Agentic AI for Cloud Troubleshooting: A Review of Multi Agent System for Automated Cloud Support
Kinjal A. Patel, Eshan Pandey, Inshu Misra, Deepti Surve · 2025
One of the growing approaches in the field of artificial intelligence is known as agentic AI. This term describes autonomous systems that are meant to pursue complicated goals with minimum interaction from humans. Agentic artificial intelligence displays flexibility, advanced decision-making capabilities, and self-sufficiency, which enables it to work dynamically in contexts that are constantly changing. This is in contrast to traditional artificial intelligence, which is dependent on inflexible instructions and tight oversight. An important step forward in artificial intelligence and contemporary software systems is represented by the development of agentic systems. This development is being pushed by the desire for vertical intelligence that is adapted to a variety of different sectors. Through their capacity for learning, flexibility, and interaction with dynamic settings, these systems improve the results of corporate operations. Large Language Model (LLM) agents, who constitute the cognitive backbone of modern intelligent systems, are at the vanguard of this transformation. They are the agents that are revolutionizing intelligent systems. The aim of this research is to create domain-specific agents to address cloud and SaaS troubleshooting concerns. A particular agent will be created for a designated cloud platform. Manage Personally Identifiable Information to hide data and improve user privacy. This review aims to discuss Agentic AI, its core components, and applications across industries. It also surveys the literature, explores solutions provided by Agentic AI for challenges related to cloud platform failures, examines LLMs as agentic workflows, analyzes the accuracy issues of large language models (LLMs), and presents the proposed methodology along with associated challenges.