ElaraHealth

Inspiration

I wanted to make mental health support more personal and accessible. In-person care isn’t always an option. It can be costly, hard to access, or intimidating for those nervous about opening up face-to-face.

What it does

ElaraHealth helps users track their daily mood, sleep, and nutrition, while offering an AI-powered therapist to guide reflection and provide personalized wellness therapy.

How we built it

We used Next.js with Tailwind CSS and TypeScript for a smooth, friendly UI, Supabase for data storage and authentication, Vercel for hosting, and Google Gemini 2.0 Flash to power our AI therapist.

Challenges we ran into

One challenge we faced was creating a truly personal experience for users. We tackled this through prompt engineering and storing chat history. We fine-tuned prompts to avoid undesirable LLM outputs, like lists, abrupt conversation endings, or lack of follow-up questions, encouraging users to open up and feel validated. To enhance personalization, we stored each user’s chat history and used it as context, leveraging Gemini’s large 1 million token context window to keep conversations meaningful and connected.

Accomplishments that we're proud of

I'm proud of how we crafted the AI therapist through thoughtful prompt engineering, making it feel more personal, supportive, and engaging, not just a chatbot, but someone we can go to in times of need.

What we learned

I learned a lot about frontend development and how to integrate Gemini 2.0 Flash into real-world applications to create dynamic, responsive AI experiences.

What's next for ElaraHealth

Next, I plan to connect the AI to mood, sleep, and nutrition data to offer even more personalized support.

Built With

• gemini
• nextjs
• supabase
• tailwind
• typescript
• vercel