Is vibe coding a bubble or skill Issue? Tactics to actually ship usable products

Speaker A: There's this whole news cycle that's happening right now that's basically saying that vibe coding is a bubble, the MRR isn't real. And I have to be honest with you, up until today I kind of bought into this narrative. But today I'm having on my friend Jacob Klug, who runs an agency specializing in lovable MVP development. These guys are super overpowered. They're building like full platform level applications in the span of like two week sprints. And he's arguing that all these people that are saying this is a bubble is actually a skill issue. And he explains in this episode the specific form formula that they use to create beautiful software that's extremely functional, that has high complexity using these no code tools. If you're new here, I'm Cody Schneider. I've been building online companies for the last 15 years. Go subscribe to this channel right now if you're on YouTube or subscribe on Apple podcasts or Spotify. It's how we keep the lights on. It's the easiest way that you can open a conversation with somebody that you're trying to get into a dialogue with. And today's episode is brought to you by graft.com graph.com is my company. It's an AI data analyst software that allows for you to build live customers, custom dashboards that are interactive and there's zero learning curve to get started. You can create these dashboards in minutes. Just connect any of your data sources, whether That's Google Analytics, four Facebook ads, HubSpot, Google Ads search console amplitude and then you can just chat with the application to build a ah dashboard you can say build me a stack bar chart showing new users versus all users over time and it will make that chart from your Google Analytics data source without you having to learn how to use the software whatsoever. You don't have to spend hours learning looker studio tableau, you can just jump right into this. And we also function as a data transfer and data storage company along with the business intelligence component. And for the more technical people out there, we're an ETL and a data warehouse and the business intelligence tool we're helping founders, e commerce companies, marketing agencies, marketing teams, hr, ah departments and recently financed through data analytics about whatever their source data is. And today I'm giving you a 14 day free trial that comes with 10 seats for your team. Just go to graph.com by using the link below and sign up for free for that trial. All right. Alright, let's get started with today's show. Jacob dude, so stoked to have you on, man. I've been looking forward to this all week.

Speaker B: Thank you for having me. I'm excited.

Speaker A: Of course, man. Of course. Um, so I feel like you are probably see just like a ton of these apps being built, um, using these no code tools. Like lovable is what you guys have deep expertise on. I'm just curious to download like, how are these most effectively used? Like what, when should they be used and when shouldn't they be? Like, from my experience, if I'm trying to build a full fledged platform with like multiple features from a scope perspective, it's not really the right way to do that. But if I need like you know, a personal tool for myself or like one, you know, almost like a tool that has one killer feature that I launch with, um, those, it seems like those are the kind of the best ways to go about it. But I'm just curious to see like what you guys see with the clients that you're working with and like, how can, you know, the people that are listening use these tools most effectively? And like, what should they, what should their frame of mind be when they're coming into them?

Speaker B: Yeah, no, appreciate it. I, um, think a lot of people view these tools, um, as kind of like these like, you know, vibe coding kind of got a bad rep when it first started. Um, it was kind of viewed as like this like very, you know, janky, um, you know, amateur way of building software. And I think that's true for any new technologies, uh, coming from like the no code bubble days of things. We have the same kind of trend where um, early days in no code you saw like really janky products being built. And then as the technology got better, as the people and kind of uh, more people educated themselves and became familiar with best practices, the tools, um, and the products that would, would come out of them just got a lot better. Um, so I think we're probably going to see a similar thing. I actually don't think it's necessarily um, like a tool issue. I don't think it's like lovable isn't to the point where you're able to build these products. I think it's just a skill issue. Um, and I think that what you have right now is a lot of people that are not devs trying to be devs. Um, now lovable has. And all these AI vibe coding tools, repla, et cetera, have made the bar significantly lower. Right? I mean like, if no code brought to like 50%, we're now at like, you know, 85% you don't of like taking the whole process. There's still that 15, 10%, um, where like maybe when you're thinking about enterprise or thinking about more sophisticated applications where you kind of do need a little bit of technical experience, um, of like just kind of getting that, getting it over the hump. I'm sure we all are familiar of like, you know that the error loop. Right. So I think that um, we technically have the capabilities to build pretty much everything. It's just a matter of, you know, of your individual skills.

Speaker A: Totally. So I mean I find this really to be the most impactful thing with all this is like come in with the plan, you know, have a PRD that you've built. Right. Um, and then understand, you know, basically like the language that's necessary to describe the thing before you even start the build process. Um, I'm curious, like what, what does your guys stack look like to actually get these out the door? So say like for example, you're uh, going to go and build, you know, an internal tool or an MVP for a client. How are you scoping that? Like what, what does that look like? Um, and then are you pulling it out of Lovable into like a cursor or cloud code or something else to kind of do like the final, um, um, you know, backend setup, et cetera, or are you doing this entirely on top of this platform?

Speaker B: Yeah, it's a good question. So I would say, you know, a lot of the planning does start within like a ChatGPT or just any AI tool. Um, I think it's really important that you, you really think through the product you're trying to build. I think, you know, it's kind of like uh, the expression of like when you have infinite choice and infinite like kind of ways to build, um, you can kind of get really scatterbrained. And so that's one of the things that we really make sure is that we clearly define what we're building first. And ChatGPT does a really good job of just kind of like prompting you and asking you questions to kind of help build that prd. Um, once you have that, um, it makes it a lot easier to kind of go into Lovable and build, uh, your first prompt. Usually you can kind of can still down something a little bit more, um, you know, shorter and, and yeah, we build, I would say it depends on the application. We try to keep it in lovable as much as possible because lovable does have like such great infrastructure behind it now. They just came up with Lovable cloud. I'm not sure if you saw that yesterday.

Speaker A: Yeah, I did. Yeah.

Speaker B: Yeah. Which means that everything is now on lovable. So you don't even need to leave lovable to do hosting or database or anything, um, or analytics even. So, you know, with that infrastructure, if you can utilize it more, you just get more of the benefits of it. We do occasionally have to move it over to cursor though, um, to maybe do, maybe debug something or add in some integration. That's just being a pain.

Speaker A: Yeah, that makes a lot of sense. So I'm curious, like, what, what are people building, like, you know, for the marketers that are listening to this, like, are there any, any things that have you come across your desk where you're like, oh, this is fascinating, this is the way that people should be doing this. A lot of what I've seen has been like, here's a personal workflow. I go and I build personal software for myself. But I'm just curious to kind of understand like how like, even if it's your own side projects or stuff that you've seen, like how can people, like just to give people a point of relativity of like what they can do with this and like how they should be, you know, how they can use this to, to automate their, their, their day to day processes.

Speaker B: Yeah, I mean I don't know if there's been like a huge difference in like the type of products for being being built, but I do think that there is becoming a trend of um, kind of what you said, like these like people almost building their own products over using an existing SaaS. I, um, think we, we just ran a cohort of um, 20 people that are all kind of building their own product and almost all of them are going after their own kind of problem set. Um, and a lot of them are, you know, maybe have been paying for an existing solution, but one they kind of want the skills that come with building on lovable. And then they also just want like to be able to have their own tool. And so what I think what we're probably going to see is this idea of like a personal os, A uh, personal like operating system. Right. Where people build maybe individual tools or maybe like one big tool where they manage their entire life through it. Right. And it's like a series of different applications. I think it's going to be really interesting. Um, and potentially, you know, I'm not sure, I'm not going to make a prediction, but it could start to rival traditional SaaS in the sense that like, if somebody just wants to build the same thing in a weekend, there's also that like, feeling of pride of like, oh, I own this. Right. And so there's almost like a little bit of a ego to it as well.

Speaker A: Totally. Yeah. I think it's kind of interesting that I, I, I have a friend who's been just going down as like P and L of his agency and like looking at their biggest costs and being like, okay, engineering team, like go and basically build our version of this, right, to like cut the costs, um, so that they can increase margin and it's cheaper for them to do that and have you know, two offshore devs that they have full time, um, working for them rather than like, you know, using these tools, um, or using the software that they're using. And so it's kind of this build versus buy is suddenly becoming like way more. I mean you're seeing this too in like these like very traditional categories, like DocuSign as an example, where it's like you can build basically a document signing application in like a couple of hours now, right, um, place, um, you know, whatever the cost is if you have a team of 100 people all using the DocuSign software. But I, I'm curious like for you, like with, you know, I imagine like you have your client work, but you're also just building stuff for yourself. Like what are the things that you've like done recently on like your personal builds where you're like, oh, I didn't like it. Just kind of like shocked you that you were capable of doing it, I guess, like using these no code tools or is there anything that you've built that's kind of been like, you know, I guess fascinating, uh, or like shocking in a way.

Speaker B: Yeah, I think one of them, I just posted a YouTube video of it. Maybe we can link it. I just built a, um, I think one of the, just the most interesting industries that I really am optimistic for for AI is healthcare. Um, and I think one of the lowest, um, I guess friction things when it comes to someone's healthcare is taking preventative measures, whether that's supplementing or lifestyle things. And I think the best way. It's so interesting that um, m. We may cut this maybe off topic, but I think it's so interesting how we measure every single thing about everything. About business, about the weather, everything is measured, except for we don't take the time to measure our own bodies really. I mean now it's starting to become more of a trend. But this idea of only Doing blood work when you're sick is kind of like, it's not very preventative, right? And so I built this tool that essentially, um, takes PDF, uh, like if you upload a PDF of your blood work and then uses OpenAI to analyze it and it will basically track your blood over time. And so, you know, if you upload it in January, then it will keep, it'll keep giving you charts and it will give you suggestions on how to improve certain things. And I think that it's just like, it sounds stupid to maybe some people and people may think that I'm like overstepping, but you really should know like some of the basic foundations of your, of your blood work. Like if someone's like, you know, what's your, you know, glucose, sugars or whatever, like, you should be able to be like, oh, like this is kind of the range I've fallen. I don't think that's uh, a crazy thing. And to have goals to like improve those things. These are very manageable things. I mean, a lot of this comes down to like an extra pill that you take a day. Um, and so I think that's like a really low friction way of like, simple tool, um, that I just built for myself. I'm not going to release it because I'm sure there's a lot of compliance stuff with it, but.

Speaker A: Totally, yeah. I think that this is interesting though, this idea of like, you know, almost like I have another friend who built like a personal CRM, right, where he views it as like I'm pulling in all my WhatsApp data, my Telegram data, all of my, you know, contact data, all my iMessage data, all my email data, just so like, I can have kind of like a holistic view of like everybody that I, he, he works in. He's like a talent manager, right? And so it's just like talking to everybody, everywhere is kind of the problem that he's facing. And it's, it's really interesting to see these solutions like the one that you described, where it's like, here's this personal thing that I'm having trouble with, right? Like, how do I go, you know, how do I go and solve this? I'm curious. Like, you know, so you, you find this problem for yourself. What's your first step on kind of building that out, like again for the audience? That's like, I want to build, you know, custom software solutions for whatever it is that I'm working on. Like one I built recently for myself was like, basically it does a, you know, research on all of our, um, uh, you know, every person that basically schedules a demo and then gives me like a synopsis right before the meeting. And it's just like literally an audio 30 seconds that tells me what I should know about them before going into the call. Right. And I listen to it right before I jump into the sales call. But I, like, you know, if you wanted to start out building, you know, this, um, you know, for whatever problem you're facing, like, where should they begin with this and like, should they just go straight into building the feature? Should I think about like, you know, um, doing the auth first and also uh, like kind of building out the, the database and then starting the features or like, is there any tactical like sequence of events that they should focus on to actually build these applications for themselves?

Speaker B: Yeah. So I think first thing is again, just having a real kind of prd, uh, or some kind of scope on paper is just really helpful. It gives you an idea of what the end goal is. Um, it also gives the AI context on really what you're trying to build. Um, as far as order, um, we've played around with different kind of orders. Um, I would say it's still very experimental as far as the right one. But I would say overall what we've seen is that building kind of the infrastructure and the base functionality and some of the biggest kind of, um, hurdles first is usually a good, good thing to do because, um, you get, you get some of the, yeah, the foundation out of the way. And so things like auth, things like your database set up now with lovable cloud, I mean it just makes it so much more seamless. So like you're not even needing to really think about it. You just basically say, hey, just like connect the front end with the back end now and it will just do it. So even that is starting to like, to decimate. But I would also take it like one thing at a time. I find that in general a good rule of thumb for AI. And this goes for ChatGPT, this goes for Claude, this goes for everything. Um, I'm curious, your experience in this too is I actually feel that there are, there's a pretty clear correlation to the longer uh, the prompt or longer the context you give it, the worse it does.

Speaker A: Um, yeah, I find that breaking it up is like the only way to make it work and like give it like breathing room in between. Like prompt chaining is the most effective way to do everything. Right. Um, um, if you're writing content, that's the best way to do it. Um, if you're you know, trying to create AI videos or you know, avatar videos, like that's the best way to like it just whatever, whatever it is, breaking it into those components and having it like work through each of those, um, is what I find like getting at the one shot things is way less likely from an alcohol, you know, a positive outcome standpoint.

Speaker B: So yeah, I think you kind of got it. You got to let the AI like you gotta give it some, some room to kind of breathe a little bit. Uh, it's kind of the same way you talk to a human. Right. It's like would you give your, someone that works for you like the most rigorous, like do this, don't do that. Like at a certain point it's just gonna, it's gonna feel like a robot. Right. And I know AI is a robot, but in fact if you treat it more like a human, I think it will act more as a human. Um, and so I think that's kind of true across the board. It's true. As you said for content writing, like if I were to try to write a Twitter post with AI and give it like a long thing, it will do way worse than if I just am like improve this and then I give it some sample and uh, it will do it really well.

Speaker A: Totally. Yeah. I think examples, providing that is the biggest thing that I'm finding to be the unlocks is giving it data of examples of high performers. Right. So say for example like we own a bunch of email newsletters. Like that's one of the businesses that, that we're, we have in the portfolio. And you know, if I just say like write a good subject line. Yeah, right. Like about this, whatever the topic is, um, that it's, it's like you know, going to be a pretty terrible outcome. Um, but in contrast I'm like Hey, here's 10 subject lines that got um, you know, our best open rates. Like write a subject line uh, for this content based off of the one that's provided. Right?

Speaker B: Yeah.

Speaker A: And the outcomes that you're going to get is just like way higher quality. And I think that, that you know, and it's in its own like right. There is an example of like oh, that's an automation or that's a software that you can basically build. Like imagine you're plugged into whatever your email service provider is. You're constantly downloading, you know, the, the, the um, campaigns that you're running with the subject lines, analyzing that data and then basically like you know, building it so that it's for any new email that you're sending, like you could have it just like you know, one shot. Basically write that based off of the data that's live, you know, streaming in the background. And these are, these are how I'm seeing these tool used like more and more is like what are these internal workflows I'm doing, How do I go and basically you know, build out almost my own operating system for, for how I approach whatever it is that I'm, I'm going about. So um, I'm, I'm m really curious like what again I know you guys see a ton of products. What have you seen be successful and like what has not been successful like based on this stuff, like is there categories that are, you know, you're kind of getting like more ah, um, you know, more success in uh, is there uh, you know, specific like you know, product scope? Like oh I have one, this has one features or it's a full platform. I'm just curious if there's like any kind of high level insight that you have just based off of like the amount of products that you guys like push through um, the agency.

Speaker B: Yeah, I mean majority of the stuff that we do are like full blown ah, like startup products. Right. So think like solo founder or early stage team is looking to build essentially their entire platform, um, with us now. I honestly like the thing is I don't think that AI has fundamentally changed. I mean obviously we've seen more maybe like AI kind of products but that's more like, you know, like just playing into the whole AI thing. I think one thing that I have noticed is that the ideas are getting worse. As in like I think that the barrier to entry is so easy now that that is one of the things that you need to be cautious of when you're building with AI is that it's almost so easy that there's almost no downside. But you know like obviously if you're doing that every weekend for you know, a year straight, you're not going to also see much upside. So um, I think it's more about picking the right ideas and still going through I think that like thoughtful process, especially if you're going to work with like an agency, um, to build something. Um, we're definitely seeing some more internal stuff and I've been surprised to see what kind of enterprises are starting to utilize it. Um, I can't really say names but I think really big companies are starting to realize um, how powerful this is. Especially when you um, you Equip them, uh, with the existing team. I, um, think that, you know, you give this to a designer to kind of whip up a prototype or something and you just get a much better result than what you used to be able to get, um, with, you know, Figma and et cetera.

Speaker A: Yeah, I think that's the thing that I'm realizing with a lot of this too is like the domain expertise is still the most valuable aspect of this, right? Like it's just a tool within the tool belt of a person that knows what they're doing. It does make the, you know, the barriers to entry, um, you know, lower for uh, an entry level person or first time, you know, builder, etc. Um, but the real kind of there's like a power law with this, right, where it's like if you again really know what you're doing and can deploy these tools, you can do so much more than was previously. I see this with AI automation in particular for marketing. If you know how to do marketing, I mean understand how to do AI automation. You basically just turn yourself from one person into a hundred people, right? From the content output that you can basically produce and just the level of just like, you know, distribution that you can create. I think that same thing is happening where it's like, oh, if you're a designer or a product person or um, you know, just like somebody that's running operations internally, you have this very specific problem that you're facing. You can go and build this software for yourself that basically solves that solution for, for the people that are trying to learn. Okay, I need, I want to use this, I need to use this. I know that this is important for like my career. I mean I just saw McKinsey laid off like eight, you know, 11,000 people or something insane that like can't adopt AI. That's basically what they said is the reason that they're like sunsetting them. Um, how, how would you like to tell that person to like go and learn this? Like is there you know, um, like courses, is there YouTube channels, like who's doing the best content around this? Or is it just like just start building with it and put an hour aside a day and build something every day and that's actually how you're going to get, you know, the most effective with it. What, what, what would you say to those people that are trying to figure that out?

Speaker B: Yeah, I think um, I think building is definitely probably the best bet. I think that um, you obviously, you learn as you go. I would say try to understand Some of the, like the base principles of what goes into building. So understand what the differences of front end and the back end understand, you know, maybe basic security measures, what role of security is. You know, how do you, how to utilize Lovable's like security features. I would say Lovable has great um, docs and kind of like community uh, as well around all this stuff. So like there's, it's really well documented. I believe if you go to like docs.lovable.dev or something like that, um, I'm sure we can. Yeah. Uh, they have really great guides on how to get started, features, etc. Um, you know, people like myself, people like, you know, Riley Brown or some other big creators. I'm sure you've done some as well. Greg Eisenberg, um, are kind of always showing up.

Speaker A: I'm not even close to you guys in this category. I do nothing in this, in this.

Speaker B: Well, you associate yourself well with the people that do. Um, but yeah, I mean, um, I don't think there's any one thing that I've noticed from kind of helping people less um, on the agency side, more on the cohort side is how the, the like the availability to it almost freaks people out. As in like people know that they can do anything and it's like this like huge like new opportunity to build things and they just do nothing because they get, become so overwhelmed. Um, like it was just so surprising to me. We were doing, we're doing like this like four week bootcamp where we help founders. And um, you know, week one I was like, so what percentage of you guys, you know, this is, you know, they paid good money for this. And I was like, what percentage of you guys have uh, built something with Lovable? And I'm expecting everyone to be like, yeah, of course. Like, you know, we've tried a lot. I would say 50% of people, like, no, not, not a single prom. I was like, wow. Um, and so I feel like, um, you know, this is kind of why agencies still exist. I get a lot of hate for people like ripping on, on me for running a loveable agency, which I get. I get the. No, I get the criticism. But you also have to understand, uh, that any new technology, um, first off, there are things you can do to obviously become better at it. Um, AI is not going to replace agencies. It's probably going to make them better, um, and more in demand because at the end of the day people are lazy and uh, they want things done for them and they want the insurance and the Comfort that comes with working with experts. But, um, anyways, long story short, I was so, so surprised by how few people just, just built something. Um, so as, as, as much as it is, like, you know, you don't want to waste your time, like, prompting away at different ideas, having one or two side projects you just work on for just the sake of just keeping up to date on the. On what's new. I think is super beneficial.

Speaker A: Totally. Yeah. It's funny, I've been finding myself just like, keeping the windows open and uh, like, bouncing back and forth between, like, stuff I'm working on and then letting the agent work on something that I'm asking it, like, in the applicant. It's kind of a. I mean, it's really. It is. It's a crazy world to be in. What I've been trying to do is like, whatever it is I'm working on currently, how can I automate that with software or like some type of AI automation? Um, so, like, while I'm doing it and like in, you know, basically in a waiting period, say, for example, I'm like, cleaning an email list. I'm like, over in the side, like, trying to figure out, okay, can I just automate the entire process or the thing that I'm doing and then hand that off to myself? Or that's just like, handled by software. So, um, yeah, I want to circle back around to like, the, the, the pro, you know, the products that you're seeing. This work. I talked to this kid, it would have been a couple weeks ago, and he, he basically built. It was like a tool in this random category that was pulling, like, public, like, weather data. Right. And then providing it, um, to, um, like, you know, an industry that he was in. I just, I just don't want to, you know, kind of, kind of blow up his spot. But I, I thought it really was fascinating, like, you know, the, the industry expertise and then the ability to like, build that custom thing. It feels like that is what's happening right now, where it's like, oh, you used to have to build, you know, software for these huge industry categories because that was the only way that you could make this work from, like, a financial standpoint. But, like, now it's like, oh, I can find, like, this very niche industry with this very specific problem and kind of build that solution. Have you seen anything like that from the companies that you've worked with or the products that you guys have built? And, uh, any insight there?

Speaker B: Yeah, yeah, definitely. We've seen some big. I, um, mean, even in the bubble days. The thing about everything going on with lovable and AI is to me it's a completely history repeating itself. Um, maybe this is bigger than um, no code and stuff so more people are talking about it. But like all this stuff is literally like going exactly the same. Um, and the exact.

Speaker A: That's so funny. You're just watching the same cycle happen, literally.

Speaker B: I mean, I would say the cycle that we're in right now is kind of like it's. We're definitely still early. Um, I think a good way to gauge how early we are is the quality of the products that are being built with the tools and the design of them. Right. Once you start seeing like really well polished design, like lovable apps, and that's really what we're trying to like. I think that's probably the hardest part, uh, of really becoming good at vibe coding is that vibe coding is great at uh, doing kind of the heavy lifting and it helps you generalize what you're trying to go for. But the second you want to be really specific and you know exactly what you want, but you just need it to work or look a certain way, that's where it becomes a bit trickier. Um, and so that's why you have developers like us where we can come and kind of just do that. But, um, but yeah, to answer the question around creators, um, you know, even, yeah, we, we've been working with creators for a while. I think it's probably um, the biggest like kind of collab of like, you know, content creators meet software now especially, it's more powerful because, you know, these creators can get kind of full ownership of the products. They're no longer needing to raise money because the cost has just gone so, so significantly down. Um, I do think though we are probably like kind of again, what happened with Bubble and no code is like, I think the pendulum will swing to a point where like every creator is creating software. Um, same with like every. I think every person is going to try to automate every single aspect of their business in life and I think eventually we will reach a point of like, okay, this is like too much. Like people will be sick of thinking uh, about automations and I think people will revert back to like kind of OG way of doing it and then again it'll swing back. It's, it's always, always is. But I think we're still, we still have like a good year of like year or two of people building different softwares, whether for themselves, whether it's a creator kind of collab xyz?

Speaker A: No, it's super interesting. Um, um, on the creator side, I have a friend that's doing something similar and they basically like find a creator and then they pitch them on. You know, here's the software that we think your audience could find valuable or like partner with them or like hey, what is the biggest pain that your audience has? Like can we you know, basically build a software? How do you, how do you structure just to learn from you? Like how do you structuring those relationships? You guys doing a rev share set up? Is it a more of a um, like you just you know basically build the product, hand it off to them and then you uh, know maintain it for them or what does that look like? Or is it different at each of the deals?

Speaker B: Yeah, we've done some like equity stuff in the past. I would say I try to stay away from them now. I think that it's a very different business. If you want to go into that business, I think go for it. But you know it's one that relies on, it's a long term business. Right. It's like you're running like a private equity fund essentially. Right. You're, you're betting on the three to five years rather than the next three to six months. And so it's really good on paper

Speaker A: and in practice it's always like very way, you know, 100x harder than actually what you think it's going to be. So.

Speaker B: Exactly. Yeah. So um, we just kind of try to keep it simple now and we just do like kind of engagements where you work with them on a retainer basis or something like that. Um, find it's, it's a win, win. And now here's the thing, like I think going back to saying like you're going to have some creators, it's the same with anything, right. Like it's like prime, uh, like Logan, Paul and KSI was like the epitome of like a creator, um, brand. Right. Like I think one of the most well executed probably plays but you have like hundreds of other uh, brands that were started by them and others that you've just never heard of again. And so it's going to be the same with software. The barrier to entry is going to be so easy that you're going to see all these creators spinning up software and I think like execution is going to matter a lot.

Speaker A: Totally. Yeah. I think it's, it's you know, kind of the dots I'm connecting with on this is like what we're seeing universally with everything in AI, like there's the, the level of slop that is coming out through the pipeline is so much larger, right? Like whether it's short form video. I just saw it. Um, I just got shown, uh, a media company that's basically doing um, like YouTube content that's like no, you know, basically like, um, uh, like no, um, whatever it's called, like no face or where they, they, they, you know, it's not an actual person. It's just a voiceover, right, with like edits in the background. Um, but you know, they're basically dropping like it's something insane. I want to say it's like a thousand videos a week right across all of these different channels. And I think that that is like what we're seeing, you know, kind of universally like in any of these industries. So it's really interesting though because like at the end of the day, like those products can work. It's kind of what you've been saying, like, good product is good product and like that it doesn't really matter how it's built. It's like matters what the end thing is. And I think that's the same thing in media and content. Like, people get really hung up for some reason on like, for example, AI avatars is one I hear. Uh, like often, like we see this in like with Facebook ads in particular, right? Like they work really well for SaaS. UGC content works. Really. Um, people get hung up on the idea though of using them, um, because they're like, oh, there's not trust built in or whatever. And it's like it's, it doesn't even matter. It just works, right? Like you like it. At the end of the day, it's like whatever makes the impact is the thing that you should be doing. So anyway, it's just kind of interesting and fascinating to see what's that's happening kind of, you know, in your world and in comparison to like the stuff that we're experimenting with.

Speaker B: But yeah, no, definitely. It's all the same. It's all the same stuff. Just different, different industries.

Speaker A: Totally. So. So talk to me about how to make an application like from a design perspective look like good. Um, yeah, like in the past, personally what I've done is like, you know, to basically take like a library, something like, you know, Radix or whatever, and um, say, okay, we're going to use all the icons from that, you know, use these components that they provide, et cetera. Is there, is there other ways that you guys are doing it, um, to, to make, you know, it feel like, not like it's a vibe coded site. Like, what, what. What is that? What are those tactical things you're doing?

Speaker B: Yeah. One of the biggest things that we've started to do is we build our own components library for each project. That's one of the first things we actually do. So we'll build a page, call it slash components or whatever, and we will first start. Usually, let's say we have a figma or something. We'll usually upload the figma and um, just as like really just screenshots or exporting them or if they don't have a figma, we have our base components library that we can use or just look for inspiration. And we'll literally just go and like, craft every button, every, you know, component that's required in the app. And um, we'll also have it print out the code and reference, like the component, um, ID so that we can reference it later. So it's like, oh, we want to change this exact button. Like, it's not changed. It knows what to reference. Right. Um, and so by the end of

Speaker A: serializing or giving a skew to everything. That's crazy. Exactly.

Speaker B: Yeah. So we're serializing and then. And then we go and we build and we, we tell it, hey, only use this components library. And it. Hopefully it does. Now, obviously it's not perfect, I would say definitely alleviates a lot more than, um, than without doing it. But, um, yeah, you end up with like a, um, much more robust process for design. I think, um, the biggest thing though, honestly is, um, AI is just really good at taking the, like, the screenshots, um, or images that you give it. And if you just say like, copy this, it does a surprisingly good job. Going back to the whole like, less is more kind of thing, I find that when I write up some like, detailed prompt, um, let's say, let's say I. You know, I have a video where I rebuilt Airbnb, just the front end, just to show how easy and close you could get it. And within 15 minutes, I tried out three different methods. So I tried out, um, the first one was like, just giving it a screenshot, just giving it a screenshot and say, copy this. Nothing else. Okay. The second one was, um, giving it to ChatGPT and um, giving the image ChatGPT and giving ChatGPT to give me a prompt to then give to Lovable. And then I did one where I tried both at the same time. Um, and the one that performed the best by far was the first one, just the image which is super interesting.

Speaker A: Um, it's so funny you say that. I literally did this this morning where we have a partner that we're working with and he's like, hey, can you give me a landing page so that when I send them from the email newsletter to the site, like it lands on a co branded landing page that it doesn't feel like a massive jump. I was like, yeah. Do you have an example of one that you did previously? He sent me the example. We took a screenshot of what that looked like and then basically just like, you know, dropped it into our um, yeah, built it in cursor, right. And then have it deployed to a branch, um, for Eng like you know, minutes later. Right. And I again like you can spend hours trying to describe this, but it's really that visual is like the most effective way. Are you using that. So you're doing the components piece. Are you doing, are you using that for um, the kind of the design as well? Like are you saying, okay, here's this example, use the components library, design it like this, you know, based off this wireframe or what does that look like?

Speaker B: Um, for like you're saying like the. More the UX aspect.

Speaker A: Yeah, more the ux and like, you know, how are you actually doing the build out of this? Or like.

Speaker B: Yeah, I would say that, that I would say is probably more human led. Um, just maybe you know, just kind of our developers, our designers having a little more say or in our cases the client tends to have a bit more of a say. Um, I think that's, that's a little bit more intuitive, you know what I mean? Like it's hard, it's hard to like it's hard for AI. I mean like lovable will do it. Um, but like is it going to be the most optimal thing ever? Probably not. And so I think that just, honestly just comes from like repetition of just like kind of building the muscle memory of like, okay, like this is how like an onboarding should flow. This is how you know, a dashboard should flow. Like I was looking at um, one of our students apps today and you know, like I was looking at the dashboard and I was like, dude, like you have like a set, you have settings, you have like notification. Like it's like it was just way too much and it's like, you know, lovable didn't do anything wrong but it's just like exactly what you told it. Yeah, it's just like, it's just a lot. And so a lot of this just Comes from playing around and experimenting. And we use. There's a couple of cool tools. There's one called Maubin. Are you familiar with this?

Speaker A: Yeah, I am familiar.

Speaker B: Yeah. Like, Mobin's cool for like, referencing this and kind of really seeing what good UX looks like. You can kind of go through every single app that you know and love and just kind of see the flows. Um, 21st dev is another great components, um, library for Lovable, which I've been really impressed with, just how accurate it translates. That's something we're really going to try to push. Um, more on is building our own kind of components libraries that we can share with people as well.

Speaker A: That would be such a huge growth lever for you guys. Like, lovable templates is such a massive growing category. Um, it's super interesting.

Speaker B: Yeah, exactly.

Speaker A: So is there any other tools? Yeah, it's kind of a perfect segue. I was going to ask you about that next. Um, like, other. Other tools that are, you know, kind of critical to what you guys are doing or things that people should be focused on.

Speaker B: Um, I think one that's like, that's like in the design category, that's a bit underrated as well is, uh, Figma Community. Um, I don't know if. I don't know why, I don't know. I don't really see people talking about this, but Figma just has like a library of like, all these designers and just put out work for free that you can just like, copy and paste. And it has like every single, like Airbnb, uber, like, it's all there, you know, you can all just. People just remade it. So I think that's really great. And you can use like, HTML to code and Fig in Figma. Ah, sorry. Like Figma to code or whatever. And you can like, copy and paste the components of the code into Lovable. So if you ever want to be like, super specific, you can do that. Um, again, like, I find that image and then kind of being more detailed after is the best strategy.

Speaker A: No, that makes a lot of sense. This is great. What should I be asking you that I'm not that, like, is kind of, you know, anything that's happening in this space that that's, you know, we haven't talked about today.

Speaker B: Let me think. Um, M.

Speaker A: Maybe something for you too. Like, we haven't really gone into, like, you run an agency, right? Yeah. This is like a part of your process is deal flow. Like, that's a huge component of this. Um, everybody is doing their own version of this. Right now where it's like build content that drives you know, volume into it. I know you have crazy LinkedIn game. Um, is there, you know, maybe we could talk about that or some of the other like um, lead gen strategies that you're using or seeing be successful for agencies that. Especially these AI powered agencies.

Speaker B: Yeah, um, yeah, I think, I think the biggest thing right now that um, I would say the most important thing when you, when you think about these tools to kind of summarize is um, you know they're going to make the barrier to entry a lot. Um, but it also means that there's going to be way more competition and the bar for what is a good product and what people come to expect is only going to get higher, um, as people continue to push the bar. And so I think the question is really now about what you work on I think is such an important part of the equation and also thinking about the distribution side. So it's like while everybody's talking about lovable and AI and all these AI tools, it's great. But okay, what are you building first off and then second off, like how do you actually get users on that? Um, these are two really big and important questions. Um, and so um, I think it's important to think through it more holistically. Um, I also think that um, if you're looking to make money with these tools, um, my bet would be to kind of, I mean just follow what I'm, I'm doing essentially which is focus on one tool. Um, I see a lot of people try to like be the masters of like everything AI from like automation to you know, like content to uh, lovable kind of stuff. Like I think uh, I think this is going to become incredibly verticalized. Um, I think you're going to have like your lovable experts which may be like more startup focused. You're going to have like your cursor experts which are going to be more maybe like developer enterprise focused. You're going to have you know, nan experts. Obviously we've already seen that. Um, and so I think you're better off being like the guy in one lane rather than being the kind of everything guy.

Speaker A: No, I totally agree. It's really interesting. It's like you know, productized services are what perform, are uh, performing the best right now. And it feels like you know, in relationship to that, that you know, that focusing down like you're talking about to the one of these single tools and riding that wave and just being like I'm as you know the thought leader was in that category. To say it in a disgusting way, that sounds terrible. But like, in reality, that's what that actually is. Um, is kind of the, the way to go about this. Um, the, uh, to come back to this. I, I, I, I know the, you know, the audience is all growth people. Like, they, they, they're trying to figure out how do they find alpha. Um, you're killing it on LinkedIn. Um, I know YouTube is something you guys are scaling up. Any tactical things you're seeing work in both of those or any other, um, channels that you're seeing be effective for you guys right now from a.

Speaker B: Yeah, I mean, we've been experimenting a lot. Um, I will say that I think, you know, being an early mover in this space has helped a lot. Just from like a, you know, obviously it's, it's doing really well and so you're kind of riding, riding the wave. Um, I would say there's a, there's a few, like, tactical things of like, I do think, like, I have two, I have two fundamental kind of principles to, uh, posting content, uh, or just marketing as a whole. Um, the first is that, um, people are selfish and don't actually care about you or your business. Um, and so I know that sounds blunt, but the reality is that people are reading content to better themselves or get ahead in their own way. And a lot of people I see, create content, um, that is, like, about themselves. And it's like, nobody cares. That's why it flops. Or they'll post a screenshot or an image of the product they're building. It's like, again, nobody cares. That's the first thing. The other one is that people are bored and want to be emotionally stimulated. Um, I think that's why you're seeing people. Like Roy Lee is the epitome of this, which is just like every single thing he does is designed to make you emotionally react, even if you don't want to. Um, unfortunately, negative emotions are easier to, um, get out of the people. And so that's kind of what he does. But if you can kind of combine both of those. Right, Which I think is what I've really been doing, which is like, uh, every post I try to add value. I try to, you know, give something to people, whether it's, you know, playbook or, you know, components or whatever. I also try to add in a little bit of controversy. And I think the space as a whole is like, so controversial right now that it's like, really easy to do. I think that, like, me Being younger and like building an agency in the space is controversial in itself. Um, I got into like a really big beef beginning of this year with like the Brett designjoy guy.

Speaker A: Amazing.

Speaker B: I love that, you know, like, like this. It's, it's just true that like these kind of things, like, people are bored in their everyday lives and if they can get a excuse to, to kind of check out for a bit entertainment.

Speaker A: Right, like, and like, it needs to be approached and thought of right from that angle. A hundred percent agree.

Speaker B: So, yeah, and be, be visual. I think images and like, stuff like that are really important. Um, it just takes up more, more on the feedback. Um, post consistently. I try to post, um, five times a week. Like, you know, Monday to Friday. Um, um. And yeah, it just kind of starts compounding. There's really no secret.

Speaker A: Awesome, man. No, it's been super valuable. Um, um, where can people find you if they want to reach out? We'll include all the links in the show notes as well.

Speaker B: Yeah. Um, I'm on Twitter @JacobsClug, LinkedIn I believe JacobClug. Uh, YouTube is something that I'm experimenting with and uh, I know Cody's giving me some notes on it, but it's something where I'm working on, so it's hard. Organic is a different breed than just typing, uh, into Twitter and LinkedIn. Um, but you can find me on there. I do a lot more deep dives on building with Lovable. Talking about even building an agency. I did a vlog where I went to Sweden hanged with lovable.

Speaker A: You can crush it right now, which is Lovable builds. How to build lovable, um, applications that don't look like AI made them that will go viral tomorrow. Right. Like, it's like all the pain points that you guys have solved, like just talking about that and turning that into content will be I a ridiculous amount of leave, like inbound to. I will come from that. Like, if you show them how to do like we see this all the time. Like show them how to do something and be like, cool. Like, if you don't want to do this yourself because it just, you know it's going to take hours of your time to actually figure this out. Um, just like hire us basically to do this right. And it's a perfect, perfect way to do it. YouTube is some of the best traffic I've ever seen. It's. I feel like it's so underrated right now, honestly. YouTube organic.

Speaker B: So, yeah, LinkedIn too. I think LinkedIn is super. Uh, people spending on X. But LinkedIn is, uh, they're really pushing content out, and so I, uh, would say it's my biggest platform. I know we spoke about this last time about, you know, uh, that's great.

Speaker A: Awesome, man. We'll include that. Uh, thanks for coming on and, yeah, we'll have you back later on, uh, just to kind of give a State of the Union and what's going on, you know, six from six months from now in the Vibe coding, uh, arena. Because I know it's. It's a lot of new companies coming into this space, too, which is kind of fascinating to see that happen. Yeah.

Speaker B: Super cool. Appreciate it. Thanks.

Speaker A: Yeah. Thanks, brother.