South Korea’s Naver and Kakao outlined 2026 roadmaps to turn their search and messaging ecosystems into agentic AI platforms that proactively execute tasks for users. Naver plans an Agent N-based AI shopping agent, search AI tab, and over 1 trillion won (about $691 million) in GPU and infrastructure investment, while Kakao prepares its Kanana-2 on-device agent for full integration into KakaoTalk and related services.
This article aggregates reporting from 1 news source. The TL;DR is AI-generated from original reporting. Race to AGI's analysis provides editorial context on implications for AGI development.
Naver and Kakao are effectively declaring that traditional keyword search and passive messaging are legacy modes. Their 2026 plans push hard into agentic AI: Naver with Agent N spanning shopping, maps, calendar, and business workflows, and Kakao with Kanana-2 embedded directly in KakaoTalk and search. This is an explicit bet that the winning consumer platforms will not just answer questions, but carry out multi-step tasks end-to-end—booking, buying, and coordinating across services with minimal user friction. ([koreatimes.co.kr](https://www.koreatimes.co.kr/amp/business/tech-science/20260102/naver-kakao-gear-up-for-agentic-ai-era-in-2026))
Strategically, this tightens the competition with global foundation model providers. Kakao’s partnership with OpenAI gives it frontier capabilities, but its real moats are on-device context, local data, and distribution to nearly 50 million domestic users. Naver is responding by spending over 1 trillion won on GPU and AI infrastructure, essentially building Korea’s own scaled agent stack in parallel to U.S. hyperscalers. For the race to AGI, the interesting shift is architectural: both companies are moving from single-model chatbots to orchestrated, tool-using agents that live inside popular apps. That will generate a massive stream of real-world interaction traces—exactly the kind of data that can later train more capable, generalist agents.

