EXTENDED ABSTRACT: MMaterial database has attracted significant attention from governments, research institutions, and scientists. However, these databases are primarily developed and maintained by governments or research institutions with substantial investment in computer professionals. Currently, promoting individual research teams to build their own material database remains challenging. This is primarily due to two conflicting issues. First, the traditional database systems are not user-friendly for non-computer professionals, while materials scientists often have limited computer skills. Second, the traditional database systems have difficulty dynamically defining and storing heterogeneous data, whereas materials scientific data exhibits dynamic heterogeneous characteristics.In 2020, we introduced the Materials Genome Engineering Database (MGED), which features a user-friendly data model and an intuitive data schema designer. This allows users to autonomously define new data models and dynamically submit data, effectively addressing the previously mentioned conflicts. MGED has been utilized to construct the National Materials Data Management and Service Platform (NMDMS) and has been adopted by numerous research teams.The powerful natural language processing capabilities of large language models (LLMs) are establishing “Chat” as the primary interaction mode in information systems. We are integrating LLMs into MGED, upgrading it to CHAT-MGED, enhancing the intelligence, efficiency, and userfriendliness.This presentation will review the development of material databases, explore application scenarios for LLMs within databases, introduce the features of MGED, and discuss the research progress of Chat-MGED.