An LLM-agent-based project management system named vProjTutor achieved an overall task completion rate of 93.3 percent in a user study evaluating collaborative workflows. Designed to assist student teams throughout the complete project lifecycle, the platform uses specialized conversational agents to support continuous planning, coordination, and monitoring in educational environments.
Published in the International Journal of Computational Intelligence Systems, the study was conducted by researchers Pau Blanco, Inmaculada Rodríguez, and Anna Puig. The authors explained that while collaborative student projects require continuous planning, coordination, and monitoring, existing educational chatbots focus primarily on administrative assistance or direct tutoring. To address this gap, the researchers investigated how Large Language Models can enable conversational agents capable of supporting complex workflows beyond traditional question-answering.
The proposed vProjTutor architecture integrates specialized LLM agents directly with external project-management tools, including Telegram, Trello, and GitHub. Through natural language interactions, this framework allows student users to configure projects, organize tasks, generate Sprint plans, monitor team progress, and interpret repository activity. By linking conversational agents to these external platforms, the system enables students to conduct comprehensive project management activities through dialogue.
To evaluate system performance, the authors conducted a dedicated user study involving ten participants to assess usability, task completion, and user satisfaction. Participants using the system achieved an overall task completion rate of 93.3 percent across the evaluated project management activities. In addition to this high completion rate, the vProjTutor system obtained an overall BUS usability score of 85 percent from the participants.
However, the user study identified specific performance limitations when handling unstructured inputs from software repositories. The system correctly interpreted 60 percent of commit messages, highlighting the need for further improvements in processing unstructured developer data. Despite this limitation, the authors concluded that LLM-agent-based architectures constitute a promising approach for supporting collaborative projects and dynamic planning in educational environments.
Subject of Research: LLM-agent-based dynamic planning and monitoring system in educational project management
Article Title: An LLM-Agent-Based Dynamic Planning and Monitoring System in the Educational Domain
Article References: Blanco, P., Rodríguez, I., & Puig, A. (2026). An LLM-Agent-Based Dynamic Planning and Monitoring System in the Educational Domain. International Journal of Computational Intelligence Systems. https://doi.org/10.1007/s44196-026-01676-3
Image Credits: AI Generated
DOI: 10.1007/s44196-026-01676-3
Keywords: vProjTutor, large language models, LLM agents, educational technology, collaborative projects, project management, dynamic planning, progress monitoring, task coordination, Telegram integration, Trello integration, GitHub integration, Sprint planning, BUS usability score, task completion rate, commit message interpretation, unstructured inputs
Cite Scienmag News
Denise Maddox. (November 16, 2026). LLM Agent System Assists Student Project Management With High Task Completion. Scienmag. https://scienmag.com/llm-agent-system-assists-student-project-management-with-high-task-completion/
Denise Maddox. "LLM Agent System Assists Student Project Management With High Task Completion." Scienmag, 16 November 2026, https://scienmag.com/llm-agent-system-assists-student-project-management-with-high-task-completion/. Accessed 17 November 2026.
Denise Maddox. "LLM Agent System Assists Student Project Management With High Task Completion." Scienmag. November 16, 2026. https://scienmag.com/llm-agent-system-assists-student-project-management-with-high-task-completion/








