AI-Driven Smart Helpdesk System for Automated Ticket Classification and Support Workflow

Abstract

Modern organizations face significant challenges in managing large volumes of IT and service-related helpdesk re- quests. Traditional helpdesk systems rely heavily on manual cat- egorization, delayed assignment, and slow response times, which increases operational cost and reduces user satisfaction. This paper presents an AI-Driven Smart Helpdesk System designed to automate ticket creation, classification, routing, and resolution workflow. The system integrates a REST-based ticketing API, a role-based web application, and an AI analysis engine that generates ticket summaries, detects urgency, and recommends resolution actions. A hybrid approach is implemented using both keyword-based NLP classification and large language model (LLM) reasoning using Meta Llama 3.1 via NVIDIA API. The system uses Flask, SQLAlchemy, Redis caching, and SendGrid- based email notifications to ensure scalability and reliability. Experimental evaluation demonstrates improved ticket routing accuracy, reduced response time, and streamlined support operations.