Gemini Guess

• 

  Gemini Guess

• 

• 

• 

Inspiration

Inspired by how kids (and anyone) love guessing games, we wanted to explore how far AI can reason in open-ended challenges. Gemini 3’s deep, multi-turn reasoning made it perfect for testing curiosity and creativity.

What it does

Players think of any concept, object, or idea — or even say it aloud or show it visually. Gemini 3 tries to guess it through deep reasoning, while the app tracks rounds, scores, and lets users replay the AI’s thought process in text, voice, and video.

How we built it

• Built as a Next.js web app for quick hackathon demo
• Gemini 3 API powers the reasoning and multi-turn guessing
• Web APIs handle microphone and camera input for voice/video
• Game state stored in memory (or lightweight database) to track top rounds
• Replays rendered visually and with audio playback for judges

Challenges we ran into

• Designing prompts for unlimited concepts so Gemini guesses reasonably
• Handling voice & video input without latency
• Keeping the game fun but fair, so AI sometimes fails in interesting ways
• Managing multi-turn reasoning for replay in a concise format

Accomplishments that we're proud of

• Showcased Gemini 3’s reasoning depth and nuance
• Created a replayable game that visualizes AI thought process
• Enabled voice, video, and text inputs in a single app
• Made an interactive experience that works for anyone, not just kids

What we learned

• Gemini 3 can handle deep, multi-turn reasoning across open-ended challenges
• Multimodal input (voice/video) makes AI interaction more natural
• Designing prompts for AI is an art as well as a science
• Replaying AI thinking is as fun as playing the game itself

What's next for Gemini Guess

• Add leaderboards and global scoring
• Expand video/voice AI interaction, maybe live-stream challenges
• Optimize prompts for faster AI guesses without losing nuance
• Mobile-friendly version (React Native) for on-the-go play

Built With

• gemini
• next.js
• postgresql