AI-Powered Smart Email Assistant for Intelligent Email Management

Abstract: Email continues to be the primary channel of professional communication, yet the growing volume of daily messages makes manual triage, prioritization, and response drafting increasingly time-consuming and error-prone. This paper presents an AI-Powered Smart Email Assistant, a full-stack web application that securely connects to a user's Gmail account through OAuth 2.0 and presents the inbox through a custom dashboard. Large language models are integrated into the backend to classify every inbound email into a meaningful category, predict its priority level, and flag indicators of spam or phishing. The system further incorporates a Smart Reply generator and an AI Compose Assistant capable of drafting context-aware replies and new emails in a tone selected by the user. Additional productivity features include automatic meeting-summary and task-extraction modules with calendar synchronization, a natural-language inbox search capability, and scheduled reminders. The proposed architecture combines a React/JavaScript frontend, a REST-based backend, secure OAuth-based Gmail integration, and LLM-driven NLP pipelines. The system is validated through structured functional and integration testing to assess whether it meets the required specifications for accurate classification, reliable reply generation, and dependable third-party service integration.

Keywords: Gmail API, OAuth 2.0, Large Language Models (LLMs), Email Classification, Spam and Phishing Detection, Smart Reply, AI Compose, Natural Language Processing (NLP), Meeting Summarization, Task Extraction, Calendar Synchronization, Semantic Search.