Inspiration
Every year, thousands of elderly people face silent emergencies — a fall, a medical episode, a moment of confusion — with no one around to help. Existing solutions all share the same fatal flaw: they require the elderly person to press a button, respond to a message, or ask for help.
But what happens when they can't?
We were inspired by a simple but brutal insight: the most dangerous emergencies are the ones that happen silently. Our grandparents in India live alone more and more as nuclear families grow. They forget to respond. They don't know how to use complicated apps. And by the time someone notices something is wrong, it's too late.
ShadowCare was born from one question: "What if the system didn't wait to be told?"
What it does
ShadowCare is a passive AI-powered safety system for elderly people. It monitors without requiring any action from the user.
- Passive inactivity detection — tracks whether the elder has interacted with reminders, the app, or responded to check-ins, without requiring them to do anything special
- Intelligent risk engine — a progressive escalation system: 1 missed reminder → watch; 2 missed + inactivity → voice check; 3 missed + no response → family alert
- "Are you okay?" voice check — the system speaks to the elder and uses Claude AI to interpret their response, even if it's confused or unclear
- Auto-escalating family alerts — family members receive real-time status updates via a clean dashboard showing Safe / Warning / High Risk
- Zero friction for the elder — no buttons to press, no forms to fill, no app to learn
How we built it
We built ShadowCare as a full-stack web application:
- Frontend: React.js with a clean, accessible UI designed for both elderly users (large text, simple layout) and family members (dashboard view)
- Backend & Real-time DB: Firebase Firestore for live status sync between the elder's device and the family dashboard
- AI Layer: Claude API (Anthropic) for natural language interpretation of voice responses — classifying replies as OK, confused, or no response
- Voice Interface: Web Speech API for speaking check-in prompts and capturing voice replies
- Authentication: Firebase Auth for secure family account management
Challenges we ran into
The biggest challenge was designing for someone who won't engage with the system. Most health apps assume an active, tech-savvy user. We had to flip that assumption entirely — building a system that works even when the user does nothing.
We also had to carefully calibrate the risk engine. Too sensitive and families get flooded with false alarms. Too slow and it defeats the purpose. We landed on a three-stage escalation model that balances both.
Accomplishments that we're proud of
- Built a working passive risk detection engine with real-time escalation
- Designed a UI that works for elderly users with zero tech experience
- Integrated Claude AI for nuanced voice response interpretation
- Created a clean family dashboard that shows exactly what matters: is my loved one safe right now?
What we learned
We learned that the best assistive technology is invisible. The moment you ask an elderly person to "use" something, you've already introduced friction. The future of elder care is systems that work silently in the background — not apps that demand attention.
We also learned how powerful Claude API is for interpreting ambiguous, real-world human responses — not just clean yes/no inputs.
What's next for ShadowCare
- Integration with wearable sensors for physical activity detection
- Support for Tamil, Hindi, Telugu and other Indian languages in voice check-ins
- Caregiver network: alert not just family but a trusted neighbor or local NGO volunteer
- Partnerships with Indian elder care NGOs and community health workers
- Offline mode for low-connectivity rural areas
ShadowCare — because emergencies don't announce themselves.
Built With
- claude
- css3
- firebase
- firestore
- html5
- javascript
- react.js
Log in or sign up for Devpost to join the conversation.