Nomi

• 

  Live agent

• 

  Scan

• 

  Check-in

• 

  Caregiver

• 

  Dashboard

• 

  Main

About the Project

Inspiration

We were inspired by how fragmented and reactive elder care still is today. Medication adherence, nutrition, mental health, and daily well-being are often tracked separately—or not at all—leading to missed warning signs and preventable health issues. We wanted to create a system that brings all of these signals together into one clear, actionable view for both families and healthcare providers.

What it does

Nomi is an AI-powered care coordination platform designed for older adults, caregivers, and providers.

It enables:

• Daily check-ins for mood, medication, hydration, and needs
• Smart prescription and product scanning to understand medications and nutrition
• AI-generated summaries and risk flags for proactive care
• Caregiver dashboards with simple, actionable insights
• Provider dashboards with adherence tracking, diet analysis, and care plan editing
• AI care plan chat to dynamically adjust patient care

How we built it

We built Nomi as a web-first platform using: Next.js + React for the frontend Tailwind CSS + shadcn/ui for a clean, clinical UI FastAPI (planned / mocked) for backend structure Dummy Data (planned / mocked) for data modeling Gemini API (integrated / scaffolded) for:

• prescription parsing
• journaling insights
• care plan editing
• recommendation generation

To stay within hackathon constraints, we focused on:

• realistic mock data
• modular UI components
• structured AI prompts for future scalability

Challenges we ran into

Scope vs. time: The vision is large, so we had to prioritize what would create the strongest demo rather than building full backend functionality Medical data complexity: Parsing prescriptions into structured, usable formats required careful prompt design and schema definition Balancing simplicity and depth: We needed to make the UI intuitive for older adults while still being powerful enough for providers Avoiding overengineering: We intentionally avoided real-time CV, full authentication, and complex models to focus on clarity and usability

What we learned

How to design AI-assisted healthcare workflows rather than just AI features The importance of structured outputs (JSON schemas) when working with LLMs How to simulate a production-grade system using mock data effectively That explainability and UX matter more than raw AI capability in healthcare tools

What’s next

Real Gemini API integration for live parsing and insights Persistent patient data with Supabase Real-time caregiver/provider notifications More advanced risk modeling and trend detection Mobile optimization for older adult usability

Built With

• dummydata
• fastapi
• gemini
• next.js
• tailwindcss