Agentic AI for Enterprise IT Support

Practice-project II: Developing a multi-agent LLM system with LangGraph and RAG to automate enterprise IT support workflows.

This group project explored the use of Agentic AI and Large Language Models (LLMs) to automate common enterprise IT support tasks, including ticket triage, issue classification, and response generation. As IT support services face increasing demand due to expanding digital infrastructure and hybrid working environments, the project investigated how autonomous AI agents could assist with routine support workflows while maintaining reliable and accurate responses.

We developed a multi-agent system using LangGraph, integrating Retrieval-Augmented Generation (RAG) to provide context-aware responses from a knowledge base. The framework coordinated multiple specialised agents responsible for tasks such as ticket analysis, issue classification, information retrieval, and response generation, simulating a realistic IT support pipeline. We also evaluated the performance of fine-tuned LLMs and compared their effectiveness against a traditional machine learning baseline using XGBoost.

The project demonstrated the potential of agent-based LLM systems to support enterprise IT service management by automating repetitive tasks and improving workflow efficiency. At the same time, our evaluation highlighted important challenges, including maintaining response consistency and ensuring factual accuracy.