I Built an iPhone Tattoo App by Talking to AI. Then I Put It in TestFlight.

I wanted to see how far I could take an iPhone app without opening Xcode and building every screen by hand.

So I opened Rork Max, selected Fable 5, and described a tattoo design app called InkMuse.

The idea was simple: choose a tattoo style, describe what you want, generate options, refine the design, and preview it on your body with the phone camera. New users would get three free generations before seeing a subscription offer.

Twenty-eight minutes later, I had the app running on my phone and a build inside TestFlight.

That sentence sounds ridiculous. The process was even more ridiculous to watch.

InkMuse running inside the native iOS preview. Screenshot from my build video at 7:31.

Rork turned the request into an 11-step plan and started working through it. The first build included:

The visual quality surprised me. This did not look like a pile of default components thrown onto a screen. The onboarding had a clear style, the tattoo cards felt intentional, and the experience already had the bones of a real consumer app.

The refinement step failed.

This is the part people often cut out of AI-building demos. I leave it in because the repair loop is the job.

I gave the agent a screenshot, showed it the error, and asked it to trace what was wrong. It inspected the logs, found the problem, changed the implementation, and rebuilt the app.

Then I tested the camera flow on my physical iPhone.

The photo try-on flow placed a generated tattoo on my arm. Screenshot from 14:39.

The result was not perfect. It was real enough to prove the interaction, and good enough to keep building.

That distinction matters. AI can give you an astonishing first version. You still have to test every path, look at the failure states, and decide what quality means for your users.

I connected RevenueCat and created a $29-per-month subscription concept. The app would allow three free generations, then present the upgrade.

For a real launch, I would still review the pricing, subscription copy, entitlement logic, restore-purchase flow, and App Store rules. The video shows the operating model, not a finished monetization decision.

This is one of the biggest changes in app building: a beginner can now prototype the product and the business model in the same session. You can test whether the value makes sense before spending months on the infrastructure.

Rork connected to my Apple developer account and produced the build that appeared in App Store Connect.

The InkMuse iOS build inside App Store Connect and TestFlight. Screenshot from 23:19.

That is a major step. It is not the same as a public App Store launch.

Before I would submit the final product, I would still:

Apple's current Developer Program costs $99 per membership year. TestFlight lets you distribute beta builds and collect feedback before the public App Store review.

The valuable part was not that AI made every decision correctly. It did not.

The valuable part was the distance one person could travel in a single working session: idea, design, generation, physical-device test, monetization concept, debugging, and TestFlight.

You still need taste. You still need testing. You still own the legal and product decisions.

The starting line moved forward by months.

Disclosure: Some links below are affiliate links. If you choose to buy through one of them, I may earn a commission at no extra cost to you.

Want to learn and build alongside other AI builders? Join us here.

This article documents my own prototype and testing session. Product features, plans, pricing, and availability can change. A prototype, TestFlight build, or App Store Connect record does not guarantee App Store approval, product-market fit, revenue, legal compliance, or business results.