Skip to content
← Back to work

Pandora's Cards

Pandora's Cards: the rebrand that got the app approved

An AI deck generator for icebreakers, road trips, and team offsites. Built as Drunk Cards, rejected by Apple as spam, and relaunched once I widened the product past drinking games.

Live at pandorascards.com ↗

The premise

Every party deck I had bought ran out by the third round. The cards stopped feeling like this group. They were written for nobody in particular. The fix was not a bigger deck. It was a deck that could adjust to whoever was in the room.

So I built one. Tell it who is playing and what kind of night it is, and it writes a deck in seconds. Edit any card that misses. Then swipe through it one-handed. Right to play, left to skip.

Pandora's Cards showing a generated card, captioned 'Create a classic in seconds'
One prompt in, a full deck out. Editing any card is one tap.

Apple said no

I built it as Drunk Cards. Under that name it never reached the store at all.

App Review rejected it as spam — the name and the branding, not the code. That is not a verdict you argue with. It is one you work around.

I could have filed off the edges and resubmitted under a duller name. Instead I used the block. A drinking app has one occasion, one audience, and a permanent ceiling on where it can be recommended. The name had to change regardless, so the product could stop being about drinking.

Apple rejected a name. What it bought me was a hard look at how small a market I had picked.

What changed

The generator stayed. Everything framing it moved.

The old build asked for a game type — Truth or Dare, King’s Cup — an environment of House Party or Bar, and an intensity that topped out at Unhinged. The current build asks for a deck type: icebreakers, story prompts, hot takes, challenges. The setting is a game night, a road trip, dinner, or a team offsite. Intensity became tone: light or bold. Only “who’s playing” survived untouched — it already offered coworkers and strangers, which in hindsight was the tell.

The deck creation screen with categories for icebreakers, story prompts, road trips, and team offsites
The same generator, pointed at a room instead of a bar. Deck type, setting, and tone each replaced a drinking category.

Same engine, wider door. A team offsite and a pre-game want the same thing from software — a prompt that makes people talk — and only one of them can be advertised on the App Store.

It went live as Pandora’s Cards on 25 June 2026.

How it works

A creator profile showing published decks in a grid
Publishing a deck is a first-class action, so the app is a small social platform with a generator inside it.

Three pillars do the work:

A card mid-play with a swipe counter reading three of fifty-three
Right to play, left to skip. One thumb, the whole game.

Design language

This is the file the rest of my portfolio inherits from. A few decisions did most of the work.

Dark backgrounds, on purpose. The app gets used in rooms with the lights down. A bright white screen burns in those rooms. Near-black is the only honest default. The portfolio took the same baseline (#0a0a0a over #000) because the same restraint reads as premium in a quiet room too.

Oversized editorial typography. Cards are read at arm’s length, across a table, by people who are not paying full attention. Anything under about 24px misses. That habit became this portfolio’s display scale — type big enough to be part of the design, not just the carrier of it.

One accent, used sparingly. A single bright colour marks the primary action and almost nothing else. That is the rule this site’s lime follows: one or two moments per scroll, never more.

Motion that costs nothing. Swiping is the whole interaction model, so each swipe has to feel substantial without dropping frames. transform and opacity only. The same budget applies here.

Tech and decisions

What I took from it

Try it

Free on iOS. pandorascards.com has the full pitch and the App Store link.