The centre of gravity of Kakao's AI strategy is shifting from experiments with standalone applications to a KakaoTalk-centred approach. It is focusing on connecting AI and existing services inside KakaoTalk rather than through a separate AI app.

Next month's developer conference, if(kakao)26, is expected to be the venue where the company makes that strategy more concrete. If last year showed the potential of combining KakaoTalk and AI, this year the main topic has emerged as how AI will connect users with multiple services.

According to the industry on Sept. 28, Kakao will hold if(kakao)26 from Oct. 13 to 14 at the Kakao AI Campus in Yongin, Gyeonggi province. The theme is "Intelligence in every connection". On the first day, Kakao CEO Chung Shin-a (정신아) will deliver the keynote speech under the theme "Connection for everyone, intelligence for one person".

Standalone app experiment ends... AI's stage moves to KakaoTalk

Kakao will end the app and PC services for its standalone AI service Canana, introduced in beta last year, on Oct. 15. That is the day after if(kakao)26 ends.

Kakao will apply technical capabilities and user feedback gained from operating Canana to other AI services. The standalone Canana service will end, but Kakao will continue to operate other AI services, including "Canana in KakaoTalk".

A standalone AI app must compete directly with global AI services such as ChatGPT for users and usage time. Kakao, by contrast, already has KakaoTalk, where everyday communication takes place. Rather than securing users again through a separate app, expanding AI touchpoints within the existing platform makes it easier for Kakao to leverage its strengths.

Kakao's recent AI services are also designed around KakaoTalk.

"Canana in KakaoTalk" is available from the Friends tab. It offers "First Talk", which uses an on-device AI model to understand conversation context and suggest necessary information first.

It supports schedule briefings, information searches, product and place recommendations, and changes to KakaoTalk settings. It plans to advance it into agentic AI that identifies user intent, suggests products or services and links through to purchases.

A similar direction appears in other services. The "AI National Secretary" introduced by Kakao and the Ministry of the Interior and Safety allows users to search for and reserve public facilities or issue electronic certificates in KakaoTalk. Kakao Tools also links Kakao's internal services with external partner services, turning user requests into actual execution.

If AI's role expands from "conversation partner" to "service connector", KakaoTalk's role could also change. Instead of users searching for needed services themselves, AI would connect the needed functions based on conversation context.

From last year's "KakaoTalk plus AI" to this year's "connection"

That direction can also be seen in the changes to if(kakao).

At last year's if(kakao)25, Kakao put forward "Possibility becomes everyday" and highlighted the combination of KakaoTalk and AI by unveiling "Canana in KakaoTalk", its own AI model and PlayMCP.

This year, it is taking that a step further by focusing on linking user context across multiple services and agents.

Sessions on the first day of if(kakao)26 are scheduled to cover how AI understands user context, AI memory that links information across services, an AI assistant available in KakaoTalk, and connections between agents.

If last year was a stage of building the foundation for "how to use AI in KakaoTalk", this year can be seen as a move to a stage of binding KakaoTalk and multiple services into a single user flow through AI.

Ultimately, the key to Kakao's AI strategy lies not in how much it grows standalone AI services, but in how naturally it connects AI and existing services through KakaoTalk. If it links searches and recommendations to actual service use, KakaoTalk's role could expand beyond communication into an interface between AI and services.

An industry official said, "KakaoTalk's user base itself is a strength, but what matters commercially is whether AI can go beyond simple recommendations and lead to actions such as actual purchases or reservations."