Know your ground before you build.

UCEY is a Canada-focused planning interface for finding underused urban land and turning it into housing opportunities. The app combines brownfields, parking lots, dead malls, rail corridors, and other candidate parcels into one map, then helps a planner inspect each site with scores, cost estimates, AI-generated memos, and optional text-to-speech.

This repository currently includes:

• authenticated entry flow
• interactive Canada map with site markers
• active / inactive filtering
• brownfield and non-brownfield candidate datasets
• right-side detail panel for site review
• Gemini report generation
• ElevenLabs audio playback

UCEY
├── Frontend
│   ├── Next.js 16
│   ├── React 19
│   ├── TypeScript
│   ├── Tailwind CSS
│   └── Leaflet
├── Backend
│   ├── Next.js API routes
│   ├── Node.js
│   ├── Zod
│   └── csv-parse
├── Data + storage
│   ├── Supabase
│   ├── Postgres / PostGIS
│   ├── FCSI datasets
│   └── OpenStreetMap / non-brownfield CSV assets
├── AI + media
│   ├── Gemini
│   └── ElevenLabs
└── Auth + testing
    ├── Auth0
    └── Playwright

The interface is organized around a single planner workflow:

1. Sign in from the Auth0-backed login screen.
2. Open the Canada map and choose a city preset or view all cities.
3. Filter the visible site inventory by land type, site status, viability, and minimum site size.
4. Click a marker to open the right-side panel.
5. Review scores, cost and capacity estimates, contamination context, and former land use.
6. Generate an AI planner memo and optionally synthesize it into audio.

Main UI elements:

When a site is selected, the side panel becomes the main inspection area. It is meant to answer a planner's first questions quickly: what is this site, how risky is it, how much housing could it support, and should we prioritize it?

The first card shows the site name, location, site type, status tag, and headline scores.

This section turns raw site data into quick planning estimates.

The Generate report button creates a short AI briefing note for the selected site. The memo is designed for planner review, not final engineering or legal approval. It summarizes:

• current site condition
• soil and contamination risk in plain language
• infrastructure context
• housing potential
• recommendation priority

The panel can turn the memo into spoken audio. This makes the selected site behave more like a voice-assisted planning brief.

The UI includes a save action in the panel for project-based workflows. This is the start of a shortlist workflow for planners, architects, and developers.

The score system is intentionally heuristic. It helps rank and compare sites quickly, but it is not a replacement for engineering due diligence.

Important nuance:

• non-brownfield entries use dataset-supplied soil and infrastructure fields
• FCSI brownfields use derived proxy scores rather than real geotechnical reports
• viability is a screening signal, not a permit-ready conclusion

The app currently merges multiple site sources into one map experience:

Raw data artifacts live under data/raw/, including:

• data/raw/fcsi/
• data/raw/non_brownfield/
• data/raw/all_candidate_sites_normalized.csv

• Canada-bounded map navigation
• city presets for Montreal, Ottawa, Toronto, and Vancouver
• merged site inventory from multiple sources
• active / inactive / all-site filtering
• land-type filtering
• site detail panel with status tag
• AI memo generation
• audio synthesis endpoint and in-panel playback
• login, signup, reset-password, and logout flow
• dark mode toggle

Potential future applications:

1. Municipal land triage Cities could rank underused parcels by readiness, contamination risk, and housing potential before starting formal planning studies.

2. Developer acquisition screening Developers could use the platform to shortlist underused parcels and compare redevelopment scenarios faster than manual GIS review.

3. Public sector funding prioritization Provincial or federal agencies could identify which brownfield or infill sites deserve grants, remediation support, or planning acceleration.

4. Scenario planning The platform could compare low-rise, mid-rise, and mixed-use outcomes across the same parcel and estimate tradeoffs in cost, timeline, and unit yield.

5. Portfolio monitoring A city or developer could track hundreds of candidate sites over time and get alerts when a parcel moves from active to inactive, or when priority conditions improve.

Near-term roadmap:

• show more site datasets directly in the production map flow
• expand coverage to more Canadian cities beyond the current core dataset
• improve field-level explanations for users unfamiliar with planning metrics
• strengthen report persistence and project-saving workflows
• add exportable PDF planner briefs
• expand AR / 3D presentation layers
• improve scoring transparency with clearer formulas and source attribution
• extend the model beyond Canada later so the platform can support other countries and eventually become usable worldwide

cp .env.example .env.local
npm install
npm run dev

Open http://localhost:3000.

npm run dev
npm run build
npm run db:seed:demo
npm run data:build
npm run data:integrate:non-brownfield
npm run ingest:brownfields
npm run db:verify

• Some scores are heuristic rather than engineering-grade measurements.
• FCSI brownfield scores are derived proxies, not formal geotechnical analysis.
• Source datasets can still contain parcels that need manual review for true redevelopment suitability.

Built for HackCanada 2026 by Daniel, Adhyan, Yeshi, and Bora.