Gather

AI-powered activity recommendations for you and your group — based on your mood, budget, and time.

🌐 Live at gatherapp.live

---

DEMO https://github.com/user-attachments/assets/f60fc8bb-f62e-4fed-868f-0c3664de9061

What It Does

Tell Gather your mood, who you're with, how much time you have, and your budget. Gather searches real nearby places using Google Places and uses Claude AI to rank and explain the best options for you — right now.

---

Tech Stack

Layer	Technology
Frontend	React, deployed on Vercel
Backend	FastAPI (Python), deployed on AWS EC2
Web Server	Nginx as reverse proxy
Process Manager	systemd
Places Data	Google Places API (New) — Nearby Search
Geocoding	Google Geocoding API
AI Ranking	Anthropic Claude API (claude-sonnet-4-6)
Domain & SSL	gatherapp.live via Namecheap, HTTPS via Certbot

---

How It Works

User fills out form (mood, budget, time, location, group type)
        ↓
Browser geolocation OR Google Geocoding API → coordinates
        ↓
Google Places Nearby Search API → real places near the user
        ↓
Claude API → ranks top 10 places with match scores + explanations
        ↓
Results displayed with Google Maps links

RAG Pipeline

This app uses a Retrieval-Augmented Generation pattern:

• Retrieval — Google Places API fetches real, current nearby places based on mood-to-place-type mapping
• Augmentation — live place data (name, address, rating, hours, price level) is injected into the Claude prompt
• Generation — Claude ranks and explains recommendations grounded in actual place data, not general knowledge

---

Mood to Place Type Mapping

Each mood maps to specific Google Place types for targeted search:

MOOD_TO_TYPES = {
    "Adventurous": ["zoo", "park", "hiking_area", "go_karting_venue"],
    "Relaxed":     ["cafe", "spa", "movie_theater", "book_store", "museum"],
    "Social":      ["restaurant", "bar", "bowling_alley", "shopping_mall", "night_club"],
    "Hungry":      ["fast_food_restaurant", "restaurant", "food_court"],
    "Romantic":    ["art_gallery", "restaurant", "aquarium", "botanical_garden", "miniature_golf_course"],
    "Active":      ["gym", "swimming_pool", "tennis_court", "park"],
}

---

Architecture

gatherapp.live          → Vercel (React frontend)
api.gatherapp.live      → AWS EC2 t3.micro (FastAPI backend)
                           └── Nginx (port 80/443)
                               └── uvicorn (port 8000)
                                   └── FastAPI app
                                       ├── Google Places API
                                       ├── Google Geocoding API
                                       └── Anthropic Claude API

---

Running Locally

Backend

cd backend
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

Create a .env file:

GOOGLE_PLACES_API_KEY=your_key_here
ANTHROPIC_API_KEY=your_key_here

Start the server:

uvicorn main:app --reload

API runs at http://localhost:8000. Docs at http://localhost:8000/docs.

Frontend

cd frontend/gather-frontend
npm install
npm start

App runs at http://localhost:3000.

Make sure the fetch URL in App.js points to http://localhost:8000 for local development.

---

API Endpoints

Method	Endpoint	Description
POST	/recommendations	Get ranked activity recommendations
GET	/health	Health check

Request Body

{
  "location": "Vancouver, BC",
  "latitude": 49.2827,
  "longitude": -123.1207,
  "mood": "Relaxed",
  "group_type": "Friends",
  "indoor_outdoor": "Indoor",
  "budget": 30,
  "time_available": 2.0,
  "radius": 5000
}

Response

{
  "recommendations": [
    {
      "rank": 1,
      "name": "Breka Bakery & Café",
      "match_score": 92,
      "explanation": "This highly-rated café is open right now...",
      "open_now": true,
      "rating": 4.4
    }
  ]
}

---

Deployment

• Backend — AWS EC2 t3.micro (Ubuntu 24.04), managed by systemd, proxied by Nginx
• Frontend — Vercel, auto-deploys on push to main
• SSL — Let's Encrypt via Certbot, auto-renews

---

Environment Variables

Variable	Description
GOOGLE_PLACES_API_KEY	Google Cloud API key with Places + Geocoding enabled
ANTHROPIC_API_KEY	Anthropic API key

---

Built By

Sayyam Singla — Third year Computer Engineering, UBC
gatherapp.live