[Edaily Reporters Lee So-Hyun, An Yu-ri, Shin Yeong-bin ] Artificial intelligence (AI) remembers users’ preferences to order products and helps them prepare for job interviews with a single phone call. In the military, it analyzes battlefield intelligence to support commanders’ decision-making. AI is evolving beyond simply answering questions to become a tool that performs actual tasks in everyday life and on the front lines of national defense.

At the “13th Edaily Global AI Forum (GAIF 2026)” held on the 12th at the JW Marriott Dongdaemun in Seoul, Kakao(035720), SKTelecom(017670), and Naver Cloud presented their AI Transformation (AX) strategies leveraging their respective strengths. Although the areas of application differed, there was a consensus that AI must be linked to tangible results.

Kakao and SKT: Agent AI Entering Everyday Life

Song Jae-ha, Chief Technology Officer (CTO) of Kakao, introduced a hyper-personalized AI strategy that remembers users’ preferences and context and even carries out actual actions.

CTO Song explained that even if a user purchases children’s shoes, they might be a gift for a niece or nephew, and that judging preferences based solely on purchase history could lead to repeated misguided recommendations. He emphasized, “A purchase is a fact, but a preference is an interpretation,” adding, “If an AI remembers me, it must also remember that I can change.”

To achieve this, Kakao is pursuing “Project Mosaic,” which distinguishes between and remembers behavior and preferences separately. The company plans to integrate “Kanana Space”—which allows users to summon the AI from anywhere in KakaoTalk—with “AI Conductor,” which coordinates services, to connect users through shopping, reservations, and payments.

CTO Song emphasized, “Memory is the basis for judgment, but it does not grant the authority to act,” stressing that user approval is required for important payments. “AI for Everyone,” led by Kakao, involves 34 companies and institutions and aims to provide a personalized “one Agent AI per person” for each user.

Kim Ji-hoon, Executive Vice President and Head of the AI Business Division at SKTelecom, presented a strategy for AI closely integrated into daily life that anyone can easily use. He noted that the benchmarks for AI competition are shifting beyond model performance to include cost-effectiveness, technological sovereignty, and the ability to perform actual tasks.

SKTelecom is preparing a system that allows users to access AI agents via phone calls and text messages without having to install a separate app or log in. The goal is to improve accessibility so that even those unfamiliar with digital services can use AI simply by providing their phone number.

He also introduced the example of a “Job Coach” that finds job postings suited to users and assists them with everything from writing applications to mock interviews. The company plans to evolve this AI beyond one-time information searches to provide support all the way through to achieving users’ goals.

Naver Cloud: “Defense AI Must Be Controlled with Woori Technology, Incorporation”

Yoo Kyung-beom, Executive Vice President of Naver Cloud, presented an AX strategy for the defense sector utilizing sovereign AI. He explained that in areas directly linked to national security—such as military intelligence and operational systems—it is crucial to secure control not only over AI models but also over data and infrastructure.

AI can be utilized for battlefield intelligence analysis, decision-making support for commanders, drone and satellite image analysis, and the operation of surveillance systems. However, he pointed out that relying on external technology carries risks such as information leaks or service disruptions.

Accordingly, Executive Vice President Yoo emphasized the need for specialized models trained on internal military data and secure computing infrastructure. He explained that high-performance models are required at the command level for large-scale data analysis, while lightweight, fast-operating models are needed on the actual battlefield.

This requires graphics processing units (GPUs). He estimated that providing AI services to service members might require about 1,000 GPUs, while advanced closed-circuit television (CCTV) video analysis could require about 3,000.

Executive Director Yoo also called for public-private cooperation to address the military’s security constraints and computing resource shortages. He emphasized, “We need to combine civilian AI technology with military data to establish a proof-of-concept environment and build an ecosystem where defense companies and startups can participate together.”