Sparks

Discovery
Collections
Profile
Post
Post

✨ Sparks

Sparks is a feed-style leisure reading app built for curious minds who want to learn about the world instead of doom scrolling. It delivers hundreds of engaging, visually rich posts from the history of desserts to groundbreaking inventions. Through a personalized recommendation and discovery system, it surfaces content based on your interests while leaving room for surprise.

We were tired of mindless scrolling and empty engagement, so we brough back leisure reading to make you feel good about consuming content again.

🤖 Gemini 3.0

Google’s Gemini 3 was central to both building Sparks and creating its content. Gemini powers an asynchronous generation pipeline that produces multi-page, Markdown-formatted articles with titles, tags, and ordered page structures. Its long-context understanding ensures coherence across extended pieces, while its multimodal capabilities help generate or match distinctive cover imagery.

During development, Gemini supported code generation, system architecture, and structured schema design across our React Native/Expo frontend and backend services. Its reasoning and instruction-following capabilities shaped Sparks’ topic bucket system, reader architecture, and metadata-driven content model. We also used Google AI Studio to quickly build and iterate on our prototype before transitioning to Mobile.

By combining AI-assisted engineering with AI-generated content, Gemini defines both how Sparks is built and how it is experienced.