How Payout was built: a class-action app vibe-coded in two weeks
A 23-year-old developer, a finance creator with a huge audience, Claude Code and Cursor. Here's the stack, the timeline and what other builders can copy.
- to build version 1
- 10–14 days
- revenue during Shipaton 2025
- $30,017
- about 50 days in
- $20K/mo
- ARR shown on May 2, 2026
- $1,003,227
Payout is a phone app that finds class-action settlements you might already qualify for and helps you file the claim.[1] It was built by Connor Burd, who says the whole thing was written with AI, and grown by his co-founder Casper Opala, a personal-finance creator.[1][5] It won first place at RevenueCat's Shipaton 2025 hackathon, and by May 2026 the founders were showing just over $1M in annual recurring revenue.[2][5] Here's how it came together, using only what the founders, their interviewers and the hackathon organiser have published.
The team: one builder, one audience
Connor started building apps about three years before Payout, with no computer science background, learning from YouTube videos and Stack Overflow. His first app, a social app called Hotspot Events, took six to eight months to build and, in his words, "kind of failed". After that he moved to utilities and tools, which he found easier to scale on his own.[3]
Casper brings the distribution. He describes a personal-finance audience of 10 million followers across YouTube, TikTok and Instagram, and the split is simple: he handles growth, Connor handles the product.[5] The Superwall Podcast summed him up as someone who turned that audience into a $217,000 app business in five months "and he didn't write a single line of code".[4]
Where the idea came from
The idea came from Casper's own brand deals. On the Superwall Podcast he talks about reverse-engineering his own brand deals to find app ideas that are already proven, and turning rejected sponsorship offers into apps.[4] The pain point was one his finance audience already had: settlement money left on the table because filing is too clunky.[5] The team's own pitch is that millions of dollars in settlements go unclaimed every year because most people never hear about them.[1]
The build: about two weeks, no hand-written code
The team's Shipaton write-up is blunt about it: the app "was entirely written using Claude Code and Cursor, not one line of code written by hand". The art and the app icon were generated too, and the developers had never made a React Native app before. They say they shipped v1 in 10 days.[1] In his Starter Story interview Connor puts it at under two weeks, compared with six to eight months for his first app.[3]
Connor laid out the process he follows, step by step:[3]
- Download around 20 apps in the niche, plus a few apps he finds beautiful, and screenshot every page of every onboarding.
- Line the screenshots up in one big Figma file, pick the best parts and redesign them in his own style.
- Write down the data before writing any prompts: a text document explaining the data, plus example JSON with notes on each attribute.
- Build the core features first and leave onboarding for later, dropping the screen designs straight into the AI so it writes each screen.
- Keep iterating when the AI gets details wrong, but don't be too picky. Shipping a version quickly beats shipping a perfect one.
“I recommend making a text document with an explanation as well as some JSON data that explains what your data actually looks like”
By the time of that interview he said he was only using Claude for AI coding.[3] Here's the stack as the team and Connor described it:
| AI coding | Claude Code and Cursor[1] |
|---|---|
| Mobile app | React Native for iOS and Android, with Expo for builds[1][3] |
| Backend | Node.js and TypeScript REST API, hosted on Vercel[1] |
| Website | v0 and Cursor; Next.js and TypeScript[1][3] |
| Subscriptions | RevenueCat, also used for pricing tests[1][3] |
| Analytics | Mixpanel and Adjust[1] |
| Design | Figma for UI design, App Store screenshots and the app icon[1][3] |
The hard parts weren't only code. The team says it had to help people file without ever "practicing law", so disclaimers and opt-ins went into every screen. Some settlement administrators take months to pay, so they added a status timeline and gentle "still waiting" nudges. And building entirely with AI meant "lots of negotiation with AI".[1]
Launch: winning Shipaton
Payout went out during RevenueCat's Shipaton 2025, a hackathon that drew 812 submitted projects. It took first place for the Grand Prize, and RevenueCat singled out how it was built: design, code and assets produced through AI-assisted development with Claude Code and Cursor. RevenueCat listed over 17,000 users, $30,017 in revenue, 1,750 paying subscribers and more than 500,000 social media impressions.[2]
Growth: creators first, ads second
Connor says the first push came from influencer campaigns: a creator already in the niche whose videos drove thousands of downloads. When a video works, you get more creators to repeat the same format, and those videos become a library of ads you already know will perform.[3] On the Superwall Podcast, Casper says two video formats are responsible for all of Payout's significant revenue. The episode also covers never saying the app's name out loud in a video, a ManyChat comment-to-DM automation with a 50% click-through rate, coordinating five creators on one day to reach #70–80 in the App Store, and reusing organic winners as Meta ads.[4]
The MRR Story write-up describes the mix as roughly 80% organic and 20% paid. It says creators are paid on performance, often $1–2 per thousand organic views; Apple Search Ads bring the highest-intent users; and paywall and onboarding changes go through 80/20 or 90/10 split tests.[5]
Timeline
About 3 years earlier
Connor starts building apps. His first, Hotspot Events, takes six to eight months and doesn't work out.[3]Sep 29, 2025
Connor starts Payout's Shipaton project page on Devpost. The team's write-up there says v1 shipped in 10 days.[1]Oct 13, 2025
RevenueCat names Payout first place at Shipaton 2025.[2]Oct 26, 2025
Starter Story interview: about $20K a month, roughly 50 days in, with over 12,000 downloads.[3]Mar 18, 2026
Superwall Podcast: $217,000 in revenue in five months.[4]May 2, 2026
The founders show $1,003,227 ARR ($83,602 a month) on a live dashboard in an X AMA, about nine months in.[5]
Lessons you can steal
- 01
Write the data down first
A short text spec plus example JSON means the AI doesn't have to guess what you're building.[3] - 02
Design with screenshots
Collect the best screens from about 20 apps in your niche, remix them in Figma, then hand the designs to the AI.[3] - 03
Treat onboarding as the product
Connor reckons 90% of users of a paywalled app may only ever see the onboarding, so he spends as much time on it as on the features.[3] - 04
Keep it small
His top advice: build something simple with a great onboarding. One to three good features is enough.[3] - 05
Pair code with distribution
- 06
Know your legal lines
In a regulated area, design the guardrails in. Payout built disclaimers and opt-ins into every screen.[1]
Sources
- [1]Payout (RevenueCat Shipaton 2025 submission)Devpost, written by the Payout team · Sep 29, 2025 · checked Oct 21, 2026
- [2]Shipaton 2025 WinnersRevenueCat · Oct 13, 2025 · checked Oct 21, 2026
- [3]I vibe coded a $20K/month mobile app in 14 days (Connor Burd interview)Starter Story · Oct 26, 2025 · checked Oct 21, 2026
- [4]Casper Opala - The Stupid Simple Content Strategy That's Making Apps Go ViralThe Superwall Podcast · Mar 18, 2026 · checked Oct 21, 2026
- [5]80% Organic. 20% Paid. $1M ARR in 9 Months. Here's the SystemMRR Story · May 2026 · checked Oct 21, 2026Compiled by MRR Story from public sources and written in the first person; the ARR figure is the one the founders showed in an X AMA.
Something wrong or out of date? Tell us and we'll fix it.