Case 04 / Concept, social game and AI
Plotline
An AI game master that turns a group chat into a multi-day mystery, so nobody has to host and everyone gets to play.
- Year
- 2026
- Timeframe
- 1 month
- Role
- Product designer: research, brand and UI
- Tools
- Figma, Illustrator, Claude, After Effects
The 30-second version
01 Problem
Every group game needs a host, and the host never gets to play. Someone has to keep the secrets and know the twist, party kits need everyone in one room on one night, and quick social games never build into a story.
02 What I did
I designed the brand and design system, 30 screens for iOS and Android, a creator studio on the web and a website. The AI is the narrator and referee, never a player. It improvises only inside beats a human wrote, and it shows its work at the end.
03 Outcome
Plotline is a concept, so nothing here is a measured result. I set targets for a pilot with 50 crews: 70% of campaigns finished, 75% of players playing daily, 30% of crews starting a new game, and fewer than one report per 1,000 turns.
Friends don't need more games. They need someone else to run them.
01Four rules
Calm by default: one clear action per screen. Every screen is a moment, built to be screenshotted. Secrets feel secret: private screens go indigo. Play first: the rules are one tap away.
02A host that never sleeps
The game master writes a chapter every morning from what the crew did the day before, so every move changes something visible in the story. The tone follows the genre and the crew's spice level, and a twist arrives when the chat goes quiet.
03Secret mode
Each player holds a secret role, so private screens had to feel private at a glance. The best tool for secrecy turned out to be a mode switch, not a new design: the same components, gone indigo.
04What I learned
Pastel accents work best one at a time. A single soft colour per screen kept 30 screens calm and readable. And an AI game master earns trust by showing its work at the end.
Next I would test whether crews finish 5-day games or need a 3-day default, how players feel when the AI plays a missing friend's minimum moves, and whether creators want revenue share or just an audience.
Case study boards (15)
