Figma to App Store in Under an Hour: The Exact AI Process
Everyone is piling into AI web apps right now.
Ross thinks that's exactly why you should look at the App Store instead.
He's a product engineer who has shipped over 20 iOS apps. In a recent Starter Story Build video, he sat down and went from a blank Figma file to a real app running on a real iPhone in about an hour. Not a prototype in a browser tab. An app on TestFlight you could send to your mom.
Forget the speed for a second. The reason this build worked on basically the first try is one step he does before ever touching a code editor, and it's the step almost every vibe coder skips.
Here's the full video, and below it, the whole playbook broken down.
Why iOS is the underpriced opportunity
The logic is simple.
Web apps got easy. Lovable, Bolt, Claude Code: anyone can one-shot a landing page and a CRUD app now. Which means everyone is doing it.
iOS is still annoying. There are no great vibe coding tools built for it yet. Mobile design has its own rules, mobile users behave differently, and the system architecture is nothing like web. Plus Apple makes you suffer through a review process.
Friction filters out competition.
→ Where building is hard, fewer people build, and distribution is less crowded.
That's the bet. And with 20+ shipped apps, Ross has more evidence than most that it pays.
Step 1: Steal an idea that's already winning
They gave themselves 30 minutes, so they picked something with one function: a mood tracker.
Not a random pick. Ross had seen an app called How We Feel blow up on TikTok. He pulled it up in the App Store live on camera:
- 4.9 stars with 23,000 ratings
- Top 100 in Health and Fitness
- Apple Editor's Choice
You tap the emotion you're feeling, add a short note, and it saves to a list. That's the core loop of a chart-topping app.
So instead of inventing something, he took screenshots of How We Feel and Daylio, AirDropped them to his Mac, and used them as raw material.
His words: it's really difficult for any designer to design from scratch. Every good designer starts from inspiration in the same market, because mobile users already have habits. Your job is to give them the experience they already understand, but better.
One feature, proven demand, existing patterns to copy. That's the starter kit.
Step 2: Design everything in Figma before writing a single prompt
This is the part most people skip, and it's the core of the whole video.
Before opening a code editor, Ross laid out the entire app as wireframes in Figma:
- A welcome screen (app name, logo, one button)
- An onboarding flow that doubles as the first mood entry
- A home screen listing past entries
- A detail screen with sub-emotions, the journal note, and a delete button
He grabbed the dark background from one competitor, the green from Daylio, and asked ChatGPT for a list of 8 major emotions with sub-emotions to fill his 4x2 grid.
Why bother, when the AI could generate screens for you?
Because when you vibe code without a plan, you end up in an endless loop of "actually, move that button" and "wait, this screen feels empty." Every correction burns time and fills your context with noise. In Figma, Ross spotted a missing element on the home screen in under a minute. In code, he would have found it three steps later and paid for the rework.
→ Design is cheap to change. Code is expensive to change. Do your changing where it's cheap.
He also points out you can get feedback on a Figma file before you build. Validation before development, even for a weekend project.
I saw the same pattern in how Mattia designed and built an app with AI in 45 minutes: the people shipping fast with AI are not prompting harder, they're planning first.
Step 3: Feed the AI one screen at a time
The stack: Windsurf as the editor, Expo for the iOS project. If you know web development, Expo is basically the Next.js of mobile.
Ross's prompting process has three rules worth stealing.
Rule 1: Warm the model up before asking for anything. His first message was a screenshot of all the screens plus a high-level description of the app, ending with "ask me any questions before we start with the welcome page." No build request yet. He wants the model holding the full picture before it writes a line.
Rule 2: One screen per prompt. He noticed that sending multiple pages at once makes every AI editor do a sloppy job. So he copies one Figma frame, pastes it, and describes the layout in plain words: logo top, subtext and button bottom. The AI still can't reliably infer element positions from an image, so he tells it.
Rule 3: Use the best model available. He ran Claude Sonnet 4 and kept Windsurf's planner mode on so the AI remembers which screen it's working on. His take: the cheaper the model, the more bugs you'll be debugging yourself later. The model fee is the cheapest part of the whole project.
First screen came back with the gradient, the title, and the button placed right. He deleted a tab bar he didn't want, then repeated the loop for every screen, using Windsurf's voice input to go faster.
Total build time for all screens: under 20 minutes.
Four years ago, Ross was writing production code at a real job. His estimate for the same app back then: a month, with a four-person team.
The first bug, and how to talk to AI about bugs
The full flow worked on the first run. Pick an emotion, pick sub-emotions, write a note, see it on the home screen, delete it. No bugs.
Then he restarted the app and it dumped him back on the welcome screen instead of home.
His bug report is a small masterclass. He didn't type "there's a bug." He wrote: everything looks great, but when I restart the app it takes me back to welcome, which should only happen when there are no entries yet.
→ Describe what happened, what you expected, and when it happens. Give the model one honest sentence of context and it will usually find the cause. Say "it's broken" and it will guess.
One prompt later, fixed.
From simulator to a real iPhone
This is where most tutorials stop, and this one didn't.
Two commands. EAS handled the App Store Connect setup that used to mean ten manual forms. The cloud build took 6 minutes 47 seconds. One submit later, the app appeared in TestFlight, and Ross held up a phone with the mood tracker running on it.
Blank Figma file to installable app: about an hour, including the two technical hiccups they hit on camera.
A detail worth knowing: TestFlight has built-in feedback. When a tester screenshots your app, they can attach a note that lands directly in your project settings. Ross uses this on all his apps. Tell your first 20 users "if anything feels weird, screenshot it and write why," and you've got a feedback pipeline with zero extra tooling.
How a free mood tracker turns into money
Fair question, because nobody pays for a list of their own feelings.
Ross's answer: ship the free single-feature version first, then charge for the layer on top. His idea here is an AI report that reads your past month of moods and notes, finds the triggers, and suggests what to do about them. That's the subscription.
The order matters. Feedback first, paid features second. When you ship one feature, users tell you which second feature they'd pay for. You don't have to guess.
One student story from his bootcamp shows the ceiling: an app that organizes your TikTok and Instagram saved recipes pulled 2.8 million views and 10,000 downloads almost immediately, because the idea came straight from watching what people complain about on TikTok.
The marketing playbook is TikTok, not ads
For distribution, Ross keeps it blunt.
Search TikTok for the existing mood tracker apps. Find their viral videos. Study the first sentence, the hook ("stop scrolling if you have a hard time expressing your emotions"). Then film a 20 to 30 second clip: their hook style, your app demo, link in bio.
The biggest consumer apps of the past few years grew exactly this way. If you want the deeper version of that playbook, I broke down how to get your first 1,000 app downloads without spending a dollar.
And budget your patience for review: App Store approval takes up to 24 hours to get picked up, another 24 for a verdict, and rejections are sometimes just reviewer roulette. Ross once got rejected, replied "that's not true, try again," and got approved.
What I'd steal from this build
→ Pick one feature. The longer you build, the longer you go without user feedback.
→ Copy patterns from top apps in your market. Users already have habits; meet them there.
→ Wireframe the whole app before prompting. Changes cost minutes in Figma and hours in code.
→ One screen per prompt, best model available, planner mode on.
→ Report bugs with context: what happened, what you expected, when.
→ Ship to TestFlight fast and let screenshot feedback drive the roadmap.
None of this needs a technical cofounder. It needs a Saturday.
FAQ
Can you really go from Figma to the App Store in 37 minutes?
The build itself took under 20 minutes and the full session about an hour, including setup and two technical hiccups. Getting on the actual App Store adds Apple's review, which takes 24 to 48 hours. TestFlight, where real users can install your app, is available almost immediately.
Do I need to know how to code to do this?
It helps, but less than you'd think. Ross is an engineer, but every step in the video is prompting: describing screens, pasting Figma frames, explaining bugs in plain English. The skills that mattered were design judgment and clear communication, not syntax. Non-technical builders ship apps this way every week now.
What tools were used in this build?
Figma for wireframes, ChatGPT for the emotion list and logo, Windsurf as the AI code editor running Claude Sonnet 4, Expo for the iOS project and cloud builds, the iOS simulator for testing, and TestFlight for distribution.
Why build for iOS instead of the web?
Competition. AI made web apps easy, so everyone builds them. iOS still has friction (design rules, review process, weaker AI tooling), which keeps the field thinner. Mobile users also pay for subscriptions more readily than web visitors, and the App Store handles billing for you.
How do you get users for an app like this?
TikTok first. Find viral videos from competing apps, copy the hook structure, show your demo in 20 to 30 seconds, put the link in your bio. It costs nothing, and the recipe-app example from Ross's bootcamp (2.8 million views, 10,000 downloads) shows the upside.
Want the numbers behind stories like this?
Every week I interview bootstrapped founders doing $100K to $10M a year on the Profitable Founder Podcast. Real revenue, real playbooks, no fluff. If you liked this breakdown, you'll like the interviews more.
