Skip to content

SidM4/HackLondon26

Repository files navigation

Meridian

An AI-powered PropTech platform that leverages a massive dataset of UK planning decisions to help homeowners instantly predict renovation approvals and discover the most profitable ways to maximise their property's value.

Built at HackLondon 2026.


The Problem

When planning a home improvement, homeowners face two major risks:

  1. Uncertainty of council planning permission -- will my project actually get approved?
  2. Unknown ROI -- will the expensive work increase my property's value enough to justify it?

Our service solves both problems using big data and AI.

How It Works

1. Instant Approval Predictions

Users enter their property details and proposed project. Meridian instantly compares the proposal against every similar planning decision made by that specific local council, providing a highly accurate probability of approval backed by hard data and historical precedents.

2. ROI Maximisation & Value Discovery

We don't just tell you if your current plan will work -- we tell you what else you could be doing. The platform evaluates possibilities in your neighbourhood, revealing extra improvements you can make to maximise your property's value. It weighs estimated construction costs against the projected increase in property value, guiding you toward the most profitable decisions.

3. Data-Backed Confidence

Whether you want to build a glass extension, a dormer loft, or an outbuilding, every suggestion is anchored by real-world data: millions of historical records, council-specific tendencies, and instant precedent matching.

Key Stats

Metric Value
Planning Records 2M+
UK Councils Covered 400+
Prediction Accuracy 94%

Tech Stack

Frontend

  • Next.js 14 (App Router) with React 18 and TypeScript
  • Tailwind CSS for styling
  • Framer Motion for animations
  • Zustand for state management
  • Chart.js / react-chartjs-2 for data visualisation
  • jsPDF + html2canvas for PDF report export
  • Lucide React for icons

Backend

  • Node.js with TypeScript (executed via tsx)
  • Google Gemini -- multimodal AI with structured function-calling for analysis
  • Ibex Planning API -- UK planning application data
  • postcodes.io -- UK postcode geocoding (free, no auth)

AI Video

  • ElevenLabs -- AI-generated video content for the landing page

Project Structure

.
├── frontend/                  # Next.js web application
│   ├── app/
│   │   ├── page.tsx           # Landing page (hero, features, CTA)
│   │   ├── analyse/page.tsx   # Main analysis page (form + results)
│   │   └── test/page.tsx      # API diagnostics page
│   ├── components/
│   │   ├── input/             # PostcodeInput, PropertyDescriptionInput, WorkTypeSelector
│   │   ├── results/           # ApprovalDisplay, PrecedentsList, SuggestedImprovements, ReportDownload
│   │   ├── Nav.tsx
│   │   ├── Footer.tsx
│   │   ├── EvidenceModal.tsx
│   │   └── HouseBuildingLoader.tsx
│   ├── lib/
│   │   ├── api/client.ts      # Axios client with mock fallback
│   │   ├── types/api.ts       # TypeScript request/response contracts
│   │   ├── constants/workTypes.ts
│   │   └── mock/fixtures.ts   # Mock data for development/demos
│   └── store/
│       └── analysisStore.ts   # Zustand store
│
├── server.ts                  # REST API server (Node.js)
├── pipeline.ts                # Analysis pipeline orchestration
├── gemini.ts                  # Google Gemini AI client
├── ibex.ts                    # Ibex Planning API client
├── tools.ts                   # Gemini function-calling tool declarations
├── postcodes.ts               # UK postcode geocoding
└── secrets.ts                 # API keys (git-ignored)

Getting Started

Prerequisites

  • Node.js >= 18
  • npm or yarn
  • API keys for Google Gemini and Ibex Planning API (for backend)

1. Clone the repository

git clone https://github.com/SidM4/HackLondon26.git
cd HackLondon26

2. Run the frontend

cd frontend
npm install
npm run dev

Open http://localhost:3000. The app uses mock data by default so you can explore the full UI without a backend.

3. Run the backend (optional)

# From the project root
npm install

# Set your API keys
export GEMINI_API_KEY="your-gemini-api-key"
export IBEX_JWT_TOKEN="your-ibex-token"

npx tsx server.ts

The backend runs on port 3001.

4. Connect frontend to backend

Create a .env.local file in frontend/:

NEXT_PUBLIC_API_BASE_URL=http://localhost:3001

Restart the dev server for the changes to take effect.


Features

  • Approval Probability Gauge -- animated display showing your project's likelihood of council approval with a confidence interval
  • Historical Precedents -- browse similar planning applications from your area, with decisions, dates, and evidence details
  • Smart Suggestions -- AI-recommended alternative improvements with estimated costs, projected value uplift, and approval probability
  • PDF Report Export -- download a detailed report with all analysis results
  • 28 Work Types across 5 categories: Extensions & Conversions, External Works, Outbuildings & Structures, Change of Use, and Other
  • Responsive Design -- fully functional on mobile, tablet, and desktop
  • Mock Mode -- full demo experience without any backend setup

API Endpoints

Method Endpoint Description
GET /health Health check
POST /analyse Main analysis (postcode, property details, work type)
POST /predict_planning Planning prediction
POST /estimate_roi ROI estimation

Example Request

POST /analyse
{
  "postcode": "SW1A 1AA",
  "property_description": {
    "property_type": "Semi-detached",
    "bedrooms": 3,
    "bathrooms": 1,
    "other_details": "Victorian terrace with rear garden"
  },
  "work_type": "rear_extension"
}

Environment Variables

Variable Location Description
GEMINI_API_KEY Backend Google Gemini API key
IBEX_JWT_TOKEN Backend Ibex Planning API token
NEXT_PUBLIC_API_BASE_URL Frontend Backend API root URL (empty = mock mode)
NEXT_PUBLIC_USE_MOCK Frontend Force mock data (true / false)

Team

Built with caffeine and conviction at HackLondon 2026.


License

ISC

About

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors