TL;DR: I have three apps live on the App Store, all designed and engineered solo: Glimpse Vision AI, AnimaEcho and Ground Control Chat. AI did not make me a better engineer. It removed the distance between having an idea and finding out if it was any good, and almost everything else follows from that.

Can one person really ship an iOS app now?

Yes, and I want to be precise about why, because the popular version of this story is wrong.

The popular version is that AI writes your app. It does not. What it removes is the tax on trying. Before, the cost of finding out whether an interaction felt right was two days, so you would argue about it instead. Now it is twenty minutes, so you just look. Multiply that over a few hundred decisions and one person can carry a product that used to need a team.

Everything else about shipping is exactly as hard as it always was.

What actually takes the time?

Almost never the code. In my experience the real weight sits in five places.

• The last ten percent. Error states, permission denials, no network, the user who taps the button twice. This is most of the work and it is invisible in every demo.
• Subscriptions. Products, receipts, restore purchases, what happens when a payment fails. Nothing about it is fun and all of it is required.
• Privacy and permissions. Camera and microphone access, data declarations, saying clearly what you do and do not store.
• The store itself. Icons at every size, screenshots that explain the product in three seconds, a description that ranks, review responses.
• Deciding what not to build. The most expensive mistakes I made were features I built well and should never have started.

If you budget for those and not for "the coding", your estimate will be roughly correct for the first time in your life.

What does the workflow look like?

1. Write the one sentence. If I cannot say what the app does in one sentence a stranger understands, I do not start. Glimpse was "point your camera at anything and ask out loud".
2. Build the core interaction first, on a real device. Not the onboarding, not the settings screen. The single moment the whole app exists for. If that moment is not good, nothing downstream saves it.
3. Use it for a week before adding anything. Living with it kills more bad ideas than any review meeting.
4. Then build the boring shell. Onboarding, paywall, settings, empty states. Now you know what they need to say.
5. Ship early and narrow. Ground Control went out at version 0.1. Real users tell you things a year of polishing will not.

Where does AI help most, and least?

Most: anything with a known shape. Networking, data models, animations, the fifty small utilities, and the entire category of work you already know how to do and simply do not want to type. Also debugging, which is where I gained the most time back.

Least: knowing what the product should be. Taste in motion and timing. Anything requiring a real device in your hand. And judgment about scope, where it will enthusiastically help you build the wrong thing beautifully.

There is also a trap worth naming. It is now so easy to add a feature that you stop asking whether you should. The discipline that used to come from effort now has to come from you.

Does the design get worse when one person does everything?

It can. The safety rail is that you have to keep the two hats separate in time. When I am building, I build. When I am reviewing, I open the app cold, on a phone, like a stranger, and I am allowed to be brutal about work I did yesterday.

Doing both at once produces the thing everyone fears, which is a product that is technically fine and has no soul. I go into the underlying craft question in design engineer versus product designer.

What would I tell a designer starting today?

Start. Your first app does not need to be a business, it needs to be finished. Finishing teaches you more about product than any amount of reading, because App Store review does not care about your intentions.

And build something you personally need. All three of my apps came from my own daily friction, which is why I never had to guess at requirements. The tooling side of how I work is in how to use Claude Code as a designer.

FAQ

Can a designer build an iOS app alone with AI?

Yes. AI handles most of the implementation, but you still own product decisions, craft, testing on real devices, and the App Store submission process.

How long does it take to ship a small app solo?

The core experience can be working in days. Getting from working to submitted usually takes weeks, because of edge cases, subscriptions, privacy declarations and store assets.

What is the hardest part of shipping solo?

Scope. Nothing external stops you from building too much, so the discipline to cut has to come from you.

Which apps has Carlos Lastres shipped?

Glimpse Vision AI, a real time camera and voice assistant. AnimaEcho, a voice AI companion. Ground Control Chat, a phone client for Claude Code sessions. All designed and built solo.

Carlos Lastres is an Apple Design Award winning product designer and software engineer based in Tokyo, a 3x TEDx speaker and a United Nations speaker. He designs products that turn attention into revenue, and ships his own apps end to end.