[Why&Next] Korea Focuses on 'Search', China on 'Action'... Why Shopping AI Assistants Differ

As shopping platforms increasingly adopt artificial intelligence assistant (AI Agent) features to enhance shopping convenience, China's AI ecosystem—capable of managing the entire shopping process on behalf of users, from product recommendation to payment—is drawing significant attention.

On April 24, the Harvard Business Review (HBR) reported that the AI assistants embedded in Chinese shopping platforms have evolved beyond the product search stage and can now handle the entire process, including payment and delivery. The publication analyzed that major Chinese consumer platforms are leading the way in the agentic commerce competition. Meituan, a leading lifestyle platform in China, launched its AI assistant “Xiaomei” at the end of last year. This assistant automatically handles everything from product recommendations to reservations, payment, and delivery tracking on behalf of the user. Meanwhile, Chinese e-commerce giant Alibaba has been enabling its proprietary AI model, Qwen, to autonomously carry out tasks such as food orders and in-chat payments for users since the beginning of this year.

Korean platform companies are also making a full-fledged entry into the agentic commerce market. However, unlike Chinese AI assistants, which automate the entire shopping cycle from product recommendation to payment, Korean platforms remain at the stage of conversationally comparing and recommending products tailored to users. When a product keyword is entered, the shopping AI agent summarizes, compares, and analyzes reviews of the products, but the user is still required to complete the transaction themselves.

Since February, Naver has been offering a beta 1.0 service of its shopping AI agent on the “Naver Plus Store” AI shopping app, focusing on digital, living, and lifestyle categories. Naver plans to add features such as real-time shopping trend analysis, automatic recommendation of related products, and adding items to the shopping cart in the future. Kakao has also linked its AI agent “Kanana in KakaoTalk (KainTalk)” with the “Reserve” and “Gift” functions in KakaoTalk. Lotte Himart is testing its shopping agent “Harvey” as well, demonstrating that the distribution industry is joining the AI agent service trend.

China’s Integrated Infrastructure... Connecting Through the Power of AI Assistants

The differences in the capabilities of AI assistants arise from the underlying shopping infrastructure. Chinese shopping platforms such as Alibaba and Meituan operate as “super apps,” integrating shopping, payment, logistics, and other services.

In the case of Meituan, every step—from product recommendation to reservation, payment, and delivery tracking—can be completed within a single platform. Alibaba’s AI assistant Qwen also coordinates tasks across Alibaba’s various services, including Taobao (shopping), Alipay (payment), and Amap (mapping). Integrated payment systems like Alipay and WeChat Pay further enable the automation capabilities of AI assistants in daily life. This means that the user interface (UI) and user experience (UX) are fully equipped for AI to seamlessly manage all aspects of the shopping process.

In contrast, Korea faces challenges in real-time integration of user login status, payment information, and delivery systems. Industry sources explain that it is difficult to automatically synchronize and manage inventory for listed products, and privacy regulations make it hard to maintain automatic login status. The structure requires re-authentication at every payment stage, creating further limitations for payment automation.

Currently, Naver and Kakao have set agentic commerce as their ultimate goal and are focusing on product recommendation services. According to Naver, "We are envisioning an overall structure that automates everything from search to payment using our own services such as Naver Pay," adding, "We are concentrating on perfectly implementing the initial stages of the service, such as product recommendations and adding items to the cart." Kakao also clarified that, while its AI assistant cannot yet automate payments, it has recently updated the “KainTalk” program to enable payments within the app.

Driven by the belief that the competitive edge of agentic commerce lies in integrated infrastructure, Google and OpenAI are also focusing on securing shopping ecosystems where AI assistants can operate. Google is developing the Agent Payment Protocol (AP2) and collaborating with more than 60 organizations, including PayPal and Mastercard, to create a payment system in which AI assistants can conduct transactions autonomously. OpenAI has also developed the Agentic Commerce Protocol (ACP), a common language that enables AI to check products between online stores, manage users’ shopping carts, and complete transactions by interacting with merchant systems.

This content was produced with the assistance of AI translation services.