Korean internet giants move beyond chatbots toward task-executing AI agents. Heavy infrastructure spending signals a shift to AI as core platform logic
South Korea’s internet leaders Naver and Kakao are stepping up efforts to deploy agentic artificial intelligence, laying the groundwork for large-scale service rollouts planned for 2026. The shift reflects a broader industry move away from generative AI that simply answers questions, toward systems capable of planning and executing tasks on behalf of users.
Over the past year, both companies focused on embedding large language models across their core services to improve engagement and efficiency. With those foundations now largely in place, the focus is turning to AI agents that can operate across multiple services, understand user intent, and act proactively rather than reactively.
What “Agentic AI” Means in Practice
Agentic systems are designed to:
- Interpret high-level user intent
- Plan multi-step actions across different services
- Execute tasks without repeated user prompts
For platform companies like Naver and Kakao, this represents a structural change in how search, commerce, and messaging services operate, effectively turning AI into an orchestration layer rather than a standalone feature.
Naver: Linking Services Into a Single AI Workflow
Naver took an early lead in this transition with the launch of Agent N in November. The system connects services such as search, maps, calendars, reservations, and content into a unified AI-driven flow.
At the company’s Dan25 conference, Naver CEO Choi Soo-yeon said the goal is to reduce reliance on keyword-based search. Instead, users interact through conversation, allowing the AI to infer intent and complete tasks across Naver’s ecosystem.
The company views Agent N not as a single product, but as core infrastructure for future services.
Commercial Rollouts and Infrastructure Commitments
After a year of testing AI features within live services, Naver is preparing for broader deployments:
- Q1 2026: AI shopping agent
- Q2 2026: AI-powered search tab
- Later: Agent N for Business, targeting advertisers and merchants
To support agentic AI at scale, Naver has committed more than 1 trillion won (about $691 million) to GPU capacity and AI infrastructure. It is also operating testbeds that connect its headquarters in Seongnam with its Sejong data center, underscoring the compute demands of real-time AI agents.
Kakao: Messaging as the Entry Point for AI Agents
Kakao’s strategy centers on its flagship messaging platform, KakaoTalk. In October, the company introduced ChatGPT for Kakao through a partnership with OpenAI, giving users access to generative AI inside everyday conversations.
Industry observers view this as an interim step. Kakao plans to deploy its own agentic AI, known as Kanana, across KakaoTalk and related services in the first half of 2026. The company has said Kanana will focus on proactive task execution and service coordination, aligning with the broader shift toward AI agents that act on behalf of users.
Why This Matters for Korea’s Platform Economy
The parallel moves by Naver and Kakao highlight a new phase in Korea’s platform competition. Rather than competing only on content or interface design, companies are racing to make AI the core decision-making layer of their services.
Success will depend not just on model quality, but on infrastructure scale, service integration, and user trust. As agentic AI begins to handle bookings, purchases, and recommendations autonomously, the companies that control these systems may gain lasting advantages in user lock-in and data leverage.
With both firms targeting 2026 for major launches, the coming year is likely to determine which platform sets the standard for agentic AI in South Korea.
