Twillion

Inspiration

We were inspired by Krillion, where players compete to come up with niche answers, and its multiplayer version, Puffer Party. We loved the idea that knowing something obscure could earn you bragging rights.

Twitch communities felt like a natural home for that experience. Every streamer has their own inside jokes, shared memories, and viewers who remember things everyone else has forgotten. We wanted to give those viewers a chance to show off their knowledge and give streamers another way to bring their community together.

What it does

Twillion brings audience games directly into Twitch through an interactive Extension. Streamers use a separate dashboard to set up and control games, while viewers participate through a compact interface over the broadcast on desktop, with a separate mobile view. Viewers type their answers privately and compete for points on a shared leaderboard.

The experience is designed around three modes:

• Niche: Answer a streamer-specific prompt with something fewer people think of. If the question asks which games a streamer has played, remembering an obscure playthrough could earn you more points.
• Hivemind: Predict the community’s most popular answer. Winning means knowing how the specific streamers audience thinks.
• Quote Completion: Fill in the missing word from something the streamer said. This tests who was paying attention.

Our current connected prototype supports Quote Completion from the dashboard through to viewer answers, scoring, and final standings. Niche and Hivemind have interface previews, with their backend gameplay still to come.

How we built it

We split the project into three parts: the streamer dashboard, the viewer Extension, and the backend that connects them.

We built the dashboard and desktop/mobile viewer interfaces with HTML, CSS, and JavaScript, using a pixel-inspired visual style. The dashboard lets streamers connect their Twitch account, control quote rounds, and monitor game progress.

Our backend uses Python and FastAPI, with Supabase PostgreSQL storing game state, questions, submissions, and scores. Twitch OAuth authenticates dashboard users, while Extension tokens identify viewers. The backend controls round deadlines and scoring, keeping submitted answers private and correct answers hidden until the reveal.

For Quote Completion, we built a caption intake pipeline and algorithmic question generation, along with spelling tolerance and answer aliases. We also created automated tests and browser fixtures to exercise full games without requiring a live broadcast. Real Twitch caption capture and installed Extension testing remain next steps.

Challenges we ran into

One challenge was keeping the dashboard, viewer interface, and backend synchronized. Everyone needs to see the same round and results, even if someone refreshes, reconnects, or submits near the deadline. We addressed this by keeping game state and timing on the backend and restoring that state when clients reconnect.

Twitch integration also introduced several moving parts: streamer authentication, viewer identity, Extension messaging, and conditional caption availability. We built a repeatable test environment so we could verify the game flow while continuing to work toward testing with real broadcasts.

We also encountered a database connectivity issue with Supabase’s direct connection. Moving to a backend-only HTTPS storage connection let us connect to the hosted database while keeping credentials server-side.

Accomplishments that we're proud of

We’re proud of connecting the core Quote Completion experience across the whole stack: a streamer can control a game from the dashboard, viewers can submit private answers, and the backend handles timed rounds, scores, and leaderboards.

We also gave Twillion a visual identity that feels playful and at home on Twitch. Building both desktop and mobile interfaces helped us think about how the game could fit into different viewing experiences.

Beyond the interface, we’re proud of the foundation we built: persistent progress, moderator permissions, reconnect support, and repeatable tests that cover complete game flows.

What's next for Twillion

We want to develop Twillion into something streamers can regularly use with their communities. Our next step is to work with a small group of streamers, help them set it up, and test it during real broadcasts.

Those sessions will help us validate the Twitch integration, understand which prompts create the best moments, and improve the setup and gameplay experience. We also plan to finish the backend gameplay for Niche and Hivemind so viewers can compete across all three modes.

Ultimately, we want Twillion to feel like a game built around each streamer’s community, where knowing the deep cuts earns you bragging rights.

Built With

• css3
• fastapi
• html5
• javascript
• postgresql
• python
• supabase
• twitch
• twitch-extensions