UrbanGreenAI

Inspiration

Growing up in Jaipur, I witnessed "Khatedar" land disputes where builders promised parks that became parking lots. This personal frustration collided with a staggering global statistic: 2.7 billion people—primarily in the Global South—have less than $1 , m^2$ of green space per capita, far below the WHO-recommended $9 , m^2$.

We call this "The Grey Squeeze"—1.5 million hectares of urban green space lost annually to concrete. The result? Children who've never walked to a park. Adults who drive 2km for groceries, generating 40% of urban emissions from trips under 3km (the "Short-Trip Paradox").

What it does

UrbanGreenAI enforces oxygen equity through a three-stakeholder loop:

Builders → Submit projects against AQI-linked green mandates. Fail? Instant rejection ("The Snap"). Citizens → Earn bounties as whistleblowers, reporting "ghost parks" with GPS + photo evidence. Government → Monitor ward-level dashboards and deploy AI-generated greening strategies. The compliance formula scales dynamically: $$G_{\text{required}} = G_{\text{base}} \times \left(1 + \frac{\text{AQI} - 100}{200}\right)$$

High-pollution wards demand more green space—no exceptions.

How we built it

Frontend: React 18 + TypeScript + Tailwind CSS + Framer Motion State: Zustand for global app state Backend: Lovable Cloud with edge functions for AI-powered recommendations Data: Ward-level datasets for Delhi and Jaipur (AQI, population, park access, green cover) AI Integration: Gemini models via Lovable AI Gateway for land recommendations and greening strategies Challenges we ran into Ghost Park Verification: How do you prove a park doesn't exist? We solved this with community validation—multiple independent reports + satellite imagery cross-referencing.

AQI Data Freshness: Real-time AQI fluctuates hourly. We use 2026 projections for stability while flagging wards exceeding thresholds.

Builder Gaming: Builders could technically plant trees, get approval, then remove them. Our solution: citizen audit bounties create perpetual accountability.

Accomplishments that we're proud of

"4-Minute Walk" Standard: Every recommendation ensures park access within 400m—breaking the short-trip paradox. The Snap Moment: Real-time project rejection when green mandates fail. No negotiations, no corruption loops. Three-Stakeholder Enforcement: A closed-loop system where each party audits the others—builders build, citizens verify, government strategizes. What we learned Policy ≠ Enforcement: India has green mandates on paper. UrbanGreenAI makes them executable. Incentive Alignment: Bounties turn citizens into infrastructure—free, distributed auditors with skin in the game. Data Granularity Matters: City-level stats hide ward-level crises. Our ward-by-ward approach exposes inequality.

What's next for UrbanGreenAI

Satellite Integration: Automated "ghost park" detection via NDVI vegetation analysis. Carbon Credit Marketplace: Let builders offset shortfalls by funding verified green projects elsewhere. Multi-City Expansion: Scale to 10+ Indian cities, then Lagos, Jakarta, São Paulo—the Global South megacities. Policy Pilot: Partner with municipal corporations to make UrbanGreenAI a mandated compliance tool, not optional software. "From Jaipur's disputed Khatedars to global oxygen equity—one ward at a time."

Built With

• framermotion
• gemini
• lovable
• react
• recharts
• tailwind
• typescript
• vite
• zod
• zustand