CareCompanion

Inspiration

The inspiration for CareCompanion came from a simple observation: millions of seniors live alone, and for them, health tracking is often a cold, clinical process of filling out forms or pressing buttons on a screen. We wanted to humanize this experience. We asked: "What if care wasn't a chore, but a conversation?" Inspired by the loneliness crisis among the elderly, we built Clara—an AI companion that feels less like a bot and more like a family member who truly listens.

What it does

CareCompanion is an empathetic AI wellness platform for the elderly. It features Clara, a natural-voice AI that talks to the user about their day. While the user enjoys a friendly chat, Clara’s underlying engine (Gemini 1.5 Flash) silently analyzes the dialogue for critical health metrics like mood, sleep quality, pain levels, and social engagement. These insights are then synced to a Caregiver Dashboard, where families can monitor trends and receive "Red Flag" alerts for potential issues before they become emergencies.

How we built it

• Frontend: Built with Next.js 16 and React 19 for a fast, "10k" premium user experience with smooth glassmorphism and animations.
• AI Engine: Powered by Google Gemini 1.5 Flash for low-latency, high-empathy natural language processing.
• Database: Supabase handles cloud persistence for user profiles, session history, and real-time data sync.
• Voice: Integrated browser Web Speech API for localized, natural-sounding voice synthesis and recognition.

Challenges we ran into

• Latency in Empathy: Getting an AI to respond with genuine empathy in real-time was a hurdle. We moved to direct REST API calls to ensure Clara’s "thinking" time felt natural, not mechanical.
• Next.js 16 Migration: Working with the bleeding edge of Next.js 16 (Turbopack) required us to proactively migrate from traditional middleware to the new Proxy convention.
• Data Distillation: Training the engine to extract clinical "Red Flags" from casual conversation without being overly alarming was a complex balancing act in prompt engineering.

Accomplishments that we're proud of

• The "Wow" Factor: Building a UI that doesn't look like a "medical app" but feels like a premium digital companion.
• Invisible Tracking: Successfully extracting health data from speech without the user ever feeling "monitored."
• Cloud Resilience: Migrating from local storage to a robust Supabase backend in record time to ensure a production-ready demo.

What we learned

• Accessibility is Design: We learned that for seniors, "simple" isn't enough—the interface needs to feel "alive," which is why we prioritized voice and large-scale, high-contrast typography.
• The Power of Gemini: The speed and multimodal capabilities of 1.5 Flash were essential for maintaining the emotional flow of the conversation.

What's next for CareCompanion

• Proactive Calling: Implementing automated wellness check-in calls for when a user forgets to check in.
• Wearable Integration: Syncing Clara with Apple Health or Fitbit to combine her transcript insights with real heart-rate and step data.
• Multilingual Support: Expanding Clara to support elderly populations who speak languages other than English.

Built With

• axios
• github
• google-gemini-1.5-flash-api
• jspdf
• next.js-16
• postgresql
• react-19
• recharts
• supabase
• supabase-auth
• tailwind-css-4
• typescript
• vercel
• web-speech-api