HeatShield AI is a climate resilience web app that creates polished, personalized heatwave response plans for vulnerable households. It is designed for social-good use cases such as family preparedness, caregiver triage, and community volunteer check-ins during dangerous heat events.

Live app: https://heatshield-ai-ten.vercel.app

Extreme heat is one of the deadliest climate risks, but public guidance is often too generic to help a real household make fast decisions. Risk depends on more than temperature alone. Housing type, cooling access, power reliability, water scarcity, symptoms, and nearby support all change what a family should do next.

HeatShield AI turns household context into a clear heat protocol. The app helps users build a practical response plan for the next few hours and the rest of the heat cycle.

The planner currently provides:

• household risk scoring
• immediate actions for the next 2 hours
• home cooling guidance
• household-specific care priorities
• emergency warning signs
• supply checklist
• community check-in script
• preparedness score
• hydration target
• check-in cadence
• day timeline for dawn, noon, and evening
• copyable volunteer brief
• scenario presets for common risk situations

This project targets a real climate-health problem that affects older adults, infants, pregnant people, outdoor workers, people with chronic illness, and households with poor ventilation or low access to cooling. The goal is to reduce confusion during heat stress and make decision-making calmer, faster, and safer.

• premium editorial interface with Instrument Serif typography
• no frontend API-key exposure
• deployable static frontend with serverless planning endpoint
• deterministic rules-based fallback when live AI is not configured
• optional AI-assisted plan generation using OPENAI_API_KEY

• Frontend: HTML, CSS, vanilla JavaScript
• Backend: Node-compatible serverless function in api/plan.js
• Runtime: local Node server in server.mjs
• Deployment: Vercel
• AI: optional OpenAI-compatible chat completion API

npm start

Open http://localhost:3000.

Optional:

OPENAI_API_KEY=your_key_here
OPENAI_MODEL=gpt-4.1-mini
PORT=3000

If OPENAI_API_KEY is missing, the app still works using the built-in planning engine.

api/plan.js     Serverless planning API
app.js          Frontend logic and rendering
index.html      Landing page and planner UI
styles.css      Premium visual system
server.mjs      Local dev server
README.md       Project overview

HeatShield AI addresses climate-health vulnerability during heatwaves. It helps users translate vague public advisories into concrete household action plans, especially where cooling, water, or support access is limited.

The app uses structured risk analysis to keep the system grounded and predictable, then layers AI-assisted generation on top when credentials are available. This avoids a black-box chatbot experience while still allowing more contextual language generation.

This project was planned, iterated, debugged, and refined using AI-assisted development, then deployed on Vercel for live judging.

1. Open the live app.
2. Choose a scenario preset or fill the form manually.
3. Generate a household heat protocol.
4. Show the risk level, preparedness score, hydration target, and timeline.
5. Show the volunteer brief and downloadable plan.
6. Explain that the app still works without live AI because of the fallback planning engine.

• live deployed URL
• public GitHub repository
• impact narrative
• AI stack disclosure
• short demo video