[Edaily Reporter SHIN MIN JOON ] The era of AI-driven agentic commerce is dawning, where artificial intelligence (AI) assistants search for and compare products on behalf of consumers and even complete purchases.

With the emergence of AI agents such as Meta’s “Muse,” the agent-driven commerce market is expected to grow very rapidly. According to global market research firm Juniper Research, the global agent-driven commerce market (based on transaction volume) is projected to grow from $8 billion (approximately 11 trillion won) this year to $1.5 trillion (approximately 2,010 trillion won) by 2030.

With the advent of the agentic commerce market, consumers’ shopping habits are also expected to change significantly. Previously, it was common for consumers to access e-commerce platforms such as Naver, Coupang, and Amazon to search for products, compare prices and reviews, and then make a purchase. However, as AI agents become more widespread, consumers will only need to specify their desired conditions. For example, if a consumer requests, “Find a gift for my parents under 200,000 won and order the item with the fastest delivery,” the AI will compare △products △prices △reviews △delivery conditions from various retailers to recommend products and complete the purchase.

The advertising market is the most vulnerable to this shift. If the starting point for shopping moves from platforms to AI, changes in the existing e-commerce advertising market will be inevitable.

Kang Jeong-soo, head of the Blue Dot AI Research Center, predicted, “AI does not view ads and purchase products the way humans do,” adding, “If the proportion of purchases made by AI selecting necessary products on behalf of consumers increases, the existing e-commerce advertising market could shrink.” He continued, “In the future, competition in the advertising market will shift from vying for top positions in search results to competing to be selected in AI recommendation results through strategies such as Generative Engine Optimization (GEO).”

Industry analysts suggest that Amazon’s decision to block Muse’s access to its site is not unrelated to this trend. Amazon recently took measures to restrict Muse’s access to its website and app. Amazon’s decision is interpreted not as mere “closed-off behavior,” but as a calculated move to protect its own advertising revenue model. Amazon sells search traffic on its site as an advertising product, leveraging its base of hundreds of millions of shoppers worldwide. If Agent AI agents from outside the platform compare and recommend Amazon products, advertisers will no longer need to pay a premium to appear at the “top of search results.” Consequently, this increases the risk that control over search, click, and purchase data will shift to agent platforms like Meta, along with downward pressure on ad rates. This move is essentially a preemptive defense strategy.

This is also an important signal for domestic e-commerce and open-market platforms. Since companies like Naver and Coupang also monetize their search traffic through advertising revenue, if Agent AI comes to dominate external recommendation channels, it could destabilize existing search ad rates and traffic valuation. An official from one platform stated, “Amazon’s block is not a one-off issue but the beginning of a power struggle among platforms over ‘who will hold the power to make recommendations’ in the age of Agent AI.”

Amid this competition among platforms, experts advise that advertisers and brands must also rethink their strategies. Kim Hee-eun, Head of the New Tech Business Strategy Center at CheilWorldwide, said, “As consumers shift toward generative AI, the advertising industry is also paying close attention to GEO,” adding, “In the era of agentic commerce, we must integrate the entire process—from data analysis and understanding consumer queries to defining necessary information, selecting appropriate channels, creating content, and verifying AI exposure performance.” For example, if a smartphone manufacturer is launching a new product, the explanation goes, it should not simply manage the keyword “new product launch,” but must meticulously analyze search terms and prompt patterns to determine what questions consumers might ask about that product.