From Agile to Autonomous: The Role of AI Agents in Software Project Management

Abstract

Digital transformation and the development of Large Language Models (LLMs) are significantly impacting the way software projects are planned and managed, with agile methodologies and frameworks (e.g., Scrum, Kanban) still requiring a high level of manual human involvement in the planning, prioritization, risk assessment, and reporting processes. This paper aims to define the concept and framework of AIA-PM (AI Agent-Powered Project Management) and explore the possibilities of applying LLM agents in Autonomous Project Management (APM) processes, with a focus on the extent to which AI agents can take over management functions such as sprint planning, progress monitoring, risk analysis, decision-making and stakeholder communication. This paper will analyze which functions in agile project management are best suited for AI assistance or full autonomy, how LLM agents can contribute to the automation of planning, reporting, and risk assessment, and what the technical, ethical, and organizational challenges of introducing the AIA-PM approach are in real teams and industrial conditions. The results include a classification of functions suitable for the degree of autonomy, a presentation of potential benefits (efficiency, transparency, scalability) and limitations (ethics, security, accountability), as well as recommendations for the future development and application of such systems, with a proposal for a structured model/framework for practical implementation.