Abstract: With the rapid development of artificial intelligence technologies such as large models and intelligent agents, and the utilization of large model decision-making to control smart home appliance scenarios, smart homes are evolving towards “scenario-based” and “personalized” directions through multi-agent decision-making technology. However, there are currently major issues such as complex smart home scenario configurations, fragmented communication protocols between devices, and deviations in understanding user intentions. To address these issues, a joint decision-making technology based on the large model decision-making framework is proposed, utilizing “one-click group formation” for intelligent agents. Through three innovative modules: MCP dynamic group formation, vertical domain large model interaction, and group definition-driven decision-making, dynamic generation and adaptive regulation of smart home scenarios are achieved. Experiments show that this technology is superior to traditional solutions in terms of cross-protocol communication delay, scenario adaptability accuracy, and user configuration efficiency, providing new ideas for the intelligent upgrade of smart home scenarios.

Key words: Smart home, One-click group creation, Vertical domain large model, Joint decision-making, Dynamic scene generation