Naver & Daum: AI Search Race Heats Up with Agent Systems

Naver and Daum are stepping up their competition in AI-powered search. Both companies are moving towards agent-driven systems that can perform real-world tasks. Naver recently launched its conversational AI search service, AI Tab, to all users. This service acts as a gateway for AI agents to handle tasks like reservations and shopping. Naver is using a product-native large language model, harness engineering, and multimodal AI for this strategy. The product-native model is designed for real-world services and can handle complex interactions across Naver's ecosystem. Harness engineering connects the AI to real-time data sources for up-to-date responses. Naver has also improved efficiency by using multiple smaller models, doubling response speed and cutting operating costs by up to threefold. They also use reinforcement learning to reduce errors by up to 30% compared to their previous model. Naver plans to integrate more features like AI Briefing and real estate services into AI Tab this quarter. Meanwhile, Daum is accelerating its AI search strategy with Upstage. They launched a beta AI Summary service last week, powered by their own large language model, Solar. This feature analyzes web documents and creates summarized search results with cited sources. This marks the first major integration of Upstage's.