Hyundai Motor's vision of conducting AI-powered consumer research in a virtual world remains a future plan. But the AI agent that will contribute to that AI simulation work is already in the office, working alongside humans. The way Korea's major conglomerates use AI agents has already moved beyond chatbot-level search and entered a stage where the agents perform specialized hands-on work directly.
According to the information technology (IT) industry on Tuesday, major Korean conglomerates are using AI agents in various fields including marketing, system maintenance, sales management, and human resources training. Until last year, the use of work AI agents was limited to "vibe coding" (writing code by entering natural-language commands) for developer roles or information search for non-developer roles. Now, however, conglomerates are building their own job-specific AI agents, and these agents are beginning to go beyond performing hands-on work on behalf of people.
The AI agent recently drawing attention as a capable employee within Hyundai Motor (005380) Group is "AIMI." AIMI's main specialty is marketing. It handles tasks ranging from simple work such as news clipping to labor-intensive work such as analyzing competitors' automotive products. In the case of competitor product analysis, it succeeded in cutting work that used to take a person more than 10 hours down to tens of minutes. Cars have the characteristic that the terms used to indicate product specifications differ by brand and by the market in which they are released. In addition, the fragmentation of information across official websites, advertising videos, and product manuals is another factor that consumes time in competitor product analysis. But AIMI used a large language model (LLM) together with a vision language model (VLM) to extract essential information and created a standard comparison table for comparing specifications across vehicle models. AIMI also participates as a member of Hyundai Motor's virtual consumer marketing project.
At Samsung Electronics (005930), a security-specialized AI agent is protecting customer-facing IT services such as Samsung Pay and Galaxy Store. Samsung Electronics' Mobile eXperience (MX) division organized a task force (TF) last December and began seeking to build AI agents. One of the AI agents developed since then is the "Security Agent." Samsung Electronics' Security Agent monitors up to 250,000 daily access records of Samsung IT services and identifies abnormal access. Last August, the Security Agent accurately caught an attack attempt originating from India and even drafted a report the same day. Samsung Electronics has set a goal of deploying multiple AI agents to bring the rate of human intervention in system operation work below 20%.
At LG Electronics (066570), a sales management AI agent is clocking in. The main use of this AI agent is training customer service staff. When offline sales outlet employees are filmed on video serving customers, the AI agent analyzes the video, assigns evaluations, and notes points for improvement. A function will also be added going forward to record customer consultations and automatically identify and document the promises made to customers within those conversations. In addition, cosmetics company Amorepacific (090430) is entrusting the first review of contracts to an AI agent when signing advertising contracts with social network service (SNS) influencers, saving working staff's time.
While conglomerates' adoption of AI agents is increasing and cases of improved work efficiency are also growing, companies are not simply smiling. Behind the adoption of AI agents lies the concern of how to set the AI's authority to perform work and the question of responsibility after the fact. In response, companies are looking for ways to gauge the appropriate level of AI adoption for each task and to design organizational execution and responsibility structures to match. This is also the reason SK (034730) Group decided to manage its organization differently depending on whether AI agents can be used.
"For companies to catch both rabbits of productivity gains and compliance system management, they must distinguish between simple AI adoption and AX," said Lee Seung-hyun, AI evangelist at Liner. "AX is possible only when a company's business process reengineering (BPR) and AI strategy come together." He added, "Detailed standards for workflow innovation regarding the guarantee of AI autonomy and AI control by task must be established."
