How VibeKit Focused on App Retention Over Acquisition to Grow from $0–1K MRR

Honestly, it started as a hack I rigged up for myself. I got a coding agent, OpenClaw, running on an EC2 box with my repo hosted right there on it, and suddenly I could iterate on my apps from my phone. Make a change, ship it, see it live, all without opening my laptop. I remember doing it from my couch, then in line for coffee, then lying in bed, just talking to my project and watching it update in real time. That was the moment it clicked. I'd always assumed building software meant being chained to a desk with a full dev setup in front of me, and here I was doing it one-handed from my phone like I was texting a friend. The catch was that it took a real pile of DevOps wiring to get there. Spinning up the server, connecting the repo, keeping it alive, none of that is something a normal person would ever do. It felt unfair that the experience was locked behind so much setup. So I set out to build the thing that gives anyone that same feeling, building and running real apps from your phone, with none of the plumbing. That became VibeKit.

The one that worked is Apple Search Ads. It puts us right in front of people who are already in the App Store searching for app-building tools, so the intent is baked in before they ever see us. Those people are actively looking to make something, which means the ones who tap through are far more likely to actually sign up and start building. It's been our most reliable source of real users. The one that completely flopped was Reddit ads. On paper, it looked great - a huge audience of technical, curious people. In practice, we got plenty of impressions and a handful of clicks, but almost none of it turned into users who stuck around. The context was just wrong. Nobody is scrolling Reddit looking to start building an app, so even the people who clicked were idly curious rather than motivated. We shut it off quickly. The lesson for us was that intent beats reach every time. A smaller, high-intent channel like App Store search will out-convert a massive audience that isn't in a building mindset, and it is not close.

The most common complaint early on was reliability. Someone would describe an app, and sometimes the first build wouldn't come up, or an app that was working would go down and need a restart. When you're building from your phone with no logs in front of you, that is genuinely frustrating, and it was our number one reason people left. We've handled it by treating every one of those failures as a bug to root-cause, not a ticket to close. We rebuilt how the system recovers from a crashed build so the AI agent actually sees the error and fixes it instead of silently redeploying. We made deploys zero-downtime so shipping a change doesn't knock your app offline. And we added guardrails so a build that keeps failing stops and tells you what's wrong instead of quietly burning your credits. The hardest part was that these users often couldn't tell us what broke, so we had to get better at detecting failures ourselves instead of waiting for a report. It's still the thing we watch most closely, every single day.

Most people think the hard part, and the valuable part, is the AI writing the code. That's the demo everyone shows, type a sentence, watch an app appear. It's genuinely impressive the first time you see it. But generating that first draft is basically commoditized now. Every tool can do it, and they all do it reasonably well. What almost nobody talks about is everything that comes after that first draft, actually hosting the thing, keeping it online, fixing it when it breaks, and iterating on it over weeks and months as your idea changes. That is the hard part, and it's where people either get real value or quietly give up. A generated app you can't run, can't fix, and can't evolve is just a screenshot. So we bet the entire company on operating the apps people build, not just generating them. Spinning up the first version is easy and getting easier every month. Living with what you built, and making it better over time, is the real product, and it's the part the market keeps underestimating.

A few things. First, reliability, and then more reliability. The bar for building from your phone has to be that it just works every time, so we're pouring effort into making builds and deploys rock solid. Second, meeting people where they are. A lot of our users aren't developers, so we're investing in educational content and in-product guidance that gets someone from an idea to a live app without getting stuck. Third, more advanced features for the people who outgrow the basics and want to build something serious. Underneath all of it, the real goal is retention. We care less about getting more people to try it and more about making the people who do try it never want to stop.

We're at 246 users who've built 148 apps between them, and the part I'm proud of is the trajectory, nearly half of those users, 117, joined in just the last 30 days. It's still early, but it's accelerating.

A mix. I have a computer science background, so I could actually build the thing myself, and I've spent years as a product professional, which taught me to focus on what people actually need instead of what's fun to build. I'm also a serial entrepreneur, which is a polite way of saying I've started businesses that didn't work out. Those failures probably taught me more than any success would have. So I wasn't starting from scratch technically, but VibeKit is the first time all of it lined up, the engineering to build it, the product sense to shape it, and the scar tissue from earlier attempts to know what to avoid.

Early on, I spent too much energy trying to get more people in the door before the product was good enough to keep them. I was optimizing ads, SEO, all the acquisition stuff, and users were showing up. But a lot of them tried it once and never came back. The mistake was treating it as a distribution problem when it was really a product problem. So we stopped treating new signups as the main goal and made retention the thing everything gets measured against. If the people who already found us aren't staying, pouring more in is a waste. That reframe changed both what we build and what we deliberately ignore.