Everyone can build an app now. Prompt Claude Code or Codex for an afternoon and you have something that compiles.
Which is exactly why most new apps die. They all look the same.
Ryan and Sarah don't have that problem. Their cat Pomodoro timer, Cat on Chair, makes about $18,000 a month, and the number one question under Ryan's first Starter Story video was not about revenue. It was "how did you design that?"
So Starter Story Build brought them back and made them show their whole app design process, live, on a brand new app. Idea, hand-drawn animation, PNG layers, Codex prompts, working iOS prototype. A couple of hours, end to end.
I pulled out every step so you can copy it.
The Couple Behind Cat on Chair
Quick context if you missed their first episode.
Ryan quit a $200K product design job. Sarah was a freelance illustrator who assumed her drawings could never live inside an app, because apps are "supposed" to look flat and minimal.
Together they built Cat on Chair, a Pomodoro timer where a hand-drawn cat hops on a chair and sits with you while you focus. 127,000 downloads. Around $18K a month. I broke down the full numbers in my Cat on Chair story, so I won't repeat them here.
The point of this episode is different. It answers the question that matters most in 2026: when anyone can ship an app, how do you make one people remember?
Their answer: the design process. Here it is, step by step.
Step 1: Pull the Idea From Something Physical
Ryan and Sarah don't brainstorm in Notion. Their ideas come from watching their own cats.
The new app they build in this episode started with a real annoyance. Sarah writes ideas in her phone's notes app constantly, then forgets they exist. In her words, it's not that different from crumpling each idea into a paper ball and throwing it in the trash.
And her cat? Her cat digs the paper balls back out.
That's the app. You write a note, it crumples into a paper ball, a little paw drops it into a jar. A shelf of jars holds your categories. When you open a jar, you get your ideas back, the way the cat returns the trash.
Notice what happened there:
→ A boring, solved category (notes apps)
→ A real behavior they observed at home (cat steals paper balls)
→ Combined into something nobody else would make
Ryan says it straight in the episode: AI made everything easier to build, so more people are building. That doesn't mean the problems are taken. It means you need an angle on a problem people already solved. Combine niches until the combination is yours.
Cat on Chair is "just" a timer. This one is "just" a notes app. Simple ideas win when the execution has a personality.
Step 2: Animate It by Hand in Procreate Dreams
This is the step most builders skip, and it's the whole moat.
Sarah's animation process for the paper ball:
- Watch cat videos on social media to see how cats actually move.
- Act the motion out herself. She literally crumpled paper to understand how a flat sheet becomes a ball.
- Draw the first frame and the last frame, then fill in the frames between to smooth the motion.
- Do it in flipbook mode in Procreate Dreams (the iPad animation app, not regular Procreate) at 12 frames per second.
- Use keyframes to move the jar up as the paper ball drops in.
- Add the detail that sells it: a little black paw appears, closes the lid, and gives it a twist and a push down.
- Color pass at the end. Black paw, bluish-green jar, done.
Total time: 30 minutes to an hour. Her estimate for a beginner with some iPad drawing skill: about a day.
A day. That's the entire cost of making your app look like a human made it.
I keep seeing founders spend weeks tweaking Tailwind themes to stand out and never once consider drawing something. Sarah spent an hour in an iPad app and produced the thing App Store screenshots get saved for.
Step 3: Export Layers, Not a Video
One handoff detail makes the designer-to-AI pipeline work.
Sarah doesn't send Ryan a finished video. She exports every frame as a PNG, separated into layers: jar back, jar front, paper ball.
Why layers matter: in the finished app, the paper ball renders between the jar's back layer and front layer. So the ball sits "inside" the jar no matter how many notes you've saved. Sarah never has to draw a jar with 1 ball, 2 balls, 14 balls. The code composes the scene from her parts.
→ Draw once, let the code compose every state.
If you're working with a designer (or you are the designer), this is the spec: frames as PNGs, one folder per layer, and a note saying what sits in front of what. That's it.
Step 4: Let Codex Build It, but Prompt Like a PM
Ryan can't code the traditional way. The new app got built anyway, in a couple of hours.
His stack for this build:
- Codex for the coding. He picked it over Claude Code for one reason: tokens. "Codex is really more generous with our token," so long build sessions cost him less.
- An app called Orca to manage multiple projects and AI sessions side by side.
- The default macOS terminal. Nothing fancy.
His first prompt did a lot of work. It told Codex to build the app in UIKit and Swift, described the one feature (write a note, hit save, play the animation), pointed at Sarah's frames, and included two instructions most people would never think of:
- Compress the PNGs before combining them into a MOV file. Playing raw PNG sequences on iOS bloats the app; a compressed video file stays small and still looks good.
- The animation is 12 FPS. State it. If you don't, the AI defaults to 24 FPS and your carefully drawn animation plays at double speed.
That second one is the kind of detail you only learn by shipping. The machine does the repetitive building, you supply the judgment calls. It's the same division of labor that makes business automations work: automate the grunt work, keep the decisions.
For anything more complex than a demo, Ryan adds one more move: he has the AI interview him first. What does this app do, what's the user flow, what are the edge cases. The AI asks, he answers, and the spec writes itself before any code exists.
And his biggest warning: know exactly what you want before you prompt. "If you can't really clearly describe it, it will just make something up that's kind of close to what you want." Close-to-what-you-want is how apps end up generic.
Why Hand-Drawn Beats AI Slop
Pat said it on camera, and I think it's the most important line in the episode: Cat on Chair works because "it doesn't look perfect. It doesn't look polished. It looks really like a human designed it."
Sit with that. The flaw is the feature.
Every AI-generated interface converges on the same rounded-corner, gradient-button sameness. Users scroll past 40 of those a day. A wobbly hand-drawn cat paw closing a jar lid stops the thumb, because nothing else in the App Store looks like it.
Sarah's old assumption (my illustration style doesn't belong in apps) was exactly backwards. Her style is the business. Ryan's AI prompting just gives it a place to live.
I wrote about the general version of this in how to design an app with AI. The tools keep changing. The principle doesn't: taste is the part AI can't generate for you.
What to Steal From This Episode
→ Mine your actual life for ideas. Sarah's forgotten notes plus a trash-digging cat beat any brainstorm doc.
→ Pick a solved category on purpose. Timers and notes apps have proven demand. Your angle is the differentiation, not the category.
→ Spend one day on handcrafted visuals. Procreate Dreams, 12 FPS, first frame and last frame first. That day is your moat.
→ Export layered PNGs, not videos. Let code compose the scene so one drawing covers every state.
→ Put production details in the first prompt. Frame rate, file compression, tech stack. The AI won't guess them.
→ Make the AI interview you before big features. The spec conversation is cheaper than the rebuild.
→ If you have a weird skill, aim it at software. Illustration, music, calligraphy, whatever. Weird is distribution now.
FAQ
What app did Sarah use to animate Cat on Chair?
Procreate Dreams, the iPad animation app (a separate app from regular Procreate, which is for drawing). She draws in flipbook mode at 12 frames per second: first frame, last frame, then the in-between frames. One animation takes her 30 minutes to an hour, and she estimates about a day for a beginner with some iPad drawing experience.
Do you need to know how to code to build an app like this?
No. Ryan says he can't write code the traditional way. He describes the app to Codex (UIKit and Swift), feeds it Sarah's PNG frames, and iterates in plain English. The skill that matters is describing exactly what you want, including details like frame rate and file compression, because vague prompts produce generic apps.
How long does this app design process take?
In the episode, idea to working iOS prototype took a couple of hours: 30 to 60 minutes for the hand-drawn animation, then an AI build session with some back and forth on the UI. A more complete app takes longer (Cat on Chair has backstories, collectibles, and a full backend), but the first working version is an afternoon.
How much money does Cat on Chair make?
About $18,000 a month, with 127,000 total downloads, built by two people with no dev team. The full revenue breakdown, pricing model, and launch story are in my Cat on Chair article and Ryan's first Starter Story interview.
Steal More Playbooks Like This
Ryan and Sarah built a real moat with an iPad, an hour of drawing, and a well-written prompt. Nothing in that sentence is out of your reach.
If you want more breakdowns like this from founders doing the numbers, not just talking about them, I interview bootstrapped SaaS founders making $100K to $10M a year every week on the Profitable Founder Podcast.
