Track 6: Agentic AI · FortyGuard Hackathon '26 · Building the World's Temperature AI
Extreme heat is the deadliest weather hazard in Nevada, and children are among the most vulnerable. School administrators and parents need real-time, hyperlocal heat intelligence to decide whether outdoor activities — PE classes, recess, sports practice, walking home — are safe.
Children's bodies heat up 3-5x faster than adults due to their higher surface-area-to-mass ratio and lower sweat rates. This makes age-specific heat risk assessment critical for schools.
CoolSchools is an AI agent that autonomously monitors temperatures around Nevada schools using the FortyGuard Temperature API® (powered by NVIDIA Large Temperature Models). It:
- Scans multiple school zones by calling the FortyGuard API (async submit-and-poll pattern)
- Classifies heat risk using WHO heat-risk bands (Safe / Caution / Danger / Extreme)
- Calculates heat exhaustion risk for students with age-adjusted sensitivity
- Generates age-group-specific warnings for elementary (PK-5) and high school (9-12) students
- Reports the total number of students at risk and those at risk of heat exhaustion
The agent runs autonomously — once triggered, it handles the full multi-step pipeline (API calls → risk assessment → heat exhaustion analysis → alert generation → report) without human intervention.
- Endpoint: POST /v1/heatmap (granularity=100, filter_type=1)
- Pattern: Async submit-and-poll — submit a heatmap request with
polygon_aoi(GeoJSON), poll/v1/status/{activity_id}, extract temperature from GeoJSON features - Data: Average, min, and max temperature per school zone from FortyGuard's Large Temperature Models
- Coverage: Nevada (Las Vegas, Henderson, North Las Vegas, Reno), U.S. coverage
- Data lag handling: Automatically queries 2 days in the past (API has ~24h ingestion lag)
# 1. Clone or download this project
cd coolschools
# 2. Install dependencies
pip install -r requirements.txt
# 3. Set your API key
cp .env.example .env
# Edit .env and paste your FortyGuard API key
# 4. Run the app
streamlit run app.py
# 5. Open http://localhost:8501 in your browser