If a non-technical person can build it in a weekend, why pay monthly for something off-the-shelf?
A former fitness coach, armed with little more than a clear idea and fifty dollars in AI credits, spent a weekend in Singapore building a personalized training app that replaced three separate tools with one. The act itself was unremarkable in its mechanics — prompts, iterations, a finished product — but its implications reach further: when the barrier between imagining a tool and possessing it collapses, the economics of software subscription begin to tremble. What was once the province of engineers is becoming the province of anyone who understands their own problem well enough to describe it.
- A non-technical user built a fully functional, personalized fitness app in a single weekend for under fifty dollars — the same price as a few months of the subscriptions it replaced.
- The fragmentation of modern fitness tracking — calories here, lifts there, form cues somewhere else — has quietly been eroding the user experience that established platforms promised to solve.
- Single-purpose fitness tools that do one thing well but integrate with nothing now face an existential question: why pay monthly for rigidity when bespoke software is a weekend away?
- Ecosystem platforms like Strava, whose value lives in community and social identity rather than mere function, appear more insulated — but isolated utility apps look increasingly exposed.
- Vibe-coding doesn't just threaten incumbents; it inverts the logic of software development itself, placing hyper-personalization within reach of anyone who can articulate a problem clearly.
For years, the author lived across three apps — MyFitnessPal for nutrition, Apple Watch for movement, Notes for lifting logs. As a former group fitness coach, the tracking made sense in theory. In practice, it fractured the experience: mid-set form checks meant scrolling through YouTube between reps, and the gym began to feel less like a place to train and more like a place to manage data.
A vibe-coding workshop in Singapore offered a different premise. Using Manus — a general-purpose AI agent, not a dedicated coding tool — and roughly forty to fifty dollars in credits, the author set out to build something better. The process required no code, only clear description: what the app should do, who it was for, which features mattered. Within thirty minutes, the AI had produced a working prototype.
The result was TrainerPro, a web app with an industrial black-and-orange aesthetic the AI itself named "Iron Forge." It housed around two hundred exercises with GIF demonstrations and coaching cues, generated eight- or twelve-week training programs with progressive overload and deload weeks built in, and consolidated what had previously required three separate tools. Early bugs were fixed by simply describing them. By the end of the weekend, it was usable.
Using it in the gym felt different — not because the exercises were novel, but because the structure was finally coherent. The author still relied on MyFitnessPal for nutrition after running out of credits, but the core realization was clear: every gap felt solvable with a few more prompts.
The broader implication cuts at the fitness app industry's foundations. Giants like Strava have evolved into social ecosystems where workouts become community and content — those platforms retain their defensibility. But single-purpose tools that do one thing well without integrating into anything larger now look vulnerable. If a non-technical person can build a personalized trainer in a weekend for a one-time cost, the monthly subscription model for isolated utility apps begins to lose its logic. Vibe-coding doesn't just lower the cost of software — it transfers authorship to the people who understand their own frustrations most intimately.
I used to live between apps. MyFitnessPal for what I ate, Apple Watch for how long I moved, the Notes app for what I lifted. As someone who'd spent years coaching group fitness, I understood the logic of tracking everything — calories, hours, kilograms. Numbers don't lie. But the system itself was a lie. It fragmented the experience. When I needed to check my form mid-set, I'd scroll through videos on my phone between reps. The gym stopped feeling like a place to train and started feeling like a place to manage data.
In February, I attended a vibe-coding workshop in Singapore. The premise was simple: what if all of this lived in one place? I decided to test it by building my own personal trainer app — something that could generate a workout program, log my lifts, and surface exercise cues when I needed them. I had roughly $40 to $50 in credits on Manus, an AI agent that Meta had acquired but which the Chinese government had just blocked from operating. Manus isn't marketed as a coding tool the way Cursor or Lovable are. It's a general-purpose AI that happens to write code. I didn't need to know how to code. I just needed to describe what I wanted.
I started with a plain-English prompt. I told the system what the app should do, who it was for, and which features mattered most — recommending exercises based on goals, tracking lifts, providing coaching cues. The scope had to be right: not so broad that it became unwieldy, not so narrow that it was pointless. Within thirty minutes, the AI had built something. I never touched a line of code. I just watched it work.
The result was TrainerPro, a web app with a grungy black-and-orange interface that the AI described as "Iron Forge" industrial brutalist. It came loaded with a library of about two hundred exercises, each with GIF demonstrations and coaching cues, filterable by muscle group, equipment, or environment. It could generate eight- or twelve-week training programs based on a popular coaching framework, adjusting for goals, fitness level, and starting weights, and factoring in deload weeks and progressive overload. There were early bugs — some exercises didn't load properly — but fixing them required only telling the AI what was wrong. By the end of the weekend, I had something usable.
When I stepped into the gym and opened the app I'd built, something shifted. I'd created this in a few hours. Now I was using it. I generated an eight-week plan and followed it exactly: bench presses, split squats, deadlifts. The exercises weren't revolutionary. The structure was. The app told me what to lift and for how long on any given day. No more flipping between three different tools. No more browsing YouTube between sets to fix my form. I did dial down the intensity on hard days, and I still used MyFitnessPal for nutrition because I'd run out of credits to build that feature. But the core insight was clear: all of it felt fixable. With a few more prompts and a bit more time, I could shape it into something that fit me exactly.
The fitness world is dominated by giants — Apple Fitness+, Strava, MyFitnessPal. They're not just tools anymore. Strava has become a social network where workouts double as content and community. Those ecosystem platforms aren't going anywhere soon. But the functional, isolated tools — the ones that do one thing well but don't integrate with anything else — look vulnerable now. If a non-technical person can build a working, personalized trainer in a weekend for fifty to a hundred dollars, why pay monthly for something off-the-shelf that can't adapt? Building an app still costs time and money, but a one-time investment could be cheaper than years of subscriptions. With TrainerPro, I didn't need separate tools to log workouts, generate programs, or buy training plans from Instagram coaches. It was all there, tailored to me. If something didn't fit, I could change it.
Fitness apps have always been built around scale — one product for millions of users. Vibe-coding flips that equation. It makes hyper-personalization cheap, fast, and accessible to anyone with a clear idea. It also opens a door to distribution. If you solve a common pain point, your app can become a side project or a business, something you share or even sell. Users often understand their own frustrations better than any product team ever will. There's also something intangible at work. Using something you built — with your own design choices and logic embedded in it — feels different from using something built for the masses. In a world where everyone is trying to stand out, that kind of customization has real value.
Citas Notables
Apps that have evolved into full ecosystems aren't going away anytime soon. But the rest of the stack — the more functional, isolated tools — look far more vulnerable.— The author, reflecting on the fitness app landscape
La Conversación del Hearth Otra perspectiva de la historia
You built this in a weekend. How much of that was actually you thinking versus the AI doing the work?
I did the thinking. I described what I wanted, and the AI translated that into code. But the thinking part — figuring out scope, knowing what features mattered, understanding my own pain points — that was all me. The AI was just the tool that made it possible.
So the real threat to fitness apps isn't that AI can code. It's that people can now build exactly what they need.
Exactly. The threat is personalization at a price point that makes sense. MyFitnessPal costs money every month. I spent fifty dollars once and got something that works for me, not for millions of people.
But you still use MyFitnessPal for nutrition. Doesn't that undercut your argument?
It does, a little. I ran out of credits. But that's the point — I could have built it if I'd had more resources. The friction isn't technical anymore. It's just money and time.
What about the apps that have become communities? Strava isn't just a tool.
Those are safe. They've built something social, something that matters beyond the function. But most fitness apps aren't Strava. They're just tools. And tools are vulnerable when someone can build a better one for themselves.
Do you think this changes how companies should build fitness apps?
I think it forces them to either become ecosystems or become irrelevant. You can't compete on features anymore if someone can customize features for free. You have to offer something that can't be replicated in a weekend.
What's the thing you didn't expect about using something you built?
The pride, honestly. It sounds small, but it changes how you relate to the tool. You're not using something someone else decided you needed. You're using something that reflects your own logic.