How We Won the AI Product Hackathon

Eighty teams. Forty-eight hours. Five thousand dollars. Here’s how we took first place.

The Problem

The challenge was open-ended: build something with AI that solves a real problem. Most teams went straight to ChatGPT wrappers. We took a different approach.

Our Strategy

Instead of starting with the technology, we started with the user. We spent the first 4 hours just interviewing people at the venue about their daily frustrations. The winning insight came from a medical student who said, “I spend more time formatting notes than learning from them.”

The Build

We built a real-time medical note structuring tool that:

• Transcribed lectures using Whisper
• Structured content into clinical formats using GPT-4
• Generated spaced-repetition flashcards automatically

What Made Us Win

The judges said it wasn’t just the tech — it was the demo. We had a live patient case walkthrough that showed real clinical utility. Product thinking > technical complexity.

Takeaways

1. Start with the user, not the API — everyone has access to the same models
2. Polish the demo — judges remember how it felt, not how it worked
3. Ship something complete — a narrow, finished product beats a broad prototype

The $5,000 went straight into our next project.