Brain Game
A developer’s pencil sketches, turned into a real app — branding, a design system and AI conversation design for a tabletop rules assistant. Built end-to-end in Figma.
The idea behind the app.
Brain Game gives tabletop players an AI assistant for rules questions mid-game. The developer brought the concept, a few hand-drawn screens and no brand — I built everything from the ground up.
Meet Jordan.
Not a hardcore gamer — just someone at a family night, a dorm room, a party. Jordan knows the basics but hits moments of genuine uncertainty mid-game: Can I stack Draw 4s? Does this card override the rules? Is that actually how Free Parking works?
Play correctly, keep the game moving and avoid being the person who stops everything to argue about a rule.
It isn’t confusion — it’s the social pressure of not knowing, and not wanting to make it a whole thing.
Game night has a UX problem.
Jordan is mid-game. Someone plays a Wild Draw 4. The table erupts — “you can’t stack those!” “yes you can, my cousin plays that way!” The game stops. Someone googles it. Three minutes later you’re back, but the vibe is gone.
Rule PDFs and BoardGameGeek forums only make it worse — they pull you further off the table. There was a clear gap for something that just gave you a verdict, fast, so you could get back to playing.
Jordan doesn’t need a rulebook. Jordan needs a verdict. Fast.
Where I came in.
I came in as the UX and brand designer when all that existed was an idea and rough paper sketches — no colors, no fonts, no voice.
Brand, screens, and a bot.
Brand identity.
A blank slate to define from the ground up. The palette landed retro and artsy: rich, muted tones, not the usual primary-color game look.
The screens.
The three screens that carry the product: onboarding, game selection and the AI chat where the real work happens.
Designing the AI companion.
Players need an answer in under 10 seconds, so every word earns its place. I used Claude to design the bot's voice: a knowledgeable friend who's confident, warm, and never over-explains.
Pre-set question chips let players ask common questions without typing — less friction when someone’s mid-game and just needs a fast answer.
“When a 7 is rolled, two things happen: first, any player holding more than 7 cards must discard half of their cards, rounding down. Second, the player who rolled the 7 must move the robber to a new terrain hex…”
“Roll a 7 → anyone with 8+ cards discards half. Move the robber to any tile and steal 1 card from a player there.”
The difference between a bot that answers questions and a bot that actually works at a game table.
A ready-to-build package.
Ready to build, not just to look at.
What I handed back was ready to build: a full identity, a working design system, defined screens, and a conversation framework any developer could run with.
Two-sentence-max responses mean players get a verdict and get back to the game.
A palette and type system that feels smart and warm at once, carried consistently across every screen.
Quick-tap chips mean returning players don’t think about how to ask. That speed builds habit.
Two firsts and one big lesson.
This project was a lot of firsts — my first contracted work and my first time designing an AI voice. Designing for a world I didn’t personally know taught me to lean on research before I touch anything. That instinct — go learn before you go design — carried directly into my work at GoGood.