Every minute without CPR reduces a person’s chance of survival by about 10%. According to the American Heart Association, more than 350,000 out-of-hospital cardiac arrests happen each year in the United States, yet survival rates remain below 12%. Bystander CPR can double or triple those odds, but panic, hesitation, and forgotten training often prevent people from acting.
Rescue was built to solve that problem. It is a smart glasses app for Mentra Live running on MentraOS that acts as a real-time CPR coach. The glasses can see what the user sees, speak instructions directly into their ear, and guide them step by step without distracting them from the patient.
How It Works
Scene safety and emergency call: A persistent reminder ensures users call for help.
Responsiveness and breathing check: Clear voice prompts guide initial assessment.
Chest compressions: A built-in 110 BPM metronome keeps rhythm, and voice prompts count compressions.
Hand placement analysis: A photo taken during compressions is analyzed locally with Gemini, which checks hand position and gives spoken feedback like “Move lower” or “Perfect placement.”
What We Learned
Designing for wearables required a new mindset. We focused on voice-first interaction, minimal UI, and short, calming instructions suited for high-stress situations. Using Gemini for analysis helped us deliver useful feedback without overloading the device.
Challenges and Future Work
We balanced helpful feedback with simplicity, optimized performance for limited hardware, and adapted the language for real emergencies. Next, we plan to explore video-based feedback, breathing detection, and AED guidance to make Rescue an even more complete emergency tool.
Built With
- bun
- fastapi
- geminiapi
- mentraos
- mentrasdk
- node.js
- pillow
- python
- react
- tailwindcss
- typescript

Log in or sign up for Devpost to join the conversation.