Enhans Starts Ontology-Based AI Analytics Pilot With IBK

The pilot project centers on structuring IBK's internal operational knowledge and data for use by AI agents, specifically through Enhans's AgentOS platform. This involves building a semantic layer to link business meaning and relationships to the bank's information-system data, applying ontology technology to organize business concepts. This foundational work is crucial for an AI system to accurately interpret and process complex, domain-specific queries, moving beyond the need for employees to handle database query commands or request materials from the IT department.

A core technical aspect of the pilot is the application of Text-to-SQL, which converts plain-language questions into the command language used to query databases. Enhans also plans to improve answer accuracy by reflecting IBK's specific business terminology and decision criteria in retrieval-augmented generation (RAG). Beyond simple data retrieval, the pilot will test the system's ability to analyze causes, draw out business implications, and present the reference data and reasoning path used in its answers.

The project, which runs through next month, will quantitatively assess results using a "golden set", a benchmark dataset prepared in advance for evaluation. Enhans CEO Lee Seung-hyun stated that in finance, building a semantic framework for data is important so AI can use domain knowledge like business terminology and decision criteria. The pilot aims to verify whether analysis results can be presented with supporting evidence, even on a bank's internal network.