How One Person Built and Shipped a Card Game App from Zero in Three Months
林政賢 ·
Seven years ago I made a deck of question cards just so interviews wouldn't go quiet. Seven years later that deck became a game that passed App Store review. In between there was no engineer — just me, one computer, and three months.
The deck from seven years ago
I filmed the channel PurpsyNight for seven years. In those seven years I sat behind the camera and asked a few hundred people questions. Some were light. Some, once asked, left the room silent for three seconds.
At first the deck was just a rescue tool. Later I realised the deck was the core of the show — it let two people skip the small talk in minutes and go straight to the places they never normally went. Relationships, late nights, ambiguity, regret: the things people "want to say but never say out loud".
Over seven years the questions grew into a huge database. Only a few minutes ever made the edit; the rest sat in my folders. That felt like a waste, because those questions genuinely work. I've seen it too many times: two friends of ten years learning something about each other because of one card; a couple talking until 3 a.m. because of one question. That isn't an effect I invented. It's what I watched happen, for seven years.
Why an app, not a T-shirt
When creators make merch, the first thoughts are T-shirts, mugs, stickers. That's just moving the logo somewhere else; it has no real connection to the show. I didn't want to sell that.
Viewers watch me ask guests with the cards, laugh, and that's the end. But what if the deck were in their own hands? What if tonight, over a drink with friends, they were the one asking? That is extending the show's experience outward.
Years ago a mentor told me: "To make money, do the kind of thing you make once and copy infinitely." I've been an industry lecturer at Wenzao Ursuline University of Languages for 11 years, and that line hit hard — a class, once taught, is heard only by the people in the room. Turn it into a product and it keeps working while I sleep.
Physical cards need printing, storage, shipping and returns, and changing one question means reprinting. A digital deck, made once, appears on a thousand phones at the same time; adding a new set is one back-end update.
Make once, copy infinitely.
This is its purest form.
I'm not an engineer. Here's how the three months went
Let me be clear: I don't come from code. My background is a decade-plus of film and VFX, plus a little design.
Three years ago I'd have gone looking for an engineer, then bled out between expensive quotes and long timelines, and most likely it would have come to nothing. But 2026 is different. AI turned the high wall between "the idea in my head" and "a product that works" into a slope you can climb.
I don't need to write every line of code. But I do have to judge the requirements, break down the steps, sign off the output, and guard the aesthetics. Those four things happen to be exactly what a decade-plus of filmmaking and 11 years of teaching built — I just used to apply them on set, and now I apply them on a screen.
In three months I didn't just build a card game that passed App Store review; I opened the whole road from development to release. The holes along the way were real:
⚠ THREE HOLES I ACTUALLY FELL INTO
One. Rejected by Apple three times in a row under Guideline 4.3(b), forcing me to re-learn what the review rules actually police.
Two. A hole in the payment mechanism, found right before release, that nearly zeroed my subscription revenue.
Three. Halfway through writing the cards, discovering dozens of questions were duplicates with a few words swapped.
Every hole hurt at the time, but every fix is general knowledge. If you're also stuck at "I have an idea but can't code", the App development & launch page collects the process we actually ran and the pitfalls to avoid.
What this series will cover
Over the next few pieces I'll write down how PurpsyNight X Card Game went from nothing to shipped. Not "I made it, so can you" motivational soup — what actually happened, how I decided, where I got it wrong.
- Rejected by Apple three times: what Guideline 4.3(b) is really about
- How one person writes a thousand cards, and how he finds out half are duplicates
- How someone who can't code uses AI to get the product built
- Designing the paywall: every trade-off in a three-tier membership
- The security audit before launch: the holes that nearly zeroed my subscriptions
The card game is live now. If you want to see what the product looks like, download it from the App Store and try it — that feeling of skipping the small talk and talking straight through to dawn.
If you're sitting on an idea of your own: next time you hit a system-development problem, don't rush to get outsourcing quotes. First define the core interaction you actually want to solve. Want to talk through how to break your project down? Book a consultation here.
About this piece: merged from two near-identical articles published in August 2026; the old URLs redirect here automatically.
Further reading: AI built me an App, then got it wrong dozens of times — the dozens of mistakes made while developing this app, written up in one piece.
Author：林政賢（Director · Gen AI creator & engineer · Founder of TangYi Studio）