I Built the App I Needed. Here's What That Actually Taught Me.

There’s a question I get from founders pretty often: how do I know if my idea is worth building?

My answer used to be theoretical. Now I just point to SoleTrackr.

SoleTrackr is a running shoe mileage tracker I built for myself. Not as a client project. Not as a business bet. As a solution to a problem I had personally, that no existing app was solving the way I needed it solved. And the process of building it taught me more about AI-assisted development than almost anything else I’ve done.

The Problem Was Mine

I’m a distance runner working toward 1,000 miles this year. I rotate three shoes, each with a different role in my training, and each with a different mileage ceiling before it needs to retire. Most running apps track your runs. None of them tracked my shoes the way I needed.

I wanted automatic Apple Health sync so I wasn’t logging manually. I wanted to assign each run to a specific shoe, not just split mileage evenly across the rotation. I wanted retirement alerts based on actual per-shoe usage, not estimates. I wanted it private, on-device, with no account required and no data leaving my phone unless I explicitly chose that.

That app didn’t exist. So I built it.

The Experiment Inside the App

SoleTrackr was also a deliberate experiment in a few things beyond just solving my problem.

The first was the subscription question. I’m skeptical of apps that put everything behind a monthly fee. So I designed SoleTrackr from the start to do its core job completely free, on-device, with no account and no subscription required. A one-time Pro unlock exists for additional capabilities, and a Sync+ tier is in the model for cross-device features that would require server infrastructure. But here’s the honest truth: if no one needs it, I may never build it. No server costs, minimal maintenance, and a clean product that just works. That’s actually the ideal outcome. The point was never to extract recurring revenue. It was to build something useful.

The second experiment was proving what’s now possible with AI-assisted development. SoleTrackr is a native iOS app written in Swift. I did not write a single line of code. Claude handled the entire build through Claude Code, and what came out the other side wasn’t a rough prototype or a proof of concept held together with duct tape. It was a production-ready app. Polished UI, real Apple Health integration, privacy-first architecture, a full pricing model. Shipped and Apple-approved.

That matters because it changes the math on what’s buildable.

Bespoke Is Back

For a long time, custom software was something only well-funded teams could afford. You needed developers, time, and budget that most solo founders and small teams simply didn’t have. So everyone settled for tools that were close enough, cobbled together workflows that almost fit, and paid subscription fees for features they only used halfway.

That calculus is shifting fast.

When AI can take a well-framed problem and produce a production-ready native app without a traditional development team, bespoke software becomes attainable for a much wider range of founders. Not every idea needs to become a venture-backed product. Some ideas just need to exist, work well, and solve the specific problem you have.

And because these micro-apps are small and focused by design, expanding them is straightforward. Adding language support to increase the total addressable market, for example, is a natural next step that doesn’t require a major rebuild. Small scope, maintained intentionally, with room to grow when the signal is there.

SoleTrackr cost me time and intentionality. Not a development team. That’s a fundamentally different world than the one most founders are still operating from.

What the Process Looked Like

This is where the last four posts come together in practice.

I didn’t open Claude Code and start prompting features. I started with the same discovery questions I’d ask any founder: what problem am I solving, for whom, and what does a version one worth shipping actually look like? For SoleTrackr the answers were tight and clear, which made everything downstream faster.

Once the foundation was framed, AI moved quickly. The architecture decisions, the Swift implementation, the Apple Health integration, the privacy model, the pricing structure. All of it accelerated because the problem was defined before the first prompt was written.

Structure first. AI second. The process doesn’t change just because you’re the founder and the client at the same time.

What’s Coming Next

SoleTrackr is the first app out of what I’m building under Genesis 37 Labs, and it won’t be the last. There are more micro-apps in the pipeline, each one starting from a real problem, each one following the same process. An Android version is also in the works for those on the other side of the aisle.

For the rest of April I’m going to keep documenting the builds as they happen. Real projects, real decisions, real outcomes. If you’ve been following this series and wondering what the framework looks like when it’s actually running, this is where it gets concrete.

SoleTrackr has been approved by Apple but hasn’t officially launched yet. Before the wide release, I’m looking for a few more people to put it through its paces. If you’re a runner with a shoe rotation problem and want early access, head over to soletrackr.com. And if you’re a founder wondering whether this kind of build is possible for your idea, that’s exactly the conversation Genesis 37 Labs exists for.