A single-page web application with three pre-loaded abandoned parcels. Each parcel shows AI-generated land reuse recommendations, two live real-time data signals, a heritage scoring boost, and a community health panel. A RAG chatbot lets anyone ask questions about any parcel in plain English.
REQ-01
✓ Build
Three Parcel Cards with Real Montgomery Data
Three clickable parcel cards — Heritage, IX Hub, Food Desert — each pre-loaded with real data: verified street address from Montgomery city surplus or GIS records, actual zoning code, actual census income figure, actual food desert status, actual landmark proximity distance. No invented numbers anywhere. When a card is clicked, the recommendation panel opens instantly with zero broken states or loading errors. Fallback data must be ready if any live API fails.
Real Verified Addresses
3 Parcels Only
Zero Broken States
REQ-02
✓ Build
AI Recommendation Engine — Top 3 Per Parcel
For each parcel the system produces exactly three ranked land reuse recommendations. Each recommendation must have: a name, a fit score (0–100) produced by the weighted math model, a 2–3 sentence plain-language explanation generated by AI using the score and real parcel data as input, and a development cost tier label (Quick Win <$500K / Mid-Term $500K–$5M / Major Project $5M+). Both layers — math scoring and AI narrative — must be working and connected. The AI receives the score as a fact and explains it; it does not calculate the score itself.
Math Score + AI Narrative
Explainable Output
Cost Tier on Every Card
REQ-03
✓ Build
Civil Rights Heritage Boost — Parcel A Only
For Parcel A (Heritage parcel near Rosa Parks Museum), the system calculates the Haversine distance from the parcel coordinates to the landmark coordinates (32.3769, -86.3120). If the distance is under 1 mile, a heritage boost percentage is applied to cultural and community-facing recommendation scores. The boost percentage and the landmark name plus distance in miles must be visibly displayed on the recommendation card. The boost must demonstrably change the ranking — a cultural use must appear ranked higher than it would without the boost. This is RISE's most unique feature and must fire correctly in the live demo.
Haversine Distance
Boost % Visible on Card
Ranking Must Change
REQ-04
✓ Build
Two Live Real-Time Signals Per Parcel
Each parcel card must show two live data signals fetched at page load — not hardcoded. Signal 1: upcoming events within 5 miles of the parcel coordinates from the Eventbrite API (event name, date, distance). Signal 2: status of at least one matched grant from the Grants.gov API (open / closed / opening soon, days remaining if open). Both APIs are free with no key required for basic use. If either API returns no results or errors, a pre-written fallback message must display silently — no error state, no blank panel, ever. Fallback JSON for all three parcels must be written and tested before demo day.
Eventbrite API
Grants.gov API
Silent Fallback Required
REQ-05
✓ Build
RAG Chatbot — Ask RISE Anything
A text input accepts free-text questions about any parcel and returns intelligent answers using a RAG pipeline seeded with the three parcel knowledge bases. The chatbot must correctly answer a minimum of 4 pre-planned questions per parcel — 12 total — that the presenter will ask during the demo. All 12 must be tested on the live deployment URL before demo day, not on localhost. The chatbot must handle unexpected questions with a coherent data-grounded fallback rather than hallucinating or returning an error message. Jolie's existing HOA RAG bot is the starting point — adapted with Montgomery parcel data by end of Day 1.
RAG — Not Plain ChatGPT
12 Questions Pre-Tested
Coherent Fallback Required
-
To start using this app you first need to clone the repository:
-
Then you will need to install all the required packages for the application. Run this command (if you're using npm):
npm i
-
Run the frontend with the following command in the folder:
npm run dev
Add a .env file to your root folder and add the following line if you are running the backend locally VITE_SERVER_URL=http://localhost:8000/