Google Storm 🌪️

Google Storm is a bilingual mobile/web platform that combines Gemini AI and the Google Maps Platform to guide communities through disasters and everyday challenges.

Think of it as Google Maps specialized for community resources: shelters, food banks, free clinics, and volunteer opportunities — all accessible through a conversational AI chatbot.

🌟 Inspiration

Living in Miami, one of the most hurricane-prone areas in the U.S., highlighted the urgent need for a tool that connects people to life-saving resources quickly.
While weather apps provide storm forecasts, they don’t tell you where to find shelter, food, or medical help. Google Storm was built to fill that gap.

🛠️ What it does

🧠 Gemini AI Chatbot → Conversational help in English or Spanish.
🗺️ Google Maps Integration → Real-time map of shelters, clinics, food banks.
🏚️ Emergency Mode → Quick toggle to show only open shelters with safe routes.
🤝 Community Contributions → NGOs and users can submit/update resources (moderated by AI).
🌤️ Weather & Hurricane Alerts → Integrated feeds from NOAA, AccuWeather, and NWS.

⚙️ How we built it

Frontend: Next.js + React + Tailwind (with Vercel/Firebase Hosting).
Backend: Firebase Functions & API routes in Next.js.
Database: Firebase Firestore for resources, pending approvals, and user data.
Authentication: NextAuth.js + Google OAuth.
AI: Gemini 1.5 Flash for intent detection, bilingual responses, and moderation.
Maps: Google Maps JavaScript API & Google Places API for locations.
Weather & Storm Data: OpenWeatherMap, AccuWeather, NOAA, and National Weather Service APIs.

Custom APIs implemented:

/api/places → Search resources (food banks, clinics, shelters).
/api/intent → AI-powered intent detection.
/api/resources → Manage community-contributed resources.
/api/hurricanes → Aggregate hurricane tracking data.
/api/weather → Fetch real-time weather alerts.

🚧 Challenges we ran into

🌐 API Overlap → Handling multiple data sources (Google Places, NOAA, AccuWeather) without duplication.
🕒 Time Crunch → Building a reliable MVP in under 48 hours.
🔐 Security → Safely exposing Maps API keys while proxying server requests.
🤖 AI Moderation → Getting Gemini to validate user submissions accurately.
📍 Data Availability → Some shelters and food banks aren’t in Google Places, so we had to seed them manually.

🏆 Accomplishments that we're proud of

Built a fully working bilingual AI chatbot + live map in <48 hours.
Created a community contribution flow with AI moderation.
Integrated multiple government and weather APIs into one seamless experience.
Designed an interface that is simple enough to use during a crisis.

📚 What we learned

How to combine AI + Maps + real-time data feeds into a meaningful, life-saving product.
Best practices for Firebase security rules and API key management.
Prompt engineering for multilingual AI intent detection and moderation.
The importance of clear UX under stressful conditions (big buttons, easy toggles, mobile-first).

🚀 What’s next for Google Storm

✅ Partner with NGOs & local governments to auto-update resources.
📲 Add push notifications for real-time storm alerts.
📡 Offline mode for when internet access is down.
📍 Expand beyond Miami to other disaster-prone regions worldwide.
🔄 Add volunteer matching and crowd-sourced population density tracking.

🔑 APIs & Tools Used

Google APIs: Maps JavaScript, Places, Geocoding
Weather APIs: OpenWeatherMap, AccuWeather, NWS, XWeather
Hurricane Tracking APIs: NOAA NHC, HURDAT, AccuWeather Tropical
Government Sources: NOAA advisories, NWS alerts
Authentication: NextAuth.js, Google OAuth
Database: Firebase Firestore + Firebase Admin SDK
Internal APIs: /api/places, /api/intent, /api/resources, /api/weather, /api/hurricanes