This dataset contains experimental results and operational logs from the project ‘Automation of the planning phase of a construction project using AI agents’. The agent-based system was designed to automate the planning phase of construction projects. The system utilises a multi-agent system (MAS) architecture, coordinated via the n8n platform and based on large Gemini-class language models, to generate work breakdown structures (WBS) and schedules based on the critical path method (CPM). The research focuses on the transition from manual planning (typically requiring 480 minutes for a single-family home) to an automated, agent-based workflow, reducing planning time to approximately 16–46 minutes — representing a 96% increase in efficiency. Furthermore, operating costs (OPEX) have been reduced by 99.9%, falling from an estimated €500 to just €0.028 per project.