A smart home platform integrating an energy-aware digital twin and an LLM-enabled conversational agent

This paper presents the design, implementation, and evaluation of the Green Smart Home (GSH) platform, a novel end-user development solution that integrates an energy-aware digital twin with an LLM-enabled conversational agent. Motivated by the growing complexity of IoT ecosystems and the need for sustainability-oriented tools accessible to non-experts, the system enables users to create automations through natural language while receiving real-time feedback on conflicts, activation chains, and energy-related issues. The GSH introduces a unified platform that combines conflict detection, consumption forecasting, contextual explanations, and an intuitive visual representation of the home. An autoethnographic study of the system in six of the authors’ homes provided insights into long-term use, highlighting increased user awareness of energy consumption, active experimentation with automations, and the value of conversational interaction, while also revealing limitations. Findings confirm the GSH’s potential to enhance sustainability, usability, and user empowerment in smart home environments.