App Review Intelligence Agent 🤖

An AI-powered product analytics tool that scrapes real user reviews from Google Play Store and App Store, runs ML models to classify and prioritize issues, and generates actionable business insights using a Groq LLM agent.

🔴 Live Demo

Frontend | Backend API

🎯 What it does

Type any app name → Agent scrapes 500+ real reviews → NLP + ML models analyze them → Groq AI generates a business intelligence report with charts

🧠 How it works

1. Data Collection — Scrapes 500+ most recent reviews from both Play Store and App Store using free APIs
2. NLP Analysis — TextBlob sentiment analysis classifies each review as positive, negative, or neutral
3. ML Classification — XGBoost model categorizes reviews into Bug Report, Feature Request, Ads Complaint, Pricing, Performance, Praise
4. Priority Scoring — Random Forest model scores each issue 0-100 based on urgency
5. Trend Detection — Logistic Regression predicts if app rating is improving or declining
6. AI Insight — Groq LLM agent synthesizes everything into an executive summary with recommendations

🛠️ Tech Stack

Backend

• FastAPI — REST API framework
• Groq — Free LLM API (llama-3.3-70b-versatile)
• HuggingFace Transformers — NLP models
• XGBoost + Scikit-learn — ML classification and regression
• Pandas + NumPy — Data processing
• Pydantic — JSON schema validation
• Tenacity — Retry logic with exponential backoff
• google-play-scraper — Play Store reviews
• TextBlob — Sentiment analysis

Frontend

• React + Vite — Frontend framework
• TailwindCSS — Styling
• Recharts — Interactive charts
• Axios — API calls
• Lucide React — Icons

📁 Project Structure

app-review-agent/ ├── backend/ │ ├── app/ │ │ ├── agent/ │ │ │ └── agent.py # Groq LLM agent │ │ ├── models/ │ │ │ └── classifier.py # ML models │ │ ├── tools/ │ │ │ ├── scraper.py # Review scraper │ │ │ └── sentiment.py # NLP sentiment │ │ └── main.py # FastAPI endpoints │ └── tests/ └── frontend/ └── src/ └── App.jsx # React dashboard

🚀 Run Locally

Backend

cd backend
python -m venv venv
venv\Scripts\activate
pip install -r requirements.txt
uvicorn app.main:app --reload

Frontend

cd frontend
npm install
npm run dev

Environment Variables

Create backend/.env:

📊 API Endpoints

Endpoint	Description
GET /analyze/{app_name}	Full analysis pipeline
GET /test-scraper/{app_name}	Test review scraping
GET /test-sentiment/{app_name}	Test sentiment analysis
GET /test-ml/{app_name}	Test ML models
GET /health	Health check

👩‍💻 Author

Niharika Chauhan — GitHub