Automate your email workflow with this intelligent agent system. Features include email ingestion, filtering, summarization, action item extraction, and customizable notifications via various channels. Powered by LangChain, LangGraph, and Gemini 1.5 Flash.

Email Filtering and Notification Agent

This project implements an intelligent email management system designed to filter, categorize, summarize, and notify users about important emails. It uses a modular architecture with several interacting agents to achieve this.

Architecture

The system is composed of the following agents:

• Email Ingestion Agent: Connects to email servers via APIs, retrieves new emails, and pre-processes them (e.g., removing HTML tags, decoding special characters).
• Filtering & Classification Agent: Classifies emails based on importance using machine learning models, filters spam/junk, and categorizes emails (e.g., work, personal, social).
• Summarization Agent: Generates concise summaries of important emails using NLP techniques.
• Notification Agent: Manages notifications based on user preferences, including channel selection (voice call, WhatsApp).

Technology Stack:

• Programming Language: Python
• NLP Libraries: Transformers
• LLM Frameworks: LangChain, LangGraph (for orchestrating agents, defining workflows, and managing LLM interactions)
• Large Language Model: Google Gemini 1.5 Flash
• Email Libraries: imaplib, email
• Notification Libraries/APIs: Twilio (WhatsApp), Blandai