I shipped two iOS apps solo — what AI changed about being a designer
For fourteen years I designed products that other people built. That's the deal, as a product designer: you shape the thing, you hand it over, engineers make it real. I was fine with that deal. Then, in 2025, I broke it — I designed and built an AI-powered nutrition app called Numi and put it on the App Store. Alone. In 2026 I did it again with Tysha, a white-noise app for new parents.
No engineering team. No agency. A designer, an idea, and today's AI tools. This post is the honest version of how that went — what AI genuinely changed about my job, and what it turned out to change not at all.
Why these two apps
Both came from my own life, which I now think is the only sane way to pick a solo project. Numi exists because I wanted to track what I ate without the misery of manual food logging — search the database, weigh the portion, give up by Thursday. The design thesis was simple: point your camera at the plate, and the app does the rest — calories, macros, micronutrients, plus context-aware guidance it builds up over time.
Tysha exists because of the 3 a.m. shift with a baby. Every white-noise app I tried had looping audio that clicked, ads (ads! at 3 a.m.!), or needed an account and a connection. So: real-time generated white, pink and brown noise, fully offline, no accounts, one-handed. The entire product is designed around a person holding an infant in the dark.
What AI actually changed
The prototype became the product
My old workflow: mockups → clickable prototype → specs → wait → review builds → polish what survived. The prototype was a promise, not a product. With AI-assisted coding tools, that gap collapsed. I could design a flow in the morning and hold a working version on my phone in the evening. Not a simulation of the feature — the feature.
Iteration got absurdly cheap
When changing a real, running app costs minutes instead of a sprint, you stop guessing. Numi's photo-logging flow went through many working versions before launch — versions I could actually cook dinner with, not click through in a design tool. Some ideas I was sure about died in an evening of real use. That kind of learning used to cost weeks of someone else's time, so it rarely happened.
The design tool changed shape
I still design screens, but a growing share of "designing" now happens in the working product itself: adjusting motion, spacing, haptics, the feel of things — details that never survive the translation from mockup to build. For the first time in my career, what shipped is exactly what I designed, because there was no translation step.
What AI didn't change
This is the part the hype skips, and it's where fourteen years of design earned its keep:
- Knowing what to build. AI will happily build the wrong thing faster. Numi isn't "an AI calorie app" — it's a position against the misery of manual logging. Tysha isn't "a sound app" — it's one-handed, offline, instant, because that's what 3 a.m. actually requires. Those decisions are the product. No tool made them.
- Taste and editing. AI output — code or UI — trends toward the average of everything it's seen. Every screen in both apps had a generic version first. The job was saying no to it, over and over.
- The last 20%. App Store review, edge cases, empty states, what happens when the camera sees a plate that's half in shadow. Shipping is still shipping. It's just that one person can now afford to do all of it.
What this means if you're a designer
I don't think every designer needs to become a founder. I do think the ground has moved under our role, in a direction that favors us:
- Designers are the best-positioned people in tech right now. We've always known what to build and how it should feel; building was the locked door. The door is open.
- A shipped side project is the new portfolio piece. Hiring teams now routinely ask for evidence of AI fluency. One live product answers every version of that question — it proves product thinking, craft, technical fluency and follow-through in a single link.
- Start smaller than feels impressive. Tysha is a "small" app and it taught me as much as products with teams behind them. Pick a problem from your own week, scope it to one core loop, and take it all the way to shipped. All the way — the learning is disproportionately in the last mile.
If you're a designer thinking about building something: my next post breaks down the actual design lessons from working on AI — Designing AI products: lessons from building Numi. And if you're starting your design career from zero, start here: How to become a product designer in 2026.