AI-powered political promise tracker. We hold politicians accountable — automatically.

Pledgely uses AI to scrape, analyze, and audit political promises made by Canadian party leaders. Every promise gets a status, a completion percentage, and cited sources. No spin. Just facts.

• 📋 Promise Database — Browse promises by politician, party, topic, or status
• 🤖 AI Auditing — Backboard AI agent searches the web and evaluates each promise
• 📊 Live Dashboard — Real-time stats pulled directly from MongoDB
• 🎥 PDF Upload — Upload PDF news articles for AI extraction to update promise progress
• ⚖️ Compare — Side-by-side politician comparison

Pledgely automatically tracks and audits Canadian political promises using AI. Here’s the end-to-end flow:

1. Data Collection

   • seedFromWeb.js pulls the official platform text for major Canadian political leaders.
   • This includes party platforms, policy pages, and other verified sources.

2. AI-Powered Promise Extraction

   • The platform text is sent to a Backboard AI Assistant.
   • Backboard extracts concrete, forward-looking promises and categorizes each by topic (Housing, Economy, Healthcare, Climate, Education, Immigration).
   • Every promise starts with a status of Pending and 0% completion.

3. PDF Uploads & Gemini Extraction

   • Users can upload PDFs of news articles, press releases, or reports.
   • The Gemini AI service parses these PDFs and extracts any political promises automatically.
   • Extracted promises are added to MongoDB alongside their source file, ready for AI auditing.

4. Database Storage

   • Extracted promises, along with their source URLs or uploaded PDFs, and politician metadata, are saved into MongoDB.
   • Parties and politicians are automatically seeded if they aren’t already in the database.

5. AI Promise Evaluation

   • Backboard AI audits each promise in real-time.
   • Each promise receives:
     • status: Pending / In Progress / Fulfilled / Broken
     • completion_percentage: 0–100%
     • ai_reasoning: a short 2-sentence explanation with evidence
     • sources: links to news articles, PDFs, or official platform pages

6. News Verification

   • To ensure credibility, Pledgely checks NewsAPI and GNews for relevant, recent articles.
   • PDF uploads are included in the verification pipeline, ensuring only verifiable information is used.

7. Frontend Display

   • The React frontend fetches all promise data from MongoDB live.
   • Users can browse promises by politician, party, topic, or status, and compare leaders side by side.

git clone https://github.com/Sarah06102/Pledgely.git
cd Pledgely

### 2. Install dependencies

Install backend dependencies:

```bash
npm install

Install frontend dependencies:

cd client
npm install

Add the following environment variables:

MONGO_URI=your_mongodb_connection_string
BACKBOARD_API_KEY=your_backboard_api_key
GNEWS_API_KEY=your_gnews_api_key
NEWS_API_KEY=your_news_api_key
GEMINI_API_KEY=your_gemini_api_key
REACT_APP_AUTH0_DOMAIN=your_auth0_domain
REACT_APP_AUTH0_CLIENT_ID=your_auth0_client_id

Add the following environment variables:

REACT_APP_AUTH0_DOMAIN=your_auth0_domain
REACT_APP_AUTH0_CLIENT_ID=your_auth0_client_id

Run this commands next to get AI-audited promises and data from the web:

node seedFromWeb.js

cd server
node server.js

cd client
npm start

The app will run locally at:

http://localhost:3001

MIT License