Crush Without Saying It: Project Story

About the Project

Crush Without Saying It is a playful, anonymous app that allows students to express romantic interest in one another without the risk of rejection. The app enables users to secretly select people they like, and if the feeling is mutual, it reveals the match. To make things even more fun, the AI-powered "Wingman" generates conversation starters after a match, helping users break the ice and start conversations confidently.

Inspiration

The inspiration behind this project came from the common experience of students feeling shy or apprehensive about expressing romantic interest. Social media and dating apps often rely on direct communication, which can be intimidating, especially for younger users. This app was created to allow students to interact in a fun, safe, and non-judgmental environment. We wanted to eliminate the fear of rejection and make the process of connecting more light-hearted and exciting.

What We Learned

Throughout this project, we learned how important it is to balance user privacy with engagement. While we wanted users to feel comfortable, we also had to ensure the app was intuitive and seamless. Additionally, integrating AI to generate conversation starters added a layer of complexity, as it required careful prompt design to ensure the outputs were natural, fun, and relevant to the users’ choices.

How We Built It

1. Frontend:
   The app was built using Next.js for fast rendering and a smooth user experience. The UI is designed with Tailwind CSS, creating a clean, mobile-first design.

2. Backend:
   The backend was built using simple API routes in Next.js to handle matching logic and to process the AI-generated messages. For quick prototyping, we used an in-memory database to store the match information.

3. AI Integration:
   We integrated the OpenAI API to generate conversation starters. The prompts were designed to give the AI context on how the user might want to start a conversation (e.g., funny, confident, or shy).

4. Matching Logic:
   The matching algorithm is simple: If User A likes User B and User B likes User A, the system triggers a match notification and unlocks the conversation starter feature.

Challenges We Faced

1. AI Prompt Tuning:
   A key challenge was optimizing the AI’s ability to generate contextually appropriate and fun conversation starters. The AI’s outputs often lacked the right tone, so we had to iterate and refine the prompts repeatedly.

2. Ensuring Privacy:
   We were careful not to expose users to public rejection. The match system was designed so that only mutual likes would result in visible matches. Non-matches remain completely anonymous, which helped protect user privacy and comfort.

3. Real-time Experience:
   Building a smooth, real-time experience where matches are revealed instantly, and users can start chatting immediately, required optimizations in both the backend (API calls) and the frontend (animations, transitions).

Accomplishments We’re Proud Of

• Instant Match Feedback: Users can feel the excitement of a match in real-time with smooth animations and a celebration screen (confetti, fun sounds).
• AI-Powered Conversation Starters: The AI-driven Wingman feature is a unique addition that sets the app apart. It provides users with several conversation options, each in a different tone, making it easy to choose a message that fits their personality.
• Fun, User-Friendly Design: The app is built with a mobile-first, playful design that appeals to Gen-Z students, making it easy to use and visually exciting.

What's Next for Crush Without Saying It

• Persistent User Accounts: Currently, the app is designed for quick use, but we plan to add user accounts to allow for saved matches and history.
• Expanded AI Features: We want to refine the AI to offer even more personalized conversation starters and fun icebreaker suggestions.
• Advanced Matching Algorithms: Future updates will include more nuanced matching logic based on shared interests or vibes.

Conclusion

Crush Without Saying It was built to be a fun, non-threatening way for students to express interest in each other, with a little help from AI. We’re excited to keep improving the app and making it more interactive, fun, and safe for users.

Built With

• gemini-a
• geminiai