TLDR: Researchers at KAIST, led by Professor Min-Soo Kim, have unveiled “Chimera,” a groundbreaking database system that seamlessly integrates relational and graph databases. This innovation significantly boosts query processing speeds by 4 to 280 times compared to existing systems, paving the way for truly intelligent AI agents capable of real-time, complex reasoning. The technology is set to be integrated into GraphAI Co., Ltd.’s “AkasicDB” for high-performance AI agents based on Retrieval-Augmented Generation (RAG).
A research team at the Korea Advanced Institute of Science and Technology (KAIST), under the leadership of Professor Min-Soo Kim, has announced the successful development of “Chimera,” a novel database system designed to overcome the limitations of traditional data management for advanced artificial intelligence (AI) applications. The announcement was made on September 8, 2025, highlighting a significant leap in database integration technology crucial for the realization of truly smart AI agents.
For years, companies have relied on relational databases (DB) for data management. However, the increasing complexity of data relationships and the demands of large AI models necessitate the integration of graph databases. This integration has historically presented challenges such as high costs, data inconsistency, and difficulties in processing intricate queries. Chimera addresses these issues by fully integrating relational and graph database functionalities into a single, efficient system.
The new system has demonstrated exceptional performance, processing queries at least 4 times and up to 280 times faster than conventional systems in international performance standard benchmarks. This remarkable speed is attributed to Chimera’s ability to efficiently execute graph-relational queries, leveraging the strengths of both database paradigms. Graph databases, unlike their relational counterparts, represent data as vertices (nodes) and edges (connections), offering a distinct advantage in analyzing and reasoning over complex, intertwined information such as people, events, places, and time. This feature makes them ideal for applications in AI agents, social networking, finance, and e-commerce.
Professor Min-Soo Kim emphasized the growing need for integrated technology that spans both graph and relational databases. “Chimera is a technology that fundamentally solves this problem, and we expect it to be widely used in various industries such as AI agents, finance, and e-commerce,” he stated.
The industrial impact of Chimera is expected to be immediate. It is poised to become a core technology for implementing “high-performance AI agents based on RAG (Retrieval-Augmented Generation),” a smart AI assistant with advanced search capabilities. The technology will be applied to “AkasicDB,” a vector-graph-relational DB system slated for release by GraphAI Co., Ltd., a startup founded by Professor Kim.
The research also comes at a time when a new standard language, “SQL/PGQ,” which extends the relational query language (SQL) with graph query functions, has been proposed to meet the rising demand for complex query processing between different database types.
Also Read:
- KAIST Unveils ‘Chimera’: A Breakthrough Database System to Power Smarter AI Agents
- RavenDB Unveils Database-Native AI Agent Creator, Revolutionizing Enterprise AI Deployment
The study was a collaborative effort, with Ph.D. student Geonho Lee from KAIST School of Computing serving as the first author, and Jeongho Park, an engineer at GraphAI Co., Ltd., as the second author. Professor Kim was the corresponding author. The Ministry of Science and ICT’s IITP SW Star Lab and the National Research Foundation of Korea’s Mid-Career Researcher Program provided support for this research.
