Naver, Kakao to Launch Shopping AI Agents in First Half

Naver and Kakao have entered final preparations for shopping artificial intelligence agents slated for release in the first half of this year. Shopping AI agents are systems that autonomously perform product searches, comparisons, recommendations, and payments on behalf of users. As the two tech giants accelerate their launches, analysts predict "agentic commerce" will rapidly take hold in South Korea.

According to IT industry sources on May 10, Naver is preparing internal beta testing after unveiling its shopping AI agent launch plans during a conference call on May 6. The shopping AI agent works as follows: when a user requests "find me a body lotion with larger capacity and lighter texture than what I've been using," the AI searches products on external e-commerce platforms and even completes the purchase.

Naver plans to roll out individual AI agents for travel, finance, and other sectors following the shopping AI agent launch later this month, then integrate them into an "AI tab." The AI tab, scheduled for release in the first half, will be positioned alongside core functions such as mail, cafes, and news on Naver's PC main page and mobile app header.

Kakao is also continuously updating its language model "Kanana," which powers its AI agent service "Kanana in KakaoTalk," ahead of official launch. Kanana in KakaoTalk is an AI agent embedded in KakaoTalk that analyzes user conversations to provide gift recommendations and reservation suggestions. The company aims to release it in the first quarter after beta testing.

Kakao plans to integrate its pilot "AI Mate Shopping" and "AI Mate Local" features into Kanana in KakaoTalk. The company will also expand external commerce services linked to "ChatGPT for KakaoTalk," which launched in pilot last year. Details are expected during the conference call following its earnings announcement on May 12.

Industry observers see it as "only a matter of time" before Naver and Kakao's AI agents handle commerce decisions including payments and reservations. Conducting everything from product recommendations to payments on a single platform offers significant advantages, notably strengthening control over the commerce ecosystem through user lock-in effects.

In the United States, the trend of delegating product searches through payments to AI agents is even stronger. OpenAI introduced an "instant checkout" feature to ChatGPT in September last year, enabling purchases from external e-commerce vendors. Google joined this movement last month by announcing partnerships with major retailers including Walmart.

The focus on shopping for AI agent applications stems from its clear monetization model, analysts say. For platform companies to increase advertising and commerce revenue, recommending products and services with high purchase probability is crucial—an area where AI agents excel. Amazon CEO Andy Jassy said last year that "customers using Rufus [Amazon's shopping AI agent] are 60% more likely to make purchases than non-users."

However, experts emphasize that ensuring safety beforehand is critical given AI agents' extensive capabilities. Ahn Sung-won, director at the Software Policy Research Institute, said: "As seen in recent safety controversies surrounding open-source AI agent 'OpenClaw,' personal data theft could become problematic when AI agent functions expand to include payments. Services must launch only after security is firmly guaranteed."