland or whatever those tools were, right? That there was a new way to build, and that new way was an integrated development environment. Do you remember this? IDE. Right? And I would tell them like, look, and that's, of course, I grew up in an IDE. So I was like, this is, why would you write it any other way? It seems silly to write code in a word processing document as opposed to in an IDE where I could hit Ctrl+Space and get all the options of code I could use. Like, I was so fast at writing code compared to those using Notepad. So I was a huge advocate. And I remember them looking at me and being like, there is no way the code I write in my Notepad There is no way the code you're writing is better than this. I am an artiste. I have defined the boundaries upon which my code will execute. I have, you know, I've used the most impressive, longstanding best practices in the industry to build this. Your platform is taking shortcuts, and there's no way that it can do the things I do. And of course, I eventually— well, some of them I never did convince. You know, you know those people, right? Dig their heels in. They're like, no, I'm gonna— I mean, we— you might know some of them now. I'm gonna build on Emacs and I'm gonna use Notepad++.

Okay.

But there's a new world, an integrated development environment world. Enter VS Code, right? Within 10 years. So we've now almost millions, tens of millions of developers are now building code in an IDE. But here's the new evolution. I now get an opportunity. In the middle of COVID to work on a model that would eventually become GitHub Copilot. And so they're like, look, a new way to code. Sound familiar? This is going to save you a ton of time. This is going to allow you to predict like the entire scaffolding of an AI, of a software system. And I— these words came out of Noelle Russell's mouth. I was like, oh, I'm pretty sure I can write better code.

Then.

And I had this—

it wasn't in that moment. Like, I was pretty like adamant during the beta testing when I was testing out GitHub Copilot. I was like, I'm not sure I want a model that can generate code. I'm not sure, you know, like this was 20— like 2019, 2020, right? So I'm like, I'm not sure that's a good idea. And then I did— I'm just being very vulnerable with you on my like naïveté at the time where I was like, I'm pretty sure not only is my code better, But I'm unique in the way that my code is written. I am special. And the way that I think about it, I'm Latina, I am a woman, I have 6 kids. Like, I just, I go through the process of engineering a solution differently. And therefore, these predictive models can't see me. Now, granted, I think there was some truth to that, but as we all know, these models can learn. And so the model I was using was on inside of like on device, right? So it was learning as I go. So within 6 months I'm like, okay, so I, I see. Okay, I see. I, I was the COBOL developer. Like I literally was like, my way's better and I know it takes me longer, but it's better like internally. And it's almost, I use this in sales conversations, it's kinda like calling someone's, you know, calling someone's baby ugly. Like you never do that. Even if, you know, you look at the baby and you're like, ooh, cute baby. Um, like it's a terrible analogy actually. I have 6 babies and they were all beautiful. At least that's what people told me cuz they wouldn't dare say the alternative. But in the same way, you're, you know, no one was willing to kind of combat me at that time. And I didn't combat those people. I just tried to show them the way I led them to water. I'm like, there's a new way to build. And now we are in a world where the new way to build is to use your words to describe the problem you're trying to solve. And the more clarity you have in that description, the easier it is for a model, an AI system, to understand what you're trying to do. My favorite way to vibe code is through Socratic prompting, right? Where I don't feed it exactly my problem. I ask it to do maybe chain of thought or tree of thought prompting with me, right? I ask it to kind of Here's the problem I'm thinking about. Help me think through what the best solution is based on the outcome I want, right? I lead with the outcome now with my AI models as opposed to my very narrow thinking. And now I've learned like over 20 years, I'm like, wow, I became the one who was like, my box is where I wanna be. This is what I own. This is what I'm good at. As opposed to seeing what my brain gets to do when freed up from scaffolding logic and writing code. And so there's a lot of controversy, uh, about this. There are people today that are like, my code's better than AI-written code. That's probably true right now, but it won't be true forever because of the beauty of self-learning models that they are gonna learn from you, from me, from those of us using this tool. We're gonna tell it what good looks like. And a lot of us are spending quite a bit of time refactoring AI code for a living. And in that refactoring, we're training these models on how to build better code the first time. So this is a temporary middle ground that we're in right now, but I think it's, it's an incredible, incredible opportunity. Now, vibe coding means that the platform is giving you the opportunity to use your words and the outcome is a software engineering asset, meaning it's a web application, a mobile application, a dashboard, but it has, some, I, I call it like an N-tier architecture. There is a user interface, there is, uh, some business controlling logic that's deciding what to do when the user makes their request. And then there's some backend where the database is stored, right? Some, that's 3 tiers, but it could be N number of tiers. So it could be multiple database layers, there could be multiple middle tiers. But the idea is like, I wanna, as the user, I wanna be able to just ask for what I want and I want the software to go through its architecture to deliver that value to me. There are lots of vibe coding platforms out there. As you know, we're gonna be playing around with Lovable. However, I'd be remiss if I didn't talk about the one I grew up on, which was GitHub Copilot. GitHub Copilot provides vibe coding, but I do think it's important to talk about different types of vibe coding. So vibe coding is a great word. I say it, I'll probably say it, I don't know, we should pro— It's not a good drinking game, but we should probably count like at the end how many times I said it. But the concept, right, vibe coding means the vibe part is not necessarily having to understand the language or syntax in which you will use to write the, to build the solution. And that is a huge freeing opportunity. Now, I started my, my life as a programmer started in Java. Java is extremely strict compared to the languages many people are learning today. And as a result, I became very disciplined in my approach, right, to building software. Today though, that's why I think Claude Code has said things like, um, the infrastructure, the code, the language is irrelevant. And for everyone who's not an engineer, you're like, well, that sounds great to me. I would love for it to be irrelevant. But anyone who is an engineer, like as a lifestyle, as a livelihood, might be like, wait, wait, wait, what did, what did we just cancel here? Um, and so, uh, no, uh, great question, Gabriella. So Gabriella asked a question, um, and there might be others. That one just happened to pop up in front of me. Uh, oops. Let me go over here. Um, yeah. Did I say Emacs? I did. Do you know Emacs? Anyway, let me go back to, I'm, I have to scroll down. Um, so It could include fine-tuning a model, but it doesn't. You actually, you want, you don't even really wanna know too much about the model in this case, right? So Claude Code, its job is to know, to build a model that knows how to write code based on your words. You're not gonna have anything to do with that. You're not gonna go in and, you know, tell Claude Code what, how to do its job. You're counting on the fact that it's going to make the engineering work irrelevant to you. You just get to say it and it builds it. What we have found though is that right now there is, uh, a high level of efficiency and effectiveness that comes when you do, do not necessarily, uh, traditional fine-tuning by like uploading a bunch of code samples to the data, to your model, which you could do, but instead using what we call instruction fine-tuning, right? Where you're actually saying, here's what good looks like through natural language. You're describing good behavior. And this is how I operate with Lovable, right? When I do interactions, I'm telling it like, I, I see what you did here. I appreciate the effort. However, I would— I actually want it to be this way, or I wanna make sure it's applying these security policies. Yeah, exactly. Context prompt engineering. Yes, exactly. I love it. Let's see, what else did I— what else did I see here? Oh my gosh. Yeah, Six Thinking Hats method. Absolutely. I love that. That's like the tree of um, thought kind of prompts, uh, process, right? Where you have multiple dyna— like multiple coherent and consistent activities happening simultaneously that are all feeding the same solution. Um, okay, awesome. So, so back to this concept of different types of vibe coding. So there is the vibe coding, like many of us who are like, I'm not an engineer. I don't even know if you ask me to build a Hello World, which There's a whole new world of that. But if you ask me to build Hello World in Java, I wouldn't be able to, or on Python or in any language whatsoever, JavaScript. Like you don't know enough about the programming language to actually do anything in it. VibeCoding has unlocked the ability to just say something and build an entire functioning application using multiple technologies, JavaScript, Angular, like, you know, TypeScript. And you don't know any of that. Again, the tech becomes irrelevant, right? You just ask for it and the model builds it. There's value in that. Again, old Noelle, Noelle 5 years ago when I was testing out this capability was like, so what happens when this thing breaks? Or worse, do you even know it's broken? Right? Like if it's functionally working, maybe okay. But do you even know like what vulnerabilities you have? Or like in the early days of Lovable, there were no security reviews, right? They were just like, they didn't actually think, I don't think they thought people were gonna like put their code in production. As a result, they've hardened the platform, right? They created, um, Lovable Cloud, which is basically like a hardened production system and dynamic security reviews, right? Against the most common security, uh, you know, attack philosophies, um, OWASP Top 10 kind of things. So I think it's interesting that to watch all of these technologies evolve, and that's why I do actually believe Cloud Code's go-to-market strategy that the engineering work will become irrelevant because it's become more relevant right in front of our eyes. Like, I've watched the platform get smarter about how it builds code and the transparency it offers users. But that's one side of vibe coding. Vibe coding for those of us who are business users who wanna build prototypes or even production systems without really understanding the technology behind it. And that might make certain people that are engineering people might make them a little uneasy, and that's okay. I was uneasy too. But the other, and, and the other alternative though, the other side of vibe coding is actually help engineers do their job well. And I mentioned this during our class in the AI Leadership Institute when we were talking about, oh yeah, we were just doing the Azure AI Fundamentals prep. My test is next week, so, but we were, I was teaching the Azure AI Fundamentals prep class, and we were talking about Machine Learning Studio. And one of the early, um, concerns with Machine Learning Studio, which actually uses AI to build AI, right? So it auto ML, if you've ever heard of this, this is like a, an automatic machine learning creation. And what I would say is like, give me your data source, right? Give me your dataset and I will, based on your dataset, I The machine learning, AutoML, will go and find the right algorithm to match the data that you have. I will do the feature engineering for you to match all the things that are the hard part, right, of getting a model up and running so you can start testing. And so similarly in vibe coding, we are seeing the same thing that inside engineering teams, it takes a minute because the first inclination is like, whoa. I don't need a machine to build my code. I'm an artiste. I build great code by myself. So there, that moment, we're still in the middle of that. Um, but as people begin to use it, they start to gain confidence and they realize, just like we talk about all the time in the AI Leadership Institute, when you, when AI is involved, we have a choice to make. It's called the human AI experience. We get to choose what the humans do and what the AI does so that they can coexist harmoniously. We choose, humans choose, adults, well, anyway, older, you know, people building stuff. It doesn't have to be adults, I guess. But I would say myself, I make this choice consciously. What am I going to keep? What am I gonna let an AI system do for me? So as an engineer, my choices are a bit different, right? So if I'm an engineer, I'm now maybe not because of my, you know, hang up on like, I think I build pretty good code that's diverse from the predictive code I'm going to get from a model, which I've been proven wrong. So don't worry, you know, I'm still— I'm always learning. I'm coachable. But in these moments when I'm not ready to give away that part that I love to do, I look for things that I am willing to give away. And some of those things, for example, one of the— it used to be like the bane of my development experience was writing test cases, like all of them, writing test cases, testing test cases, and then building code to the test. It became so non-creative that I was not at all enamored with software engineering. I was like, why do we even do this? A machine could do it, is literally what I used to say, not thinking that eventually a machine would do it. But this is the type of engineering work that most developers don't enjoy, writing the test cases that are going to test whether or not the code they are writing is working well. So having an AI build test cases, Having an AI do code reviews. I don't, I don't know if you are an engineer, but reviewing code is not the most exciting part of your job. It certainly isn't the most creative. Like you're looking at code, typically not yours, and you're trying to find bugs in it. Now, this reminds me of a time where I was working with a hardware company. They were building out silicon and their engineers, right? The FPGA engineers were staring at silicon and marking them like, this is good, like with their eyes. Looking at hardware and identifying whether or not it had been soldered correctly, like on an assembly. Like, I was like, wait, what are you guys doing? And they're like, yeah, because, you know, in order for us to do the next step, we need to know if it's good or not. And I was like, and meanwhile, they're stacking up on the side of their desk while they're doing this analysis. And so at Microsoft, we were like, hey, maybe we can help you out here. Maybe we could use computer vision, right, to look at that and then tell you which ones to be worried about so that you can shift your attention to higher value work. I don't know, like building the engineering that's gonna actually make that silicon valuable. And so that's what they ended up doing. But in their day-to-day, they were so in the weeds of doing the work, they never could get out of, like I always say, like when you're in, um, when you're trying to see the picture, but you're inside the frame, right? It's very difficult to see the picture of what you're doing when you're like in the work, building your plane while flying it. Sounds glorious and fun and a little scary, but it's very difficult to see like, is the plane built correctly? Does it even look right? Like, are the wings even— like, it's very hard to see any of that when you're just like in it to win it, like in the grind. And so that's where AI can be helpful, is to be this like outside perspective, or let it do the grind on the inside so you become the outside perspective. You get to choose. That's the benefit. But in Vibe Coding, uh, on the engineering side, it's very different 'cause we're not asking it to, you know, build us something and we have no knowledge of how it's being built. We're asking it to partner with us. That's why it's called GitHub Copilot. It is based on this concept of, do y'all remember Extreme Programming? Or yeah, like pair programming. And there was a whole philosophy on engineering. It's beautiful. It meant that you always got to work in pairs. It turns out we engineers don't like working in pairs, so it didn't take off. But, but it, the benefit of working in pairs is that you could work on the same bit of code together and get twice as much done. And then you could switch and review each other's code so you could get through the cycle of code reviews faster. Turned out though, We don't like doing it. Um, so it didn't really take off. I mean, some people like doing it cuz it's, it, it became its own little like, you know, um, pop culture cult thing. Uh, but for the most part, most engineers did not apply that, that methodology. But that's why, and one of the challenges was, was actually with human interaction. If I'm gonna pair with someone, I have to like, I mean, just as a software engineer, I tend to like putting on my headphones. And just, you know, like, like knocking something out. In pair programming, I have to talk to somebody and like plan with them and coordinate time with them and exchange code. Like, it was not, it's not really part of the ethos of like most software development teams. And so it, it might match your culture, but in most cases, at least at Amazon and Microsoft, it was not, it was not a good culture fit, this process. So instead, in 2018, they were like, what if We built a model that could do that pair programming work, right? But you, it wouldn't be a human. And it turned out engineers were like, well, I like that, right? Like, I, I love the idea of having a machine that I can partner with, that I can send stuff to, that can review my code. And then, uh, again, I, as I mentioned in the RSA conference, speaking with, um, Meta's CISO, talked a lot about one of the top use cases they use AI for is for code reviews and using AI to do the code reviews. And the reason why this was useful at RSA, which is a security conference, is that those code reviews mark a gateway point where bugs get released into production, right? So if you review code and you're like, looks pretty good, if you're a human, you likely have only caught 25% of the bugs that are in that code, just because that's the statistically how much you've been able to capture. If you are a machine, It gets better. You catch about 50% of the code, but you still loo— right now, again, this is 2026, but you still lose, right? You still have 50% of those defects going to production, which are vulnerabilities, uh, increasing your attack surface cybersecurity-wise, right? So that's extremely dangerous. So what can we do to help that developer get higher and higher levels of review and coverage for the code they're shipping? And that is where a lot of companies, I mean, if you ask like AWS or Microsoft, that those lines of code that this is, these, uh, um, machines are building, they're not all product feature code, right? A lot of it is the work that engineers either deprioritized at the end of the project. You might know this if you've run engineering work before, but at the end of the project you're like, we have 6 weeks to test., and then you get to the end and you're like, we can make it in 4 weeks. We could do it in 2 weeks. I mean, really, we only need 2 days and then it launches and you hope you got some testing in. So that testing window that was at the end of every project that would often get crunched as resource and time slipped, that no longer has to happen because it happens adjacent to the engineering cycles. And that I think is pretty cool. So these are very different sides of Vibe Coding. Today we are gonna focus A little— well, I actually don't think it matters. On Sunday, we're definitely wanting to equip all of you. If you have an idea to vibe code, you know, build a solution, whether you are an engineer or not, like it has no correlation to your desire to build something cool on Lovable. But the last thing I wanna do is over— take a chance to give you a few tools outside. I mean, I've mentioned Lovable almost as much as vibe coding today, but that, that's only 'cause they've got this really cool campaign that they're running and I love it and she builds and it's awesome. But there are many other platforms that offer vibe coding opportunities. Claude Code is one of them. As a matter of fact, I feel like Claude Code, or I should just say Claude, right? Claude, Sonnet, 4Six, Opus, they predispose to coding. So when I'm— I used to maybe 2 weeks ago be able to go to Claude and be like, help me with this copy. Like I'm just asking for copy edits. So I paste in my landing page copy and I'm like, can you help me with the landing page copy? Can any of you guess what ends up happening? It builds out the fully functioning hosted webpage, and I didn't ask for any of that. Like, I'm like, no, I was just wanting to tweak and make my copy high converting. So it's indexing on the idea that if you ask me for something, I'm just gonna build it out. Which is, it's fine. Maybe that's what you want. In my case, I already have it built and I'm just trying to edit the copy a little bit. And so every time I ask it for copy edits, it builds me an example of the page that copy will sit on. And I, I don't know, I don't think I trained it to do that. I'm pretty sure it's just, that's its, that's its default capability. So that being said, don't be shy about using Claude, right? To say, hey, I wanna build a landing page, or hey, I wanna build a, you know, a, a lead magnet to give to my clients. That's, these are things I'm gonna encourage you to do on Sunday is like, if you have a business or if you are a thought leader, I think every one of us falls into one of those two buckets. You might not know it yet. You might not have said out loud to anyone that you have ideas that other people might benefit from, but I'm gonna say it for you. Like, every one of us has lived experience that is worth sharing with other people. And now you have an opportunity not only to share it, but to actually get paid for it, even if it's $100. Um, I will tell you, overnight, people go to my website and they buy What is it called? The Upskilling Guide for— it's a generative AI upskilling guide. I wrote that like 2 and a half years ago. I update it all the time, but it's not a new thing, right? Gen AI. People download it every single day and it's like $9. And so every morning I wake up, I'm like, huh, people are still buying an upskilling guide. Like, it just humbles me, right? It reminds me that not everyone— it like, there's a few of us that are like in the AI world, but most of the world is still like, how do I get up to speed on this?, and so that's what I'm gonna focus on is like, no matter where you are in your AI journey, there are a bunch of people who don't know what you know. And so maybe monetize just that. Um, so that's what we're gonna do on Sunday, but there's lots of other tools. So I mentioned Claude Code. I would love to, um, have the chat. I know there's at least 6 or 7 of you. Um, if you are interested in sharing, unless it's lovable, and then I'm gonna ask you to wait till Sunday. Um, but if you would like to share your vibe coding tool of choice, and talk about it for a little bit. I would love for you to raise your hand and, uh, give everybody— like, if you have a different tool that you use, and I know you do. Yeah, Renee, and then Joe. All right, so we'll start with Renee. Uh, go ahead, and then we'll go to you, Joe, and then Nitin, and then Dawn. Okay. All right, good. Let's go.

Hey everyone, happy Friday! Great to be here with you.

Um, Renee, before you get started, let me just guide everybody on— if you want to make sure you see all the speakers, don't forget to go up into your view little box if you're in the Zoom and drop it down to multi-speaker, and that'll make sure that no matter who's speaking, they pop up to multi-speaker for you. If you just do the gallery, it might be— you'd be like, where is this person?

Where—

who's talking right now? So multi-speaker helps you get that. Thank you so much, Renee. Okay, continue.

Sure. So anyway, great to be here. Happy Friday. So I've, um, I have used Lovable, and— but then I've actually taken that code or other code that I've written, and I or I've started from scratch, and I really like Anti-Gravity. I've used Anti-Gravity. I've also used, um, I use Vercel, and then I've connected all 3 of those platforms again just with ideas. And I'll cut like and paste different things, screenshots, um, to and just show the application what I want, and it'll be like, oh, and you want this and you want these letters and this font and all of that stuff. And sometimes Noelle does really cool stuff and I'll screenshot her page and then I post it in my application. So you can also do it with Hostinger. You can build a landing page, a webpage even within the Hostinger AI. But yeah, AntiGravity is probably my favorite and it's made by Google and it's integrated really well with Firebase, Supabase, which are backend databases, Airtable, GitHub, and Stripe for payments. So I've enjoyed it very much and it can keep you up a lot though. If you, if you don't wanna ever sleep, start live coding.

Exactly. I remember introducing this to a startup founder and, uh, they almost immediately, like, I, I was like, you could just ask for what, and again, startup people, entrepreneurs like you and me, right? We're idea people. So that we've now, you know, ignited our ability to build out our ideas like immediately. Which is very hard if you are an idea person to not build it when you get the idea. And so you end up building all the time. Um, and then of course, you know, one other thing.

So when you're doing it, you also sometimes on the side, I'll have like the YouTube videos on how to do what I'm doing, like walking me through it. So that's a really solid plan. It helps you expedite your learning. But like one of the guys one day was like, yeah, get, get the, um, buttons to move and shake and be bright. And so I was like doing that at the same time. Like, I want you to be brighter anyway.

Yes.

Yeah. And y'all know me and my confetti, right? So I'm like always like, make it a celebration. How do we— I always add confetti to any achievable outcome in my— any of my apps. It's very Noelle Russell Russell. So awesome. Yes, absolutely, Maurice. So thank you for sharing that, um, Renee, like being able to do side by side. Uh, and one of the reasons I use the Arc browser is because it lets me do that very, very well. And I can save my side by side so I can have, uh, cause I, of course, um, have ADHD. So I might start one and then get an idea and then go to another one, but then I want to go back. And so it allows me to store all of my side by side video and lovable projects. Thank you, Sherylyn. That's amazing. So thanks for— Sherylyn threw in all of the links to all of the tools mentioned so far. So I want to go to Joe next, unless you— oh yeah, I don't know if you put your hand up, but I know you did physically.

I did, yeah. So I have been— I'm kind of an odd bird. You probably already know that. So I kind of focus on, as an architect, going documentation first. So I, I actually design what I'm going to build in documentation and then, and kind of tweak it. Just this most recently, I have been, I've been using Antigravity as well. But I, today with the release of OpenAI GPT-5.4 and some of the benchmarks that they've been getting with 5.3 Codex, which is basically blowing away every other model right now. I was really excited to install Codex, the app on my Windows box. It does not run on my, my Mac because I'm a little bit— I have a little bit of an older Mac. It doesn't have the Apple Silicon on it. But yeah, so I just— and I bought a course from Udemy on how I can get the best out of Codex. So that is something that I'm going to be looking at, but mostly for code reviews. So going back and testing and stress testing and doing code reviews on the code that I write. And most of the code that I'm writing, I'm writing in ChatGPT.

Yeah. So tell me a little bit more about what you mean by documentation for those people who are not familiar with the general term? Like, is it in Notion? Are you putting it in Word? Like, what are you talking about?

So, no. So the way I work is I like to organize my ideas, whether it's a solution or if it's a particular set of code. I use AI to talk it through and actually challenge my ideas. And then what I do is I take that and I create an architectural document, and I just say, give me a diagram, give me, you know, you could tell it what you want in the architectural document, but I want the specifics that I can pass to a model to start by coding. So give me all of the details in this document. I'll go in and check it and make sure that it, it, it got the right idea. And then I'll put it into Joplin. I use Joplin as an open source note taker, organized note taker. And there's in the community, I did a session on Joplin just to help people out, but it's really easy to like hit the little copy button in your browser session, whatever browser session it is, paste it right into Joplin. And, and so I use that. Quite often as my, as my documentation tool, then I can, I can go back and reference that architecture, I can improve on that architecture, I have an audit trail of all of what I've done in the architecture, I put my artifacts along the way. So in other words, if I have evidence of that I did this part, this step, I grab the evidence, I put it in there so I know what I've done so I don't have to go back and repeat it.

Yeah.

And for those of you who are following along, uh, you get to see a little sneak peek into everything that Joe is building in his, uh, he's kind of documenting it in the community. So, um, definitely really cool. Uh, there was some questions about the tools, but, uh, Sherlyn has at the ready, uh, your, the Joplin app reference. So thank you, Sherlyn, for that. I was like, well, that was fast. Uh, so that's awesome. Um, and yeah, so, uh, a couple— and of course Anti-Gravity is in there. Um, but you talked about— you mentioned this last time, you didn't mention it now, but it's a good segue to Olga, who's next. Um, but you're using Vercel somehow, some way, are you? Or have you just tested it, Joe?

Um, Vercel? No, I haven't. I have not.

Oh, okay. I was curious. I thought maybe you did, or maybe you met—

oh, maybe not.

Maybe Okay, cool. Well, either way, we're gonna— I heard, I saw, um, that, uh, Olga, you use, uh, Vercel and Cursor, which are two tools we have not talked about yet. So I thought maybe you could share a little bit about how you are using those.

Uh, sure. Yes. Um, actually I usually start with Miro because I—

Oh yeah, of course.

Yeah. When I work with clients, uh, I usually start with collaborative workshops where we co-create some ideas, concepts together, and we document them on Miro. And they can be just Post-its, they can be in any format. And then because I come from product design background, I'm used to designing wireframes first. So if we're talking about complex apps, like if we, you design a website for whatever, a shop, you don't need that level of thinking. But at the moment, for example, I'm designing a very complex data-heavy product, which is very exciting. And I want to control every detail of the interaction. So I designed wireframes in Miro and then I exported them as images and imported them into v0, the Vercel. Tool, and it's particularly good at creating prototypes out of the image you are uploading. So, it splits it into the right divs, boxes, CSS, everything really quickly within seconds. And then you can write code there with precision because when I was trying to create a very precise, like, layout in Lovable, it just didn't work. It didn't give me that level.

No, exactly, exactly. It's more That's why I like the trajectory we're on, right? Because lovable, it's when you don't even know what that means. Like you're, you're not trying to create a very precise design, but there are some people here who have this very specific vision. And more importantly, they, they're, they underst— they've done this work. They were trained in this work, right? Like they went, they got a degree in this work. And so you have a level of fidelity, high fidelity that an average user wouldn't have. And it actually is what makes it special, right? It's what. Actually brings unique, a unique value proposition to the, what you're building. Cuz there's only so much we can do when we're like, build me a website. You know, I don't know. I, I think I, I use words like make it minimalistic, whatever that means. Um, but I think there's such a high fidelity when you understand these design terms and frameworks and philosophies, some you've defined, but also some that are just kind of classic use cases and design patterns that only an education, you know, in those, that, that area would provide you.

Yeah. And what I'm really enjoying is that in the past, like 15 years ago, when I was doing it manually with teams, we would need to design every single state of every single interface and show how they will connect. And now I write it in V0. I say Well, add scroll item, scroll icon so that we could switch between the months and the data has to move to the next column as we switch between the months. And I just describe it and it does it in less than a minute. And in the past it would be hours and it's actually building it.

It's, um, yeah, as opposed to like, uh, you know, like Figma, right? Where it kind of looked real, but it definitely wasn't real. But now you actually have like the scaffolding code behind it. So it is a, you know, it's productizable from that perspective.

Yeah. And then at some point I export it into Cursor. And then Cursor is creating production-level code. So in Cursor, I add all the final nuances of interactions. Actually, at the moment, I'm experimenting with scraping off the styles from the client website and applying them to Cursor, in Cursor, just by replacing CSS file. And that's it. And that's turning my wireframe into a visually designed site.

Yeah, that's pretty cool. I love it. And so, I think One of the, you know, the, the benefits of hearing from others that are doing this is that what we're talking about when we say vibe coding, the hope is to draw you into the idea that you can just say what you want and something will get built, and that that something will likely be better than anything you've thought of before. Um, certainly it'll be faster than things you've thought of before, but, but that's just the beginning. And that in order to actually build something of high value that is you know, going to actually be valuable in production to generate revenue, save time, that it takes a lot more than that, right? So you're seeing like Joe and Olga, like, you know, there, there's a process, um, and that process is uniquely yours, right? And uniquely ours to define. Um, however, no matter what, that process is infinitely less time than it used to take and requires— now I say this gently, requires less humans to do the work, right? In the past, I would have had to send an email and wait 3 days, and then maybe they got back to me, or I'd have to send another email and be like, did you see my thing? Can you get that back? Like, it is allowing us to be more autonomous as individuals. Um, and in— at, um, at Amazon, Jeff Bezos created this concept, like, I think he called it— was like 2-pizza teams is the term he used. It's actually in Nitin's really cool Amazon like history thing, but—

Yes, it is.

Yeah, I saw it. I was like, what? That's so cool. But these two pizza teams, the intention of it was if you go to the Amazon website, there's like 200 web services all managed by different teams that are loading and deploying simultaneously on that website. So if you look at it, it's obvious there's lots of bars, lots of moving pieces, lots of different— like there's stuff that's devices, stuff that's content. They come from different parts of the company. None of them are— they're not merging all that to a single code base. They are all merging to separate mains that are all publishing simultaneously, which it's a miracle kind of that it doesn't all— at least one part of that page isn't broken all the time. Um, but the reason for that is because number one, of course, everyone has a set, a set of principles that they build to, but the most important part of that is that each group is autonomous. Each group is building against the same set of rules, but they have everyone on the team to get something out the door and published to where the, the outcome published, right? The outcome sent to where it needs to go. Meaning that on that team they have the person who's gonna do the security review, they have the person who knows the legal review, they have the person who understands how to build the UI, they have the person who understands how to build the backend. They understand they have a full deck. They have a full-stack dev person. So these two pizza teams tend to be 8 to 10 pe— people. And the reason they call it two pizza is that he would say you never wanna have more than two pizzas could feed, which again, some of you might have heard of teenage young men. Um, and, and so I was like, oh no, like some of these guys can knock off a whole pizza like at one time. So we, we took it with a grain of salt what he was referring to, but he just meant Like you could order 2 pizzas and everyone would get a slice. So that means 8, maybe 2, right? But 8 to 10 peeps, people on the team max. But each one of those people would have, have to have enough autonomous capability and single-threaded ownership of the thing that they were responsible for so that the process can move all the way to production. And they never had to escalate. They never had to call another team. They weren't reliant on any. Anyone. And this is why I like that role. I was a solution architect and my job was like the glue that held it all together, right? So I had skills across multiple areas so I could just, you know, we called it like SWAT team style. I would just swoop in and fix stuff when it was broken, whether it was in product management, security, um, you know, uh, I've patched code, but I've also built code on Alexa. I was all over the place. I helped them deploy code, but I also helped them build the pipeline. I also helped them write skills for Alexa, which is more like app development. But because of my well-rounded nature, I was like, this is so awesome. 'Cause now I, now I'm useful. 'Cause I like to do all the things. I don't wanna be stuck just doing one thing. But that's how this autonomous structure really suited me very well, which is why I was so successful there. So anyway, I did want to give another, I thought I saw another hand go up around Vibe Coding, but if you'll have to raise it again if you want to say something for me. Okay, great. Don, yes, I thought I saw you. So, and you're welcome to raise it again if you did and put it down. But Don, tell me, and you're a visual storyteller, so you are more on kind of creative side of the house. So I'd love to understand a little bit more about how you're using Vibe Coding and the toolset that you're using.

So I was a bit reluctant at first because I do like the traditional style.

Did you put your hand down? I can't find you.

You.

Wait, you can't hear me?

Yeah, I think— no, I can hear you. You just put your hand down, so it's hard for me. I got you now.

Oh, you want me to raise it back?

No, I got you.

You're good.

That's a weird angle. Okay, anyway, um, hey everybody, welcome to the stage. Yeah, so, um, so yeah, I, I, like Renee, have been using anti-gravity now for about, I don't know, I'd say maybe a month and a half or so. I've been through a few ups and downs with it, but I feel like the real benefit to this tool particularly is just in terms of it's got this recording interface. If you have an idea, you can literally just hit the record button. You don't have to type unless you want to, and you can start to just kick out your own thoughts, your own idea to build the structure. I will say if you can now, in hindsight, I think if you can establish some form of an MD file, I think that gives the machine a bit of framework as to what it is that you want to structure. It'll probably help you a lot. That's the one thing I think I didn't realize at the time was I just, started speaking and I was like, oh, all right, do this. And then it's great. But, you know, the key thing is that— hey, do you guys hear screaming in the background?

Sorry, that's my children.

Oh no, that's mine. That's why I was worried if you—

oh, oh, okay, then no, we can't hear anybody.

Okay, great. Yeah, we got, we got, uh, Xbox.

I'm like, oops, I thought my microphone was better than that.

Yeah, I know, we've got like, uh, you know, Xbox, WWE going on over there.

No, okay, or nothing.

It's madness. So anyways, so, um, and you— all these things you guys can, you know, you can totally search up. But, and it doesn't hurt to have skills. So, um, I don't know, I, I'm, I'm gonna try to— there's a lot of stuff I'll just quickly talk about, but you, you may just want to kind of dig into it. So, um, uh, Sue, Sue, I think, you know, acquiring skills. So basically the idea of skills is are there particular types of skill sets that you might need for building this code? Like in terms of design or testing or, I don't know, particular types of reasoning. Adding particular skill sets help. Then I got a little bit deep as far as setting up the MCP. You can do things like, you might have an idea from, I forgot, was it the chicken or the egg? But you can use Stitch, for example. So Google Stitch is sort of like a nice interface tool that's pretty funky as far as like being able to help you interface, you know, come up with some really interesting design ideas. That would help you maybe create a mobile app per se, or even a really cool web page. In my case, I'm building a visual storytelling tool because I want to be able to have a more active role in how I create stuff and let the machine do— I want to do more of the thinking and let the machine do a little bit more of the prompting piece. That's what I'm using it for. I will say this about Antigravity. The best way to look at Antigravity is it's essentially a container. Within that container, you can use the Claude models, the Gemini models, or you can completely decide, I'm a Claude code person, I just want to use Claude code. You can specifically route it to Claude Code, or— and I think Joe was talking a little bit about Codex. So if you decide like, hey, you know, um, I just want to use, you know, what OpenAI's got, you can still do that within, um, with, within, uh, Antigravity. So, and if you want to get funky, you can even use Open Code if you feel like, hey, I've got a bunch of other models I want to use, you can do that too. So some of this might sound like overwhelming. I mean, I think the best way to look at it is if you, you know, all these things burn tokens. So if you don't want to use it, it's a magic day, right? So if you don't want to, if you've, if you've got a limited budget and you want to keep going, you can easily switch. And I'm not— I'm sure there's probably some cons to going to jumping from piece to piece to piece to kind of like work on your code. But you know, it's better than nothing if you need it. But I think also if you have— I, so I have a Gemini subscription and it gives you a generous amount of tokens to kind of use different models from Gemini Flash to, I don't know, what the heck.

Yeah, um, it was cool. Yeah, Google 3.1.

Yeah, but then you can also use the Claude models within So, you know, it goes, it does go quickly. I think you want to, I don't know, you got to be a little bit conservative, I guess, about how you use the tool. I will say this though, it's painstakingly slow in some cases if you've got like this grandiose idea. And so, and that's, I think that's just the way it is. Like spinning up a website, it's pretty easy. It's actually, it's, like too easy, to be honest. I think I spun, I spun up a website for somebody and it took me like 15, 20 minutes and it was like, wow, this is awesome. Not right.

Well, thank you so much. Yes. Thank you for sharing. Um, I think it's always nice to see the world through different perspectives, different people's use cases. Um, we are coming to the end. I do think it's always nice to get a chance to kind of throw together, but I'm going to show you something first. Hold on. Um, I want to— so I did throw in the chat, uh, the link for Sunday. So you do have to register. That's just so we can send you the replay, uh, of our little live session, but also send you all the links and resources we're going to talk about. We are using Lovable. Lovable, that, uh, block of time, 24 hours, opens at midnight, um, on Saturday night and goes all the way till You know, 11:59 Sunday night. And I, for those of you who have never done this before, I would, you know, carve out a couple hours just to go in and play around without having to worry about exactly what Don was talking about. Um, awesome.

Yay.

Uh, and it's free. Like this is part of being in a cool community that wants to support your learning. The most important thing that I think Vibe Coding offered me was a tool I could use to show people how to go from idea to execution. Using AI as a tool, and it's very impressive, right? So I'm gonna share with you right now, uh, one of the things that today, and give you the thought process we went through today. Uh, so we're on a call prepping for a client. Um, and you can see here we're on a call and we're building out, like, we're consultants, right? So we're going in, we're gonna like build out a spreadsheet to collect all the metrics and like good old-fashioned consultants do. And I was like, Or we could vibe code ourselves a platform that can collect all this information. And now, even 2 years ago, it was good. It, it would've done this then. But what's different now is of course MCP and connectors that I actually in the past would've been able to vibe code a website pretty easily, but then connecting this to real data so that it could do real-time data access, that would've been a little harder because I would've had to actually integrate that. But that's not, the case today. Almost all of the APIs that I'm using to collect this data is coming from an, an API that is available that I can access either via API or via MCP. In many cases, these services are creating MCP endpoints as well. So it's so cool. Let me make it a little bit bigger so you can kind of see it. Um, but the purpose of this, for those of you who took our challenge, some of you were with us for the Agentic Challenge where we built our first AI employee, you can still go through that challenge today. Um, if you're in the I Love AI community, I think it's free. Um, I hope so. Uh, but if you come through the website, if you just happen upon it from a YouTube video, it has a charge to it. So if you're listening to this and you're like, wait, I have to pay, just message us, join the free community and we'll give you access. Um, that's like part of the benefit of being like with us cool people. Uh, but we don't want just everybody, anybody, right? We wanna make sure that you're interested. Excited, and most importantly, looking for a community to grow your AI experience. But this was like the difference in thought process when we're doing a project, right? Is that rather than thinking, going like re— kind of sitting back and relying on the way I did things 2 years ago, build a spreadsheet, build a PowerPoint, create like, create these assets, build a Word doc, create a participant deck, like all the things that I would have spent hours probably doing, if not longer. I now immediately think, let me just try to build out. Now here's where my other, you know, my, my philosophy on minimum remarkable product becomes so, so valuable. And I defined this probably 10, 11 years ago. I delivered, meaning I hosted over 140 hackathons between 2013 and 2015. That's a lot of hackathons. I will have you know, by the way, and what that meant, I was working for Amazon Alexa. My job was to push voice into every hackathon I could. So I was at MIT, um, Hacks, uh, I mean all of 'em. Like I was at Harvard, I was at AT&T Hacks, I was at, in Vegas like too long, too many times that year. But the whole purpose of it was to build something in a really short period of time. And my job was to push Alexa on everybody to be like, and add voice and make sure you could talk to your tech. And of course you just mentioned, Don, how nice it is to just click a button and talk to your tech. Like, that was the outcome. This was 12 years ago we were doing it. But one of the, um, opportunities that, like, we did during our challenge last, I guess last week, week before last, was that we created 6 agents. And now I have sold, right, as a consultant. I'm like, some of these, uh, CEOs don't have time to build those agents, configure the agents, control the agents, manage them every day. So I offer a consulting service. You can actually find out what that looks like., at aibusinessschool.com, not to become a customer, but just you can see how I positioned myself. Some of you might wanna do the same positioning. There's plenty of work out there. Um, but I positioned that, that product set and sold, you know, a bunch of them. And so today we're now in week 2 of some of these clients and I was like, we should have a way of demonstrating value. And instantly I was like, let's build a dashboard and then let's connect each one of these. So if you go in, you could see like some of the details, uh, associated with this. Maybe I'll make it a tiny bit bigger so you can see some of these words. But, but like what, what I wanna do is say, hey, you've, you've used these 6 agents for a week. Here's how your numbers have changed, right? Here's how many more people are engaging with you. Here's how many leads you got. Here's how many sales you got from those leads. Here's how many, um, you know, uh, in this case, newsletter subscribers you have, right? Because part of my service is building out their, not only their brand, through these activities of social media and newsletters and all that, but then also adding my specialty of creating a go-to-market strategy for their business, like changing how they're approaching the market and then measuring how well that's performing. So this is— I just thought it was cool. This is an example of VibeCoding. You can see here, um, this is the prompt: build me a client agent performance dashboard that tracks this stuff, right? Like, there's a little bit more to it. Um, but I just asked, like, build me a dashboard. I gave it all the metrics I was interested. And this was V1. Most of my V1s look like this because I didn't give it any guidance at all on what it should look like. Now, this is in Lovable. I also want to show you something that I vibe coded in Claude. Now, this I was actually asking— remember earlier I was talking about, um, let me go up here— I would like to work on my landing page copy. That's it. Like that. This was testing. I wanted to have a couple examples I could show you live. I wanna work on my landing page copy for a new accelerator. We have accelerator launching in like 23 days. It's actually almost full, but I was like, let me just demonstrate it so you could see an example. This actually looks kind of nice. It caught my vibe. I was like, I'm a golden black person. That's Hamilton style. And I also, you know, the, the fonts are pretty cool. So I'm not mad at it, but I didn't give it any guidance on look and feel. I just let that run. But I gave it the information that's on my website, and I don't know if I have it up here. I must somewhere. Let's see if I can find it to click on it. I don't think I can.

Do do do do do do do do do.

I guess I could just— oh, there it is. I guess I could just search for it. AI startup. Oh, goodness. Where's my keyboard? AI Startup Studios, right? So this is the website and I just grabbed the main 6-week program from here. Um, but I literally was like, can you just make me like make the words better? That's kind of what my question was. And instead it was like, kabam, here's your whole functioning website. And it's not terrible. I was like, this is kind of nice. Maybe I'll swap, maybe I'll swap it out. Um, so this, these are examples of what I'm thinking you wanna play around with over the weekend, right? Like what ideas do you have? What have you already done that maybe you could improve? What are ideas you have that are in some Moleskine notebook that you've just not thought about or had time to build? The, this is the chance you have to just use your words. And Don did cli— you know, uh, provide the very important level of transparency that every word you type. Is it's going to operate in a, a currency called tokens. And those tokens do cost real money and you can go through them very quickly. And so for me, this is a business investment, the, the service levels that I have with all of these, uh, platforms. But for you, if you're just starting, take advantage of the opportunity we have to get access to this platform. Um, and of course get a chance to even maybe monetize something on Sunday. So I wanted to mention that we have a few people with their hands raised. Kim, absolutely. Kim Gaskell is one of our— the builders in our community that is so gracious and shares a lot of what she's building. So we'd love to hear from you, whether it's a question or hopefully your perspective on what you're doing with Vibe Coding. Oh yeah, there you go.

Yeah, let me get unmiked. Sorry about that.

And then I will stop sharing and you can go.

Go for it. Yeah, I just want— you were asking what we were using. You know, I use Pay Me GPT and Chatbot Builder, but I use ChatGPT and Claude every day in just my thinking process and finalizing the code that I need to swap out. Like almost everything that I do now, I think in— you think in dashboards when you, when you do some— I think in agents, like, oh, I could build a an agent for that. And so everything I think about, I'll go build an agent that does that little tool or task. And in Pay Me GPT, I can build the landing page in 30 seconds and I can add an agent to it and it's done. It's there, it's hosted. So I don't have to mess around with all the backend stuff and I love it. And now with the new browser agent, I can just, I have it hooked up to ChatGPT. And so I can go in there and just tell it what kind of an agent I want. It builds my flows and everything for me in the platform.

I love it.

Yeah, it's such a fan. Crazy good. But I've used all the Lovable, Anything, Bolt, Base44, Bolt, and, and every one of them are kind of the same for me. It's that UI is great and it'll give me the framework, the wireframe or whatever term you want to use. And then, but then what do you do with it?

Yeah.

You know? Productize it. Yeah. And so, yeah, the cost can get up there. I actually did a video the other day on my whole process of doing that Knights of AI thing that I did. Yeah. And so I showed what tools I used, the process I went through on each platform that I used AI to help me code certain things or write the prompts or whatever. And then the cost associated with everything that I iterated each time.

So did you post that in our community or is it in your community?

No, I just put it up on my YouTube.

Oh, awesome. Well, throw your YouTube in the chat.

Yeah, I can, I can put it on in the community.

Yeah, either way. Yeah. If you put the YouTube in there, then we can follow you from there too. But yeah, I would love to see it. That's amazing. Amazing.

So I do stuff like that every day with, you know. Yeah, I love it. And I've been here. Do you use skills in Claude? Have you built any? Oh, okay.

Yeah. As a matter of fact, I asked when I first, when Skills first launched, I was just like, if I was going to build a skill, Claude, which one would you have me build? And it's— so I built one like immediately. But Skills, I think I talked about this, maybe you weren't around, but I had talked about Skills having originally come, of course, from Alexa. And then Microsoft adopted it in Azure AI Search like years ago. But it's a great term for AI because it is the concept of reinforcement learning and, uh, transfer learning, right, of capability to a model. So it's pretty cool.

Right.

Um, yeah. So I really like it.

So I'm a big fan. But yeah.

Well, thank you so much. And yes, I think we're offering two more tools. We, and for those of you who don't know, we do have a class. It's called, um, Agentic AI Practitioner. It does lead to a certification at the AI Leadership Institute. It's part of our Essentials program. And we go through these tools and build them. And I will tell you, we've had mixed reviews on like how hard it was, but I think it's good to do hard things. And I think the best way to do hard things is with people who have your back. And so is it hard? Yes. Um, is it good to do hard things sometimes? Yes. Is this the one hard thing you're gonna wanna do? Maybe not. But if you're like wanting to be part of this, like just know there are kind of safe spaces where you can learn. It's difficult. Your brain will hurt. But I've almost come to like, I think, Kim, you're the same way. Like, we kind of like, yeah, we get that sensation of like bursting at the seams and we're like, this is where it's at.

And, and guys, I don't code. I am not a coder. I'm not an engineer. I'm not. So I learned all of this by doing these classes and building things that didn't take any code, but in doing that, I'm really getting pretty good with code now and understanding different things that I can do and just replace certain things. And it's because I've just eased into it and I just keep asking the questions of these LLMs. They tell me, oh, you need to do this. And I'm a 5-step start person. That's my framework I use on everything. And so I just ask the 5 questions and it helps me through.

So Amazing.

I love it.

And like, yes, like Olelo said, uh, you're very inspiring. It's so cool. I'm so glad you chose to, uh, thank you very much, journey, because it's very cool. And new tools. So all of these tools, like, that's why it's good to take advantage of these free tutorial, like the free trials, and just test it out, because some tools are just kind of made the way you think because the developers who built them thought like you, and you won't know it until you try it. And that's my hope that this session we're doing today will just inspire you to go try some tools out and figure out, you know, which ones meet your needs. So Sandra, good to see you. Would you mind coming off mute and saying hello? Either maybe share what you're doing, is that what I hope we are going to hear, or you can ask a question either way.

Yeah, I'm excited about the Lovable Marathon because I'm in the process of building a website, but I'm a creative person and I don't want to have like a generic thing that it's spitting out something that I don't like. It's not in my— aligned with my artistic skills. So the question that I have is, before this logo marathon starts, what are the things that I need to get from AI to structure for me the steps so I'm ready to build it instead of asking for things and then changing them on the fly? I just want to go with 99% of an idea of what I want to do.

Oh my gosh, this is a great point. And something that I probably talk loosely about, but Olga had mentioned her process, which is I do, and Joe even, right? There's kind of a mental process of defining what you want, right? This problem identification and, and scope definition, like what are you trying to do? And then making sure that you're not trying to do too much, right? Minimum remarkable product, something small. And then describing it from a business function and, uh, business requirements perspective. Like, what's the outcome? How's, how do you know if it's done? How do you know if it's working well? Like, really digging into the thing you're building and how you know it will be successful. And then you can even use, which I use ChatGPT to help you, or Claude or some other tool to help you think through, like, Socratically, right? Like, what am I not thinking about that will help me? And use that as a, a scoping mechanism, right? When you talk to an LLM, what am I not thinking about when I'm building this that I, that will help me as I use a Vibe Coding platform? And it'll start helping you craft. And, and again, you have some very specific guidance that you wanna provide when it comes to design or creativity or how the model renders content. So you can provide that and even ask, uh, a model like an LLM in preparation, What's the best way that I could make sure that I give this instruction set to a machine and it understands it? And it'll reformat it, right, for you to ensure that you've put things— because LLMs are funny about the order in which you say what you say and the way that you say it. So really partnering with an AI. But, but I always say build out like the business requirements, which is more on how do you know that you did a good thing and that it's working and it accomplishes the task. And then the functional requirements, which is what Joe was talking about, where you actually say, what does each thing need to do and look like and document that. I use GitHub, so I put all of my documentation in GitHub, but you could use Word, Notion, like anywhere, but have it written down. The benefit of that is that because we're in a world of prompt engineering, everything that you document, you can now, it creates a bit of portability for you, right? So let's say you build it on Lovable, which is usually I will go into Lovable or I will go into my LLM of choice. And say, now that I've talked to you about this, can you write me a lovable prompt?

Right?

Cuz it knows what lovable is. So I will even ask it, and that's how I do get to like a 90% solution cuz I'm taking all of this thought work and partnering work and then I'm having it, um, build a lovable prompt based on that conversation. And that will get you much further so that when, yeah, when you, when the gates open, you are starting with something that's much further along and then you can focus on those areas that you really wanna narrow, right? The instruction fine-tuning, as we call it, where you want to narrow its behavior.

Yeah.

Would, would you say that ChatGPT is better at that conversational stage where it's extracting the ideas from my head instead of going to Claude? Because Claude to me looks more structured.

Yeah, I think, yes, I would say I always use ChatGPT to write my lovable prompts. Judge me if you will. But that's what I use and it's very hard for me. I'm in a very tough spot right now. But, but yeah, that's what it— that's exactly what I use. And only because I've been using ChatGPT for like 4 and a half years. It knows when I, when I ask questions, it understands the underlying context of my questions. I have, you know, just a huge, huge amount of tokens invested in ChatGPT. So as a result, my answers are better. I also go to Claude and it just, it doesn't do, and I do both, right? So one of the things that I do do is I do use, uh, a model aggregator to test out MRP prompt styles. So, um, I use Poe.com. It's very, I, it's, I think I paid $20 a month. Um, oopsie. Uh, but maybe Sherilyn, you could throw that in the chat for me. But Poe.com gives you access to like 20 major models and you could just throw that prompt that you, you know, the, the result into 6 or 7 models and just see what it comes up with. And that'll give you a feel for maybe it's, you know, maybe using even Gemini. Gemini in the last 2 releases has gotten very good. My habit isn't there, so it's a, you know, I have to like change the way I think to go over there. But every time I go over there, I'm like, oh, this is pretty good, but it's not so good that I'm like gonna change my habits. But it, those are the things I would do to kind of test things out.

Uh, can I ask you a final question? Once we are done with the Lovable build, um, how can I extract whatever Lovable created and put it into my Claude space? Because I have the co-work and everything in there.

Yes, great, great question. So I'm going to share my screen once again, go to Lovable, and remind you, um, so when you are using Lovable— maybe I will make my screen a little bit bigger. Oh my gosh, never mind. Oh, Oh yeah, there we go. That's the way I wanna go. So notice way up here there's a little Git Kitty. So this is the OctoKitty from GitHub. When you click on this, this will— so if you don't have, we do have this class inside the community. Um, if you don't have a GitHub repo or a GitHub account, create a GitHub account. It, it's free to a degree, like anything else. If you wanna create private repos and build a business on it, obviously you'll probably have to pay for it. But I think I pay $4 for enterprise, so. A month. So it's, it's economical, but if you're not looking to take on another fee, GitHub, when you connect it to GitHub, it's gonna take all of the code you built and sync it to GitHub. So now every, like all the assets that you've developed are now there, and now you can point any, any LLM to that.

Yeah.

Or any other system. Some people actually choose to build from there and then use GitHub Copilot or another VS Code, extension to help them actually manage the code from that perspective. Like they use this to get kind of to a production prototype, move it to GitHub, and then they build from GitHub. That's one option, but this is great, uh, to take advantage of for sure. And I will be doing that tomorrow in our, our session is that we'll play around for like an hour and a half. And at the end I will demonstrate creating a GitHub account, um, and then connecting it to your Lovable account. So you can move that code, uh, to GitHub. And I would do that anyway just because it's good to be able to step through that code without being tied to an L—

an— a—

a live coding provider. You never know what can happen. Yeah, great question, Sandra. Thank you for asking. That was a great question. All right, awesome. So we've had some fun today. We actually decided to, uh, take, um, some amazing, uh, conversations. I like doing it this way, but I know I always want us to meet and greet in person. But we, we are at time, so I hope you enjoyed our session today. Let me remove the spotlight so Sandra's not like in the spotlight the whole time. But I hope you enjoyed the, the overview of vibe coding, the reason why it's useful, my experience in this space, and how very far back this concept goes, and how much we really are trying to amplify the ingenuity of humans. Right? Like Vibe Coding is meant to allow people who could never have gotten their ideas to market the ability to do that. And I hope you see the huge opportunity, huge, incredible opportunity that we have on Sunday to just play around with this tool. I always say like, if, if you know someone who's kind of got their hands crossed about AI and is like, I'm not sure about this, this is a great chance to be like, let's go build something. Just, and if you want, create a little build party, right? Like get them on a Zoom. You don't have to join mine. I'm, I'm actually going to a live She Builds Lovable party at 9 in the morning or something on Sunday. I'll be very tired, but I'm going because there's something to doing it together. That's why I love AI. Our community is so important. And of course, AI Leadership Institute, so important because it creates a space for us to ask questions, do so without fear of asking something silly or feeling embarrassed. Because we're all on the journey, right? We're all learning this. And the more people we can bring along, the better. There is a digital divide happening as we speak, as people decide, you know, whether they're gonna use AI or not use AI. But there is an opportunity to say, oh my gosh, but what if that nonprofit that you've been dreaming about for 10 years, what if that's a Vibe Coding session away? Like that, that's pretty magical. And so I want people to at least know what is possible. And of course, what we can do safely, responsibly, and securely. So if you are not part of our community, head on over to noelle.ai. Very simple. It's a very simple website. It'll show you the opportunities to work with us. One is our free community, I Love AI. I hope you will join us. The other is AI Leadership Institute. That's where we do all of our classes, all of our certifications. So if you're looking to take this a little bit further or a lot further, get hands-on experience and work with me live. There is an opportunity for you to do that. And most excitedly, we are launching the Business Accelerator, which is for small business owners. So if you are looking for a way to get your first 10 clients, AI enabled, AI powered, that's what we're doing. If you have questions about that, just reach out to me on LinkedIn or YouTube or wherever you're watching this from. Um, and I'll give you the details. But, uh, but it's gonna be fun. We're about to build some crazy businesses, generate some massive value, and most importantly, get our ideas off the pages of our mile— you know, I don't know if you've seen, I have like hundreds of Moleskine notebooks, hundreds of them full of ideas, and now's the time. So, so get your idea into the world by vibe coding just a little bit and see where it takes you. Thanks for being here, everybody. I hope you enjoyed this session, and I hope I see you on Sunday. Otherwise, I'll see you all next Friday. Thank you, everybody.

Bye.

Hi, hi, Noelle.

Um, I left you a little message there. Sorry.

No problem, Gabriella.

Uh, okay.

Thank you very much. I will look at it. Bye everybody. Amazing. And don't forget, the notes will be sent to you, so we have all these amazing links. They're going to be in the replay section, so you'll get to see them all. It's amazing.

Thanks, Noelle.

Wonderful session.

Good to see you, Nitin. Thank you so much.

Thank you.

All right, talk to you soon. Bye everybody. I'm gonna save the chat so that I can read all this stuff. There's a lot of chat in here. Y'all are so awesome. Okay, save the chat.

There we go.

All right, awesome. See you later. Are you good, Joe?

Okay, awesome.

See you guys. Bye.

Oh, gotta end with some fireworks. I should have done that way earlier.

Bye, Joe.

See you later.