AI agent specialist Dalpa said on the 21st that it supplied an AI-based monitoring solution to the real-time chat-style community inside Kiwoom Securities' mobile trading system (MTS) "HeroS#."
Kiwoom Securities on the 16th introduced a chat community in HeroS# that allows investors to communicate in real time while viewing stock charts and quotes. Given the nature of financial services, large-scale traffic occurs simultaneously and the requirements for personal information security are high, making a real-time AI monitoring system essential.
Based on Kiwoom Securities' internal community operation guidelines, Dalpa built an AI monitoring system that analyzes user chat content in real time to determine whether to restrict it. It classifies chats that require restriction for community operation—such as profanity, the spread of false information, and expressions that trigger political or social conflict—by type, and is designed to identify abnormal messages in an average of within 1.5 seconds. After the first-stage AI auto-classification, a dedicated operator conducts second-stage review, ensuring both accuracy and operational stability.
In particular, given HeroS#'s characteristic of having a large monthly user base, with hundreds of chats per second occurring simultaneously in a high-traffic environment, it applied a real-time API architecture based on asynchronous processing and an auto-scaling infrastructure. Even when traffic concentrates, such as at the start of trading, it operates stably without service delays or interruptions, keeping the overall system error rate below 5%.
Through this, Kiwoom Securities has established a structure in which AI conducts first-stage monitoring of the massive volume of chats that is difficult for internal staff to review directly, and dedicated operations staff can conduct second-stage reviews under the same standards. The company noted that this both increases consistency in community operation standards and significantly improves operational efficiency.
Kim Do-gyun, CEO of Dalpa, said, "This project is a case that shows AI can operate stably in real time even in a financial environment that simultaneously demands high traffic and strict standards," adding, "By introducing AI monitoring, we supported operations staff to run the community more consistently and efficiently."