Where To?

Inspiration

Where To? started as a very personal frustration. We've all been in that group chat — someone suggests going out, everyone agrees in theory, and then 40 messages later nothing has happened and the vibe is already dead. We wanted to build something we'd actually use ourselves on a Friday night. That was the only brief we needed.

What it does

Where To? is a real-time, Jackbox-style multiplayer party game that turns group indecision into a competitive, comedic experience. Each player champions their pick — a restaurant, a film, anything — and defends it through 3 rounds of AI-generated, topic-aware questions. Everyone votes, a winner is crowned, and their pick becomes the group's actual plan. Plans finally leave the group chat.

How we built it

We built Where To? as a mobile-first multiplayer web app using React, TypeScript, and Socket.IO — every player joins on their own phone via a 4-letter room code, no download required. Claude Haiku powers the comedic question generation and the swipe-card preference matching for players who don't have a pick in mind. Google Places surfaces real nearby venues. We used Cursor and Claude Code throughout to build and iterate rapidly, and Figma Make to generate the initial wireframes. Both the AI and Places services are fully mockable, which meant we could demo reliably under any conditions.

Challenges we ran into

The biggest was a pivot. We originally started in the Health Impact track, but quickly realised we were designing solutions for clinical workflows and patient journeys we didn't fully understand. Pivoting mid-hackathon was a hard call but the right one. Once we moved to a problem we'd personally lived, everything clicked faster. Technically, getting real-time state sync reliable across multiple phones on the same room — including reconnects and host migration — took more careful thinking than expected.

Accomplishments that we're proud of

A fully playable, polished multiplayer game built in under 24 hours that works on any phone with no install. The AI question generation genuinely surprised us — the questions are funny, specific to each player's actual pick, and they create real debate. We're also proud of the bot system, which lets a single person demo the entire game flow solo without needing a group present.

What we learned

-Build what you know. Domain expertise matters more than a good-sounding idea. The healthcare pivot taught us this the hard way -Build things you'd actually use yourself. The best filter for any feature decision was simply: would we want this on a Friday night? -Scope ruthlessly. One tight user journey done well beats five half-finished features every time -Mock everything for demos. Reliable demos win hackathons as much as good ideas do

What's next for Where To?

The game mechanic works for any group decision — movies, holiday destinations, weekend activities. The immediate next step is expanding beyond restaurants and making the topic fully open-ended. Longer term, Where To? is a natural fit for Zymix's Mini App ecosystem — a one-tap experience launching directly from a group chat, no friction, bridging the gap between the group conversation and the real-world plan.