Autonomous AI Agents for Automated Data Analytics
Department of Artificial Intelligence and Data Science Ajeenkya DY Patil School of Engineering, Pune , India · India
Abstract
Traditional data analytics face challenges due to tedious SQL-based querying, reliance on manual data cleaning, and static dashboards, making the process time-consuming and user-intensive. Autonomous AI agents provide an alternative, wherein Large Language Models (LLMs) are used for analytics ranging from natural language questioning to planning, analytical coding, and task execution requiring less human input. This paper studies autonomous AI agents for automated data analytics, describing the key components of such an agent, including the underlying LLM core, planning, reasoning, memory, and tool usage modules. Additionally, this paper describes the applications of such an agent throughout the various stages of the data analytics life cycle, including collection and cleaning, exploratory data analysis, modeling, and reporting. Following a review and analysis of seven recent works on the topic, this study identifies five core areas for future research: error handling in data cleaning, autonomous decision-making, data security and privacy, multi-modal data processing, and enhanced analytical capabilities. Finally, this paper provides a list of requirements for developing such a system for practical deployment.
Keywords
How to Cite This Article
Rights
BODHIVRUKSHA JOURNAL OF DIVERSE DISCIPLINE