📧 AI Email Intelligence Platform

An AI-powered full-stack email intelligence platform that transforms raw emails into structured insights.

The system automatically:

• Classifies email intent
• Calculates priority scores
• Generates summaries
• Extracts named entities
• Identifies action items and deadlines
• Generates context-aware reply suggestions
• Provides analytics through an interactive dashboard

Built with a React + Flask full-stack architecture and a classical NLP pipeline (spaCy, scikit-learn, TF-IDF, rule-based intent detection) rather than an LLM. See Why Classical NLP Instead of an LLM? for the reasoning.

---

Table of Contents

• Live Demo
• Project Status
• Features
• System Architecture
• Database Design
• Tech Stack
• Why Classical NLP Instead of an LLM?
• Project Structure
• Installation & Setup
• Deployment
• Environment Variables
• API Documentation
• Screenshots
• Engineering Highlights
• Future Improvements
• Author

---

🌐 Live Demo

Frontend: https://ai-email-intelligence-platform.vercel.app

Backend API: https://ai-email-intelligence-platform.onrender.com

Repository: https://github.com/kavya608/ai-email-intelligence-platform

Note: the Render free tier spins down after inactivity, so the first request after idling may take up to a minute to respond.

---

📌 Project Status

🚀 Deployed — Backend on Render, Frontend on Vercel

Implemented:

• Backend REST APIs
• NLP processing pipeline (classification, priority, summary, entities)
• React dashboard
• Email analytics
• Full-stack deployment (Render + Vercel)

Planned:

• Gmail/Outlook integration
• LLM-based enhancement
• Authentication and user accounts

---

✨ Features

Backend

✅ Email ingestion API ✅ Batch email processing ✅ NLP-based classification ✅ Priority scoring ✅ Spam detection ✅ Email summarization (TF-IDF) ✅ Named entity extraction (people, orgs, locations, dates, money) ✅ Action item & deadline detection ✅ Reply generation API ✅ Dashboard statistics API

Frontend

✅ React dashboard ✅ Email listing interface ✅ Search and filtering ✅ Sorting and pagination ✅ Email details view ✅ Analytics cards & charts ✅ Responsive UI

---

🏗️ Architecture

React Frontend (Vercel)
      │
      │  Axios REST API
      ▼
Flask Backend API (Render)
      │
      ▼
AI Processing Engine
      │
      ├─────────────┬─────────────┬─────────────┐
      ▼             ▼             ▼             ▼
  Intent        Priority      Summary        Entity
Classification   Scoring     Generation    Extraction
      │             │             │             │
      └─────────────┴─────────────┴─────────────┘
                          │
                          ▼
                   SQLite Database

---

🔄 Email Processing Workflow

1. User submits an email
2. Flask API receives the request
3. NLP pipeline processes the content:
   • Intent classification
   • Priority calculation
   • Summary generation
   • Entity extraction
   • Action detection
4. Results are stored in SQLite
5. React dashboard displays insights

---

🗄️ Database Design

Processed email intelligence is stored using SQLAlchemy ORM. Each email record includes:

• Sender information
• Subject and content
• Category
• Priority score
• NLP-generated summary
• Extracted entities
• Action items
• Deadlines
• Spam classification

---

🛠 Tech Stack

Frontend

• React
• Vite
• React Router
• Axios
• Recharts
• Lucide React
• CSS

Backend

• Python
• Flask
• Flask-SQLAlchemy
• Pydantic

AI / NLP

• spaCy
• Scikit-learn
• TF-IDF based summarization
• Rule-based classification engine

Database

• SQLite

Deployment

• Render (Backend)
• Vercel (Frontend)

---

🧠 Why Classical NLP Instead of an LLM?

The first version of this platform uses classical NLP techniques (spaCy, scikit-learn, TF-IDF, rule-based logic) instead of an external LLM API.

Reasons:

• Zero API dependency
• Faster inference
• Deterministic, reproducible results
• No usage cost
• Easier debugging and testing

The architecture is designed so an LLM-based understanding layer can be added later without a rewrite (see Future Improvements).

---

📁 Project Structure

AI_Email_Intelligence_Platform/
│
├── Backend/
│   ├── app/
│   │   ├── ai_engine.py
│   │   ├── models.py
│   │   ├── schemas.py
│   │   └── main.py
│   ├── instance/
│   │   └── email_intelligence.db
│   ├── requirements.txt
│   └── run.py
│
├── frontend/
│   ├── src/
│   │   ├── components/
│   │   │   ├── Sidebar.jsx
│   │   │   └── StatCard.jsx
│   │   ├── pages/
│   │   │   ├── Dashboard.jsx
│   │   │   ├── Emails.jsx
│   │   │   ├── EmailDetails.jsx
│   │   │   └── Analytics.jsx
│   │   ├── routes/
│   │   │   └── routes.jsx
│   │   ├── styles/
│   │   │   ├── Dashboard.css
│   │   │   ├── Emails.css
│   │   │   ├── EmailDetails.css
│   │   │   ├── Analytics.css
│   │   │   └── SideBar.css
│   │   ├── App.jsx
│   │   ├── main.jsx
│   │   └── index.css
│   ├── package.json
│   └── vite.config.js
│
├── screenshots/
│
└── README.md

---

⚙️ Installation & Setup

Run the project locally for development:

1. Clone the Repository

git clone https://github.com/kavya608/ai-email-intelligence-platform.git
cd ai-email-intelligence-platform

2. Backend Setup

Navigate to the backend directory:

cd Backend

Create a virtual environment:

python -m venv venv

Activate the environment (Windows):

venv\Scripts\activate

Install dependencies:

pip install -r requirements.txt

Run the Flask server:

python -m app.main

Backend runs on:

http://127.0.0.1:5000

3. Frontend Setup

Navigate to the frontend directory:

cd frontend

Install packages:

npm install

Start the React application:

npm run dev

Frontend runs on:

http://localhost:5173

---

🚀 Deployment

The live version of this project is deployed as follows:

Backend — Render

cd Backend
pip install -r requirements.txt
python -m app.main

• Hosted as a Web Service on Render
• Uses the requirements.txt for build and python -m app.main (or a configured start command) to run
• Note: free-tier Render services spin down when idle and cold-start on the next request

Frontend — Vercel

cd frontend
npm install
npm run build

• Hosted on Vercel
• Framework preset: Vite
• The frontend's Axios base URL is configured to point at the deployed Render backend API

---

🔐 Environment Variables

The project currently runs on SQLite by default (instance/email_intelligence.db).

Optional .env configuration inside Backend/:

DATABASE_URL=sqlite:///instance/email_intelligence.db

For the deployed frontend, configure the backend API base URL (e.g. in frontend/.env):

VITE_API_BASE_URL=https://ai-email-intelligence-platform.onrender.com

(the /emails path shown in the Live Demo link above is one specific endpoint — the frontend should point at the base URL without the path)

---

🔌 API Endpoints

Method	Endpoint	Description
POST	/emails/ingest	Process and store a single email
POST	/emails/batch-ingest	Process multiple emails together
GET	/emails/	Fetch emails (pagination, search, filtering, sorting)
GET	/emails/<id>	Fetch complete email intelligence for one email
DELETE	/emails/<id>	Delete an email
POST	/emails/reply	Generate a context-aware reply suggestion
GET	/dashboard/stats	Fetch dashboard analytics and statistics

Get Emails

GET /emails/?page=1&limit=10

Supports pagination, search, filtering, and sorting.

Get Email Details

GET /emails/<id>

Returns complete email intelligence: category, priority, summary, entities, and action items.

Example Response:

{
  "category": "Work",
  "priority": 8,
  "summary": "Meeting scheduled for project discussion",
  "entities": {
    "people": ["John"],
    "organizations": ["Google"]
  },
  "action_items": ["Confirm attendance", "Prepare project update"]
}

---

📸 Screenshots

Dashboard	Emails

Email Details

Analytics

---

🧩 Engineering Highlights

• Designed and deployed REST APIs using Flask, hosted on Render
• Implemented a modular AI processing pipeline (classification, priority scoring, summarization, entity extraction)
• Used SQLAlchemy ORM for database management
• Integrated the React frontend with the Flask backend via Axios, deployed on Vercel
• Implemented pagination and filtering for scalable email retrieval
• Built analytics endpoints for dashboard visualization
• Built reusable React components, routing structure, and API-driven frontend workflows using Axios

🌍 Future Improvements

• 🤖 LLM-based email understanding
• 📧 Gmail API integration
• 📬 Outlook integration
• 🔔 Smart notifications
• 🧠 Advanced AI reply generation
• 🔐 User authentication
• 📱 Mobile application

---

👩‍💻 Author

G M Kavya Frontend Developer | Full Stack Developer

GitHub: github.com/kavya608

---

⭐ If you find this project useful, consider giving it a star!