[15p-S2_204-1~18] 23.1 Joint Session N "Informatics"

5:30 PM - 5:45 PM JST(8:30 AM - 8:45 AM UTC)

[15p-S2_204-15] Autonomous Multi-Agent AI System for Thermoelectric Materials Research and Knowledge System Construction

〇Yusuke Hashimoto¹, Takaaki Tomai¹ (1.Tohoku Univ.)

Keywords:

AI agent,Thermoelectric materials,RAG

Thermoelectric materials are expected to play a key role in applications such as waste heat recovery; however, the selection of optimal materials is hindered by the vastness of the exploration space. In this study, we develop a hierarchical multi-agent AI system based on large language models, enabling automated thermoelectric materials research using the Starrydata2 database and the construction of an expert knowledge infrastructure via RAG/GraphRAG. The system design and its effectiveness, incorporating a quality assurance framework, will be reported.