'Geum.Yong.Sa.' Service Where AI Explains Difficult Financial Terms Simply

The Kakao Pay team developed 'Geum.Yong.Sa.', an AI-based financial glossary service, to resolve the hurdles users face due to difficult financial terminology. This service goes beyond simply explaining terms—it also recommends related financial products, taking into account scalability toward a business model.

The key features consist of the following three stages:

• Simple: Provides short term explanations within 20 characters in a tooltip format.
• Detail: Provides easy-to-understand detailed explanations using examples and bullet points.
• Products: Recommends insurance product information related to the term, organized in a table format.

Technically, the service uses an Agent in Amazon Bedrock to classify requests, and has built a RAG (Retrieval-Augmented Generation) structure via Amazon Knowledge Base. The knowledge base is divided into static and dynamic data according to purpose, such as insurance terms and disease terms, to optimize response efficiency.

The overall architecture was designed centered on Amazon API Gateway and AWS Lambda, with each Lambda calling an agent in Amazon Bedrock to generate optimized responses tailored to user needs.