Loading summary
Transcript27 lines
- [00:01]
A
Vibe coding, all the rage on the Internet. Every single person you know is probably Vibe coding something, but is it a really critical skill for us marketers to learn and anyone else who wants to grow their business? Should we be learning how to Vibe code and learn to work with AIs to build things? Well, in this episode we're going to go from no idea a rough idea of our audience to get in an incredible idea and build in something for that audience and in hopefully under 25 minutes. So let's get into today's episode of Marketing against the Grain.
- [00:38]
B
Before we get into today's show, here's a quick word from HubSpot. Cutting your sales cycle in half sounds pretty incredible, but that's exactly what Sandler training did with HubSpot. They used Breeze, HubSpot's AI tools to tailor every customer interaction without losing their personal touch. And the results were pretty incredible. Click through rates jumped 25%, qualified leads quadrupled and people spent three times longer on their landing pages. Go to HubSpot.com to see how Breez can help your business grow.
- [01:26]
A
All right, so let's start with a really good LinkedIn post from Jason Lemkin. It really caught my attention because he is in the depths of Vibe coding and I am in the depths of Vibe coding. And actually I had a post that was just about to go live when I saw his. And so like he talks about the fact he was at five days with Vibe coding with Replit. For everyone who wants to know what is my Vibe code stack, he's using Replit. I have a team account with Replit. I have a team account with Lovable because I have some people who are actually better developers than me working alongside me and I'm going to get into that point and a little bit of time. We are also using Cursor. We are not using cursor code. I've heard it's incredible. So that's really what we're using is Claude Replit, Cursor and Lovable and team accounts with Lovable and Relet. Now he's using Replit and basically he talks about the fact that he was Vibe coding all weekend.
- [02:15]
C
Super addictive.
- [02:16]
A
He was burnt talks about the fact that he thinks he can get pretty far with Vibe coding. But there is still a lot of challenges with getting to a kind of point where you can create something that.
- [02:27]
C
Is production ready, that you can just.
- [02:29]
A
Create an app via Vibe code in and ship something and it's production ready app. Let's go and show you one of the apps I'm building now, again, I'm trying to give some equal airtime to Replit and Lovable because they are both great tools. I use them both. For this episode, we may actually do a little bit more work in Replit. I could have done it in Lovable, but we have Anton hopefully coming on the show and I'm going to actually use a lot of Lovable. Anton, the founder of lovable, just raised $1.8 billion, no big deal. Done it in their very first year of business. Pretty absolutely amazing. And we're going to have Anton, hopefully on this show we're going to feature a lot of Lovable. And so in this, I thought I would give Replit a little bit of airtime and do some Vibe coding in Replit. But this is one of the apps, right? So I can talk because I'm actually doing this. There's a lot of people talking about Vibe coding. I think online they're not actually building real apps. And so you can see this is just one of the apps. I have others in Replit and others in lovable. This is 26,000 lines of code so far. Quite an impressive code base. Thank you very much. Lovable. Now we're going to get into actually giving you a walkthrough and showing how all marketers could be builders in the future. And I think that's going to be true. I think AI enabled marketers are going to have to build code powered experiences versus content experiences and we can show an example of that. But for someone who has been Vibe coding for thousands and thousands and thousands and thousands and thousands of lines of code each and every week, it is a really addictive thing. I think it's probably the most addictive thing that I have ever done in my career. Now I graduated from university and computer science. I spent some years being a very average software engineer. I've talked about that a lot in interviews and other things. I always wanted to build. I started my first company in third year in university. It was a community site. You could basically build communities around cities at advertisers, businesses. We went to door to door and we sold ad sales and we had to code things. And really my stumbling block each and every time was I just was not a great coder, right? I was great at the ideation. I was great at understanding how to bring things to life. I was great at understanding how to put things into the product that would help me grow the product, distribute the product. I just could not code right. I would sit there hours at nighttime reading My code in books and I would read, read, read, read, read, code.
- [04:50]
C
Code, code, code, code.
- [04:51]
A
And then I would wake up in the morning and I would be like, I don't remember anything. I literally don't remember anything that I did last night. Anything that I read. It just does not stick into my brain. But I'm still quite technical and that's why I did okay when I went into marketing, because I had a good blend of creativity and technical skill set. And I think that's like, actually quite hard to find. And so vibe coding has been really addictive because I can sit there at nighttime and actually build things, get minimal viable products, and actually work alongside the AI to bring ideas to life. Now, I do find it really challenging at times to solve bugs. And even though I am quite technical and can dive into code, what I have found is the AI at times will fix one bug, create several more bugs, and it becomes a game of whack a mole. And you're like, oh, like, fix that bug, fix this bug, fix that bug. No, I didn't tell you to fix that bug. I didn't tell you to do this. Please just fix the thing I want you to fix. And luckily I have my bro who's very technical and can fix things for me because he is working on these sites with me. And I have another good friend who is working with me on all of these kind of different websites. And that person is an incredibly experienced developer. So I do have help and I think if I did not have help, I could not get as far as engineering with some of these apps. And so what about the marketeer today? If you're watching this and you're thinking to yourself, well, should I start coding things for my audience? And I think what I call code powered experiences, which is building lightweight tools and these kind of things for your audiences, will become a big part of your marketing playbook. But how do you do that? Well, let's show an example of how simplistic that could be, because I think if you are not trying to build fully fledged apps, which I am doing, and you were just building lightweight tools for your audience, then I do think a marketeer can actually integrate this skill set into what they want to do and use that playbook to grow their business and build things of value for their audience. And so we're going to get into it now, the first thing we're going to do is we're going to pick a company to build a lightweight app for. Now, this will be a first version because I'm on the clock. I have about 15 to 20 minutes to do the show. I have not prepped anything. So I want to show how easy it is. I've hard stopped. I have to be somewhere. But that's a good thing because it's going to force me to go through this and show you how easy it can be in real time versus having a bunch of thing prepped. Now there's four steps we're going to go through. We're going to pick the company, then we're going to build an icp, an ideal customer profile. Then we're going to use the ideal customer profile and we're going to use a prompt to take that ideal customer profile and create a list of potential code powered experiences, which are really lightweight interactive tools that I can build for that audience to attract them towards my brand. These code powered experiences, we're going to have a prompt do that for us and then we're going to take one of the best ideas and we're going to have Replit on this occasion build it. But again, I cannot emphasize this enough. Replit and Lovable are both incredible tools and, and I am not taking sides in the AI assistant coding wars because I think there is a giant market when everyone becomes builders and these tools are going to be and already are incredible companies. Okay, so let's build the icp. Now we're going to do something fun because there is, you know, competition heating up in the AI coding space. Lots of great companies. We're going to pick Relet because we're going to build on Relet and we're going to take the ICP ideal customer profile of a marketeer who wants to be a builder and integrate that into their skill set using replit, which I think is pretty fun. So let's use Perplexity Pro Labs and we use the Labs feature. You can see down here, it's a great product. Now you are probably not an AI degen like me and have subscribed to every single tool and in some cases team accounts. So you can use whatever tool you have access to. It would be okay. I love Perplexity Pro Labs for a lot of the research things that I'm doing. And so in this case what we're going to do here is say that you are an AI market research analyst. You're going to create an ICP for the builder marketer. Now I actually think this is a real Persona for both lovable and replit. So if anyone is watching from those brands, think this is a great Persona for you. All It's a real one. And so we're going to go through and we're going to talk about the job role. Their ambition is really to use Vibe code in AI environments. We're going to talk about Replica specifically here for this example and the fact that they want to create these apps for list growth and lower CAC and all of these different reasons. And then we're going to say, hey, give us this snapshot. We want to include all of these things like their goals, the pain points, the motivations. We need this to actually build potential lightweight apps that we can build for this Persona. And that's going to be the next step. So let's kick this off and we are going to get going to build this app icp. Now, AI is really incredible to build a detailed icp. This is just a very lightweight one. If I was building this really for Replit and I was internal to that company, obviously I would have access to a lot more internal data unstructured and structured and then external data unstructured and structured. That I would use AI to craft the perfect ideal customer profile for the builder marketer and then I would use continual feedback from that group to actually refine it over time. So I don't want you to think that this is a recommendation for how you get the ideal ideal customer profile, but it's good enough for the examples that we're going to go through. So let's kick this off. Okay. And I will just say if anyone for perplexity is watching, I actually do have some contacts so I'm going to hit them up. But I really want access to comment now. I do wonder if the reason I haven't got access to comment is because why I'm in the eu. Everything seems to take a little bit longer in the eu. Okay, we're back. Okay, so this is the icp. The builder marketer represents a new breed.
- [10:46]
C
Of growth demand gen marketers at startup.
- [10:47]
A
To mid market companies who leverage Vibe code in AI environments like Replit to ship lightweight web app calculators and micro SaaS. Lead magnets. Pretty true. They sell critical challenges of demonstrating value through building, not just telling. Again true. Achieving faster list growth, lower CAC and differentiated brand authority. Lower CAC is really coming from these kind of lightweight apps. There is a moment in time right now where your real virality is through the product demo. If you can have really cool product demos primarily for AI tools, you can get a ton of virality that used to be from the internal app metrics like inviting people all the Things kind of Dropbox invented but it really is through the product demo now. So we have some really great stuff. Right. We had the key market insights where the average B2B CAC reaching this number marketing their site and predictable demand engine as their number one pain point. Like really cool stuff here technology the emergence of vibe coden popularized by Andre Caparte in early 2025. Talk about here traditional competitive advantage. Yes, traditional lead magnets are given away private market. These kind of code powered experiences are much, much better strategic implications for vendors, for marketers, market timing, recommended engagement strategies. So we've got enough here to run the next prompt. So here is a pretty killer prompt actually. All right, so what does this do? It's an elite marketing code experience researcher. Basically what it's going to look at is your ICP and it does a lot of work itself to try to figure out what are lightweight apps that it can build for or the marketer can build for this ICP to attract that ICP to the brand. And it's going to do some pretty great things where it's going to pitch them and and then stack rank them and it stack ranks them by relevance, virality, potential effort and confidence in terms of being successful. So let us run this. So basically what we're doing now is like we have an icp we want to build lightweight apps. For now, the thing you would want to do is only do this for ICPS that would use lightweight apps. Right. There's no point doing this for an ideal customer profile that these kind of.
- [12:49]
C
Things would not be relevant.
- [12:50]
A
For now I would say replit building for the builder marketer that is a perfect Persona to build these kind of tools for. So we've got the right ICP to do this and now it's going to take that ICP and say, well, we don't actually have any ideas like what are good examples of lightweight apps we can build for these folks. And so we're going to run that. We're going to run this in ChatGPT03. Still my model of choice. Anyone who saw the Grok for breakdown, I am still continuing to use Grok 4 side by side because it was so impressive on the benchmarks. But O3 is still my model of choice right now.
- [13:33]
B
Hey everyone, we'll be right back to the show, but first let me tell you about a podcast that I love. It's a podcast called I Digress. It is hosted by my friend Troy Sandage. It's brought to you by the amazing HubSpot Podcast Network. The Audio destination for anyone in business. And I Digress is great. It's got 30 minute episodes and the podcast is all about helping you eliminate complexity, complications and confusions in your business. It's really heavy in frameworks and strategies to help you scale and sustain your success. And you know, recently Troy had a really great episode. It was called Branding Secrets Leveraging style to change the narrative, Boost, influence and profits. And that was with his guest toy Sweeney. It's really good. You should check it out and you can listen to that or all the episodes of I Digress anywhere you get your podcast.
- [14:29]
A
All right, we're back.
- [14:31]
C
Okay, so we thought for a little bit of time and so here it recaps the ICP snapshot.
- [14:36]
A
So again, this is our builder, Mark.
- [14:38]
C
Tiers and then gives us some insights about why that is a good ICP to go after. And so it says things like the low code no code market has reached 45 billion in 2025. It shows you the source that it got this from. The budgets are already earmarked for DIY build tools like Replit. So for example, 84% of learner devs use an AI code assistance. So that's a big population of these folks are going to need tools like replit and others. I thought that's kind of cool. Pulled out Karpathy's vibe code and meme that went viral. This kind of started everything in February. Cultural signal. Non traditional coders feel empowered to ship apps.
- [15:17]
A
Right.
- [15:18]
C
And then it's going to talk about some of the reasons it's picked. I'll show you now it's picked an idea around calculators. Calculators yield higher quality leads versus quizzes in 2025 tests. Points to ROI calculators as both food conversion plays. So tries to pull out some insights on the kind of ideas it's going to share. And I took a quick look. These are good ideas. Most of these ideas for Replit are going to be things that Replit would build that the marketer can then replicate to generate demand for their audience, which makes total sense. So if you're trying attract this kind of builder marketeer, you would build them an array of different templatized apps that they could easily copy and paste and use the AI code and assistant on Replit to tweak for their audience. So it has things like this ROI snapshot builder. So this comes back to the calculator. It really allows the marketeer to create these ROI calculators for their audience in a really easy way. So it pulls out the elevator pitch, the description we put in here, the path to replit. So how does someone actually end up signing up for a replit account? A lot of it is they can fork that template in their replit workspace. It goes through the exact user journey here that you can discover. You answer some prompts to create your own calculator, then you fork it, tweak, copy and embed and capture emails. So it goes through a lot of the core features, all of these different things. Even brings out a similar reference point which is the website grader for HubSpot. We know this works because we've done something similar. I thought this was really cool actually. They've called the lead magnet Arcade. Really it's a template of different lead magnets that are really easy for a builder marketer to come in and clone different aspects of it and remix them and create something for their audience. So again, really, how do you help build a marketers create these code powered experiences through replit, fork it and then customize it for their audience. And then by doing that you create replit accounts. Now what we're going to go down here and do this one here. So this is a tool that allows you to create a Persona just by simply pasting in a blog URL or a website URL. Again, very similar kind of thought process here that you would be able to export the Persona deck and even customize the Persona tool. But you would need a replit account to do that. But to begin with, you could come in, use it and then say, actually I want a version of this for my audience. I'm going to build something like this for my audience. I can fork that in my replit account, customize it and then use it to acquire demand for me. So I thought this was a pretty cool one. It's minimal, viable enough that I think we can probably just create something in repl.it so what I would normally do then is take this idea and then work with ChatGPT03 to come up with a prompt to build a first version of this in repl it to see.
- [18:13]
A
How it would work.
- [18:14]
C
And again showing you how easy this would be, right, That I have a greater grade, whatever code, powered experiences work for my audience. And the important part is how does this generate demand for you? How will people come in and use this tool? And then you will acquire demand from it. So we're going to do the Persona inside scraper and so we've already kind of done a quick prompt and we did put that into replit. So you can see here.
- [18:39]
A
I worked with it to kind of.
- [18:41]
C
Adjust some things actually and then worked with it to create a prompt specifically for Replit. And so we went through and talked about what a minimal viable version. Now I would say this is a very minimal viable version. There is some tweaks I would make. This is the first version that we've created. I will say huge kind of shout out to the Replic team. I do think they've made some changes recently and their code and assistant just seems to be incredible right now. And so it really did one shot this I haven't actually tweaked it. So again you will be able to do this pretty easily in replit and then tweak your lead magnet, your co powered experience in any way you want. But the kind of core goal here is to show you how easy we went from developing the ICP to great in these kind of code powered experiences to create a minimal version. So we're going to put in the HubSpot CRM and then we're going to generate this Persona. So Replit did everything, it didn't even adjust the prompt or anything because what it's really doing is running this through an LLM to create the Persona so all of this can be improved and so it generates this kind of buyer Persona here. So it's small business owner has pain points, budget constraints or limiting growth. Difficulty in managing customer data challenges in creating remarkable customer experiences. The reason it's got SMB is because it's a free CRM for startups, small businesses, complications with data migration. All of this is pretty true need for scalable CRM solutions and it has the goals that you can streamline business operations. So it has pretty good goals and then it has buyer triggers and this is really good actually Experience budget limitations. They need a centralized CRM desire to improve the customer experience. They're outgrowing their current CRM. They're seeking cost effective business solutions. Again all of this is pretty good if you want to make the buyer Persona much much better. I could easily play around with the LLM and the other thing I would do is I would upload customer call transcripts and I would combine those two things to create this buyer Persona. So you can see how easily I can actually iterate and rapidly make this better. I actually think I will build this into a tool which is your ICP tool. It's given me a great idea actually how to iterate on this really fast and so you can view the raw JSON here if you want to download it that way. I haven't actually tried this. Oh, this is cool. So this is the lead magnet part, right? So we were trying to figure out how do you actually acquire demand. So in this case, if you want to download your buyer Persona, you actually have to go into the repl.it account to be able to do that. Now I actually think what you could easily do is if I was really going to build this tool as a.
- [21:07]
A
Lead magnet tool, which I think I.
- [21:08]
C
Am going to build it for my newsletter, the AI marketing generalist, to allow people to come in and use the tool if they're signed up to my newsletter. What I will probably do actually is allow you to add your URL, upload some content like customer call scripts and things like that, and then generate a Persona and show some of it and gate some of it. And then actually you can get the rest of it by providing your email address or whatever kind of lead magnet CTA you need. But that's just how I think marketers become builders. Because it's going to be as easy to create code powered experiences as it is to create content powered experiences. Things like we've traditionally done blog posts and PDFs and all these different things. What's actually really interesting is AI is better for a marketer at creating code powered experiences right now because it objectively knows how to create code and create apps and there's not a lot of perception and taste whether it did that thing or not. It actually either built the app or did not, where it's much harder to do content because content still needs a lot of human taste, human point of view, human creativity. Also there's a lot of perception about like what is good content to one person may not be good content to another person. Where building an app is somewhat binary in that you built the app or you didn't build the ad, it's functionable or it's not functionable. And so I actually think AI is really incredible for marketers to add code paired experiences to their arsenal of tools and marketing playbooks to acquire demand, acquire users, all of those different things. So that's the episode trying to give you a idea of how quickly you can go from, hey, I have a general idea of who my ICP is, but I can fine tune that VIX Labs. I can then just upload that using the prompt that I showed you that we can make sure we give away that will grade that for you and create these kind of five different types of code powered experiences that would be great for your audience, that you can build for your audience and then you can easily go into a replit we did that. Or you could go into a lovable of that your tool of choice. Build a version. Make sure you're really clear on what's the path to someone using that app to then acquire in demand for your core product or service, and you've got another great tool or tactic to add to your list of marketing playbooks to acquire demand in this AI age. That's the episode. Hope you enjoyed it. Until next time, this is marketing against the grain.
- [23:38]
A
Sam.