Nexon Engineer Blazes New Trail in Game Data Analysis with AI

The Nexon Developers Conference (NDC) 2026, the largest knowledge-sharing conference in the domestic game industry, has been underway near Pangyo, Gyeonggi Province, since the 16th. On the 18th, Kim Seo-yeon, a data engineer from Nexon Korea's Data Integration Services team, unveiled 'AI Search,' a system designed to make game data analysis easier, during a session titled 'AI Search: Beyond Dashboards to Agents – A New Paradigm for Game Data Analysis.' This service moves away from complex existing methods, utilizing artificial intelligence (AI) to help game developers quickly and accurately find the information they need within data.

Engineer Kim pointed out that simply feeding raw data to AI has limitations in understanding business context. She also highlighted three structural limitations of dashboard-based analysis: lack of guidelines, complex development procedures, and content overload.

To overcome these limitations, Nexon developed its proprietary AI agent service, 'AI Search,' based on its internal data platform 'Mono Lake' and Snowflake environment. AI Search operates by understanding the user's query intent and routing the task to the appropriate agent. AI Search independently handles the planning of analyses, query writing, and the derivation and systematization of results. It demonstrated analytical capabilities including cohort analysis and competitor trend identification through linkage with unstructured text data, as well as suggestions for optimal update timing.

Nexon cited its long-established standard datasets and 'AI Ready' data refinement work – adding metadata to enable AI to intuitively utilize data – as key factors in enhancing AI Search's performance. The company is continuously optimizing prompts by adopting the 'Claude 3.5 Sonnet' model.

Engineer Kim stressed that general-purpose AI infrastructure alone is insufficient for differentiation. She emphasized that accumulating 'AI Ready' data imbued with a company's unique business context and developing agents will become differentiated assets in the future analytical platform market. Nexon currently monitors AI systems in parallel with its established dashboard systems, maintaining monitoring agents and human review processes.