Agentic AI home energy management system: A large language model framework for residential load scheduling

Abstract

The electricity sector transition requires increased residential demand response capacity, yet Home Energy Management Systems (HEMS) adoption remains limited by user interaction barriers. Large language models (LLMs) offer potential to address these barriers through natural language interaction; however, existing implementations employ LLMs as rule generators, optimizers, or parameter extractors rather than autonomous coordinators for multi-appliance scheduling. This paper presents an agentic AI HEMS where LLMs autonomously coordinate scheduling from natural language requests. A hierarchical architecture combining one orchestrator agent with three specialist agents is developed using the ReAct pattern, enabling dynamic coordination. Evaluation across three open-source models (Llama-3.3-70B, Qwen-3-32B, GPT-OSS-120B) using real Austrian day-ahead electricity prices reveals substantial capability differences for multi-appliance coordination. Llama-3.3-70B successfully coordinates all three appliances across all evaluation scenarios to match cost-optimal benchmarks computed via mixed-integer linear programming, while Qwen-3-32B and GPT-OSS-120B struggle to coordinate all appliances simultaneously despite achieving perfect single-appliance performance. The demonstrated system enables natural-language-based scheduling without technical parameter specification, offering a pathway to address configuration complexity barriers that currently limit residential HEMS adoption. All system components, including complete agent prompts, orchestration logic, and simulation user interfaces, are released as open source to enable reproducibility and further development.