SnapQuest

• 

  landing page

• 

  page after uploading image

• 

  3 games to choose from

Inspiration-Games are fun, but they’re usually static. We wanted to explore what happens when a game adapts itself to the player’s real world context. With the launch of the Gemini 3 family, we were inspired to build something that blends multimodal AI, creativity, and interactivity not just another chat interface. SnapQuest was born from the idea that any image could become a playable world.

SnapQuest turns a user uploaded image into an interactive mini-game experience. The app analyzes the image context, generates a themed scenario, selects or designs suitable game mechanics, and adapts visuals such as backgrounds and story elements dynamically. Each upload creates a unique “quest” that can be played instantly using only a mouse or keyboard no special hardware required.

How we built it-

SnapQuest is built with Flask for the backend and HTML/CSS/JavaScript for the frontend. We use the Gemini 3 Flash API as the core intelligence layer. Gemini 3 is responsible for multimodal reasoning, including interpreting image context, generating game scenarios, adapting difficulty parameters, and creating dynamic narrative content. To ensure stability and performance, game mechanics are implemented as safe, deterministic mini game engines, while Gemini drives creativity and decision-making. The system includes caching and graceful fallback logic to handle API limits reliably.

Challenges we ran into-The biggest challenge was working within API rate limits, especially for multimodal (image) requests. We also had to balance creativity with safety allowing Gemini to generate rich game logic without executing untrusted code. Designing an architecture that felt “AI-generated” while remaining stable and demo-ready required careful separation between AI reasoning and gameplay execution.

Accomplishments that we're proud of- Successfully integrated Gemini 3 as a central component, not just an add-on

Built a full end-to-end interactive experience in a short time frame Designed a hybrid system that gracefully adapts when AI services are limited Created a game that works for any user, on any laptop, using just a browser

What we learned- We learned that the most powerful way to use large models like Gemini isn’t to let them control everything, but to let them reason, design, and adapt, while traditional code ensures reliability. We also gained hands-on experience building multimodal AI systems under real-world constraints like quotas, latency, and user experience expectations.

What's next for SnapQuest

Built With

• flask
• gemini
• html
• python