Kakao is expanding artificial intelligence (AI) commerce experiments inside KakaoTalk. It recommends people to send gifts to on the first screen of the Friends tab, finds places that match the context of a conversation and then links users to reservations. The aim is to keep demand discovered in KakaoTalk inside the app and bundle recommendations, search, reservations and payment into a single flow.

But simply having AI suggest friends to gift or places to visit has limits for monetisation. What Kakao must prove in the second half is transaction completion: users receiving recommendations in KakaoTalk, moving to actual purchases or reservations and linking through to payment. The market’s benchmark for evaluating Kakao’s AI strategy is also shifting from whether features are launched to whether they drive payment conversion and contribute to profitability.

From gift recommendations to place reservations

Kakao recently added a “Special Friend” feature at the top of the KakaoTalk Friends tab, increasing exposure for its gifting service. AI recommends suitable recipients for gifting by considering anniversaries, friend relationships and usage history. It also includes a function that identifies gift demand based on conversation context in KakaoTalk, usage patterns and gifting history. If a user sends a friend “Good luck on your exam,” it shows gift recommendations related to exam support.

Kakao has previously increased touchpoints for gifting, including expanding menu exposure for friends whose birthday it is. It is seen as an effort to strengthen commerce profitability. Last year, Kakao’s annual integrated commerce gross merchandise value was 10.6 trillion won, up 6 percent from a year earlier, and annual commerce revenue rose 7.8 percent to 951.0 billion won. Commerce revenue accounted for 11.7 percent of Kakao’s total revenue.

AI commerce does not stop at gifting. Kakao has also strengthened the place agent function of “Kanana,” an AI service within KakaoTalk. The key is expanding recommendations beyond restaurants to travel, culture and everyday convenience, and enabling everything from recommendations to exploration, sharing and reservations to be handled within KakaoTalk. Users receive recommendations tailored to conversation context for tourist sites, accommodation, exhibitions, performances, cinemas, gas stations, convenience stores and auto repair shops, and some can be booked directly within KakaoTalk. If users register a visit schedule for a specific area, the schedule briefing also provides information such as local restaurants, parking and menus.

Kakao is expanding these functions to turn KakaoTalk into the starting point for transactions. AI captures demand that already arises in KakaoTalk, such as a friend’s birthday, exam support or travel schedules, and connects it to products, places and reservations.

If this structure works, KakaoTalk becomes a demand discovery channel rather than a simple messaging app. If users discover needed products or places in KakaoTalk conversations before opening a search window and then complete reservations and payment right away, Kakao can expect not only ad exposure but also brokerage fees, payment fees and an increase in commerce gross merchandise value. Conversely, if actual payment after a recommendation happens in an external app, the revenue Kakao can secure is limited.

Results hold up, but payment remains

Kakao’s core business results are relatively solid. According to consensus from financial information firm FnGuide, Kakao’s second-quarter revenue is estimated at 2.0492 trillion won and operating profit at 226.3 billion won. That would be up 1 percent and 21.7 percent, respectively, from a year earlier. Revenue growth is not large, but operating profit growth is forecast to reach double digits.

The gap is linked not only to the improvement itself but also to the effect of streamlining affiliates. Kakao’s number of affiliates fell to 94 at the end of last year from 147 in May 2023. Even without a sharp rise in revenue, costs fell, resulting in a structure in which operating profit increases relatively more. Such cost cuts, however, are separate from the performance of the AI business itself.

The problem is that these results are not persuading the market. Ho-yoon Jung (정호윤), an analyst at Korea Investment & Securities, said the mainstream view has become to see the internet industry as low-growth domestic stocks as Kakao’s overall growth rate slows. Eun-jung Shin (신은정), an analyst at DB Financial Investment, also said cumulative users of “ChatGPT for Kakao” are estimated at about 11 million, but traffic indicators and external partner connections have not yet been activated. Both analyses suggest Kakao needs to show that agentic AI can lead to actual revenue to dispel low-growth concerns.

Kakao has already built experience linking external partners in some services. “ChatGPT for Kakao,” a chatbot-style service developed in collaboration with OpenAI, has completed links via KakaoTools with Olive Young, Musinsa and Saramin, but remains at the level of conversational information search and lookup. End-to-end commerce that extends to payment remains a task to be proven in “Kanana in KakaoTalk.”

Related services are expected to be unveiled soon. Seok-oh Kang (강석오), an analyst at Shinhan Investment Corp, said he expects Kakao to unveil an agent in early August that can handle search, recommendations and payment within KakaoTalk. The key, he said, is whether Kakao can bring into KakaoTalk commerce markets beyond existing gifting, including apparel, cosmetics and travel.

Still, brokerages are split on the potential of Kakao’s AI commerce. Dong-hwan Oh (오동환), an analyst at Samsung Securities, said both tracks, Kanana and ChatGPT, are delivering weak results. He said Kanana is designed in a way that does not fit the context of a messenger chat window, while “ChatGPT for Kakao” has failed to generate meaningful traffic due to a lack of differentiation from standalone apps. He pointed to the need to redesign from the consumer’s perspective rather than the supplier’s.

Chang-young Lee (이창영), an analyst at Yuanta Securities, interprets the same situation differently. He said Kakao is already equipped with a broad range of services including shopping, maps, pay, mobility, music and webtoons, and has about 50 million monthly active users (MAU). He said the “first-mover advantage” in which the side that retains users by offering diverse experiences dominates the market can be enjoyed by Kakao from the most advantageous position. Views are split between those pointing to a lack of completeness and those focusing on existing assets.

The competitiveness of Kakao’s AI commerce lies not in the number of products but in the context generated within KakaoTalk. Conversations with friends, schedules, relationships and gift history are data that general shopping malls find difficult to have. For this data to lead to actual transactions, recommendations must feel natural, the payment process must be short and users must not feel advertising intervention. Kakao’s challenge is closer to user experience than technology.

Yu-jin Seon (선유진), an analyst at LS Securities, said: “For the service to fit within the frame of AI monetisation, it needs to meet tasks including optimised targeting, a user experience differentiated from existing commerce and an actual contribution to profitability.”

Ultimately, the key to Kakao’s AI commerce is not whether it makes good recommendations but whether it leads to payment. Special Friend and the place agent are only starting points for finding demand within KakaoTalk, and it will be assessed as reaching the monetisation stage only when external partner links and Kanana-based agent commerce lead to actual transaction conversion in the second half.