Generative AI for Software Project Management: A Systematic Review

Abstract

Fast progress in Generative Artificial Intelligence (GenAI) and Large Language Models (LLMs) after the introduction of ChatGPT to the public sector in late 2022 has brought changes in the way knowledge-intensive tasks, such as software project management (SPM), are being conducted. The current study represents the systematic literature review (SLR) of 78 scientific papers, reports, and empirical studies published from 2019 to 2025 concerning the use of GenAI and LLMs as autonomous or semi-autonomous assistants during the SPM process. Based on the PRISMA 2020 guidelines, SLR consolidates evidence related to six main SPM stages: project planning and scheduling, risk identification and mitigation, requirements engineering, effort estimation, documentation generation, and stakeholder communication. As a result, a layered architecture of GenAI-assisted project management information system is proposed together with the introduction of GenAI Utility Model (GUM) that clarifies the connection between task complexity, human intervention, and utility of LLMs. Moreover, the comparison of five most advanced language models—GPT-4o, Claude 3.5 Sonnet, Gemini 1.5 Pro, LLaMA-3, and Microsoft Copilot—is provided based on the capability radar. The review concludes that LLMs show high efficiency in documenting, improving requirements, and assisting early project planning; however, there remain many challenges like hallucinations, lack of reasoning capability, inadequate understanding of the context of specific projects, and ethical issues especially when involved in crucial decision-making in regard to schedule and budgeting. Moreover, the research highlights eight main research gaps and suggests directions for future research in terms of both empirical and design-related studies. In general, the research provides very useful insights for software companies implementing GenAI solutions, developers of AI tools, and researchers engaged in software engineering and natural language processing domains