An LLM-Agent-Based Dynamic Planning and Monitoring System in the Educational Domain
Year: 2026
Type: article
Abstract: Large Language Models (LLMs) are enabling conversational agents capable of supporting increasingly complex workflows beyond traditional question-answering. In education, collaborative projects require continuous planning, coordination, and monitoring, yet existing educational chatbots mainly focus on tutoring or administrative support. This paper presents vProjTutor , an LLM-agent-based project management system that assists student teams throughout the complete project lifecycle. The proposed architecture integrates specialized LLM agents with external project-management tools, including Telegram, Trello, and GitHub, allowing users to configure projects, organize tasks, generate Sprint plans, monitor progress, and interpret repository activity through natural language interactions. A user study involving ten participants evaluated usability, task completion, and user satisfaction. Participants achieved an overall task completion rate of 93.3%, and the system obtained an overall BUS usability score of 85%. However, the system correctly interpreted 60% of commit messages, highlighting the need for further improvements in handling unstructured inputs. The results suggest that LLM-agent-based architectures constitute a promising approach for supporting collaborative projects in educational environments.
Authors: Pau Blanco, Inmaculada Rodríguez, Anna Puig
Institution: Universitat de Barcelona
Language: English
Subfield: Education
Field: Social Sciences
Domain: Social Sciences
SDG: Quality education
Open Access status: gold
Estimated APC paid: $2,490
Awards: G786884508, G9001382452
OpenAlex ID: W7166830234
