Sonara

What inspired you

We wanted to make meeting new people feel natural and stress-free. Traditional dating apps can feel like a marketplace, so we set out to create something that felt more like a conversation with a trusted friend—one that helps you discover genuine connections without the pressure.

What you learned

Building this project taught us how to combine cutting-edge AI services with a smooth user experience. We deepened our understanding of real-time voice processing, learned to optimize large-language-model prompts for speed and relevance, and gained hands-on experience managing authentication and storage in a production-ready environment.

What it does

Our app talks with you like a close friend—getting to know who you are and what you're looking for. Once it understands your personality, it showcases the similarities between you and other users. It can also help you plan a date based on when you want it and what you'd like to do. This helps you find more like-minded individuals more efficiently and takes the stress out of planning the date.

How you built your project

Under the hood, we use an opinionated stack for reliability and speed: Next.js 15 with Turbopack, Tailwind for design tokens and instant dark mode, and Supabase for auth and storage. We stream Gemini for quick text results and capture voice for richer context; we also write transcripts so voice interactions are accessible and reviewable. Our UI components are consistent, accessible, and responsive; the itinerary lands as readable, scannable blocks instead of a wall of text.

Challenges you faced

The biggest hurdle was integrating real-time voice capture with streaming text responses while keeping latency low. We also had to carefully design database schemas to handle both chat history and itinerary data without performance bottlenecks. Finally, tuning the LLM prompts to balance personality detection and concise output required many iterations and testing sessions.