Broad adoption, results not yet quantified
SK Telecom is extending the use of artificial intelligence across its entire workforce to turn widespread adoption into measurable productivity gains, as South Korea's largest mobile operator seeks growth in AI services and data centers beyond its traditional telecommunications business. According to what the company told the Korea Herald, more than 7,500 employees use AI agents every month, more than 4,000 have created one of their own, and about 1,500 agents are in regular use, employed to prepare reports, analyze mobile network data, and automate routine tasks.
The figures come after a company initiative launched in March to encourage employees to develop AI tools for their own work. SK Telecom is now seeking to bring these tools to the level of entire departments, a more demanding task that requires systems to access corporate databases, communicate with existing software, and handle sensitive information. Kim In-soo, head of the AI Board, explained to journalists that most of the problems that individuals can manage have been solved and that difficulties requiring organizational support are now emerging.
The technical data case and the confidentiality limit
A practical example concerns the large volumes of technical data accumulated in mobile network operations. Many had been designed for conventional software systems, making it difficult for AI agents to understand what the individual fields represent. In a project involving about 9,000 databases and 300,000 fields, the company used AI to generate explanations of technical labels: SK Telecom says the accuracy of the generated responses rose from about 40% to more than 80%.
Kim estimated that processing that data with external AI models would have cost between 100 million and 200 million won per year, or between $75,000 and $150,000; the company instead used its own models and its own computing infrastructure, but did not disclose the actual savings. The executive also described the confidentiality issue: one identifies a task in which AI would be highly effective and then discovers that the necessary data cannot be provided to the models. For this reason, SK Telecom developed a system that keeps sensitive records separate from the models, allowing agents to perform tasks such as managing employees' medical expense claims without accessing the original data.
Agent selection and the missing verification
The company is also assessing which agents justify their operating costs, rather than merely encouraging employees to create new ones. "Our goal is not simply to reduce the number of agents, but to produce results," Kim said. The strategic context was set out earlier this month by CEO Jung Jae-hun, who outlined plans to use AI productivity gains and redeploy employees toward growing businesses.
The most relevant and unverified point, however, remains: despite broad adoption, SK Telecom has not disclosed how much workplace AI initiatives have reduced costs or improved productivity at the company level. Asked about the financial impact, Kim said employees complete individual tasks more quickly but declined to provide specific figures. The case, reported by the Korea Herald, thus shows already extensive adoption and a series of technical and organizational problems that have been addressed, while proof of the overall economic benefit remains to be documented.
