Korea’s AI Reality Check: Strategy Over Hype

AI in the Wild: Korea’s Giants Show Their Work

Korean tech’s biggest names just put real numbers behind their AI bets—and the results are messier and more instructive than any demo. At the Korean Marketing Association’s spring conference on May 23, executives from Naver, Kakao, and Woowa Brothers (Baedal Minjok) laid bare how AI is reshaping commerce, not through vague promises but through concrete conversion lifts, behavioral data, and deployment struggles. A parallel commentary from startup advisor Lee Bok-yeon, published on May 26, cuts through the noise from the other direction: founders are asking the wrong question when they wonder if they should “add AI.” Together, these two perspectives frame a critical pivot—Korea’s AI conversation is shifting from FOMO to functional accountability.

Naver, Kakao, Baemin: The Scoreboard Tells a Story

Naver’s shopping AI agent delivered the session’s headline figure: purchase conversion rates 85% higher than traditional search. The number is striking, but it’s the context that matters. Naver’s search behavior has undergone a structural change. Since 2023 and the mainstreaming of LLMs, 7-character-or-fewer keyword queries are falling while 15-character-plus queries have doubled in a year. Users aren’t typing “wireless earphones”; they’re describing a noisy commute and asking for ANC recommendations. Naver is betting its agentic commerce layer on that shift, moving from matching keywords to solving context-rich shopping problems inside a standalone app, Naver Plus Store.

Kakao’s play was narrower but no less telling. A simple copy tweak to a “public wishlist” marketing message boosted public registration from 50% to 90% —an Ai-powered nudge redesign, not a model revolution, that unlocked a social commerce flywheel.

“The thing that matters in marketing is whether AI affects performance,” said session chair Lee Dong-il of Sejong University. “Efficacy in performance-adjacent areas must be confirmed to gain momentum for company-wide diffusion.”

Baemin’s contribution was a war story: a multilingual international app that had languished for four years collapsed into solvability with an LLM. It’s the less glamorous truth of enterprise AI—boring integration work, not magical blue-sky demos, unclogging pipelines that stalled for years.

The Other Voice: “Should We Add AI?” Is a Trap

While Naver and Kakao presented performance metrics, a sharp external critique appeared on the same topic days later. In a column for Outstanding, startup mentor Lee Bok-yeon called out a pattern: founders are stuffing AI into pitch decks not because the business demands it, but because investors might expect it. The wrong question, he argues, is “Should we add AI?” The right one: “What is the most effective way to solve our users’ problem right now?”

Lee’s point lands hard against the backdrop of the marketing session. Kakao didn’t need a foundation model to get that 90% wishlist rate; it needed a behavioral hypothesis and a rapid experiment. Naver’s 85% conversion uplift matters because it ties AI to a core metric, not because the term “agent” is hot. Baemin didn’t sell a dream; it swapped a broken process for a working one. The column’s subtext is that AI deployment without a commercial nerve is just expensive cosplay.

The Missing Pieces and What Comes Next

Notice what’s absent from both reports: customer acquisition cost, retention impact, or margin effects. The public numbers are top-of-funnel—conversion, registration, query length. That’s not a criticism; it’s a signal of how early we are. Companies are still proving that the new interface works before tackling the harder question of whether it builds durable advantage. The next wave of AI performance data will need to show not just that people engage differently, but that they spend more, churn less, and cost less to serve.

For Korean commerce, the trajectory is clear. Search is becoming conversational. Agentic intermediaries are taking over recommendation and transaction execution. And the companies that win won’t be those with the flashiest AI label, but those that bury the technology so deeply inside a better customer experience that nobody asks what’s powering it.

References

• [커머스BN] 네이버·카카오·배민의 ‘쇼핑+AI’ - Byline Network
• “우리도 AI를 해야 하나?”는 잘못된 질문입니다 - 아웃스탠딩