What happened
SK Telecom disclosed that more than 7,500 of its employees use AI agents monthly, with over 4,000 having built their own tools. Approximately 1,500 agents are in regular use for automating routine tasks and analyzing mobile network data. The company is now moving from individual tool usage to department-wide integration, addressing challenges such as data accessibility and security. SK Telecom reported that using its own AI models to explain technical data fields improved answer accuracy from 40% to over 80%, avoiding estimated external processing costs of $75,000 to $150,000 annually.
SK Telecom announced that over 7,500 employees use AI agents monthly, with more than 4,000 employees having developed their own AI tools. Approximately 1,500 of these agents are in regular use, assisting staff with report preparation, mobile network data analysis, and the automation of routine tasks. This follows a companywide initiative launched in March to encourage the development of AI tools for job-specific applications.
The company is currently transitioning from individual tool usage to broader departmental integration. This shift requires AI systems to access company databases, interact with existing software, and manage sensitive information. Kim In-soo, head of SK Telecom's AI Board, noted that while individual-level problems have been largely solved, more complex issues requiring company-wide support are now emerging.
A key technical challenge involves the company's vast technical data, much of which was designed for conventional software and lacks clear semantic labels for AI interpretation. In a project involving roughly 9,000 databases and 300,000 data fields, SK Telecom used AI to generate explanations for technical labels. The company reported that this process improved the accuracy of AI-generated answers to related questions from approximately 40 percent to more than 80 percent.
SK Telecom stated that processing this data through external AI models would cost between 100 million and 200 million won ($75,000-$150,000) annually. Instead, the company utilized its own AI models and computing infrastructure, though it did not disclose the exact savings achieved. The company also developed a system to keep sensitive records separate from AI models, allowing agents to process tasks like medical expense claims without direct access to original data.
Despite the high adoption rates, SK Telecom has not disclosed specific figures regarding cost reductions or companywide productivity improvements. When asked about financial impact, company representatives stated that employees are completing individual tasks faster but declined to provide specific quantitative metrics on overall efficiency gains.
Source details: koreaherald.com
Why it matters
This report provides concrete, quantified evidence of enterprise AI adoption at scale within a major telecommunications carrier. It highlights the practical transition from individual experimentation to organizational integration, revealing specific technical hurdles such as data field interpretation and sensitive data handling. The disclosed accuracy improvements and cost avoidance estimates offer a realistic for other enterprises evaluating the return on investment of internal AI initiatives. It demonstrates that while adoption is high, measurable companywide productivity gains remain unquantified, underscoring the gap between tool usage and tangible business outcomes.
The report offers a rare, detailed look into the operational realities of enterprise AI adoption, moving beyond high-level announcements to specific usage statistics and technical challenges. The figure of 7,500 monthly active users indicates a significant penetration rate within a large workforce, suggesting that AI is becoming a standard part of daily operations rather than a niche experiment.
The disclosed improvement in data interpretation accuracy, from 40% to over 80%, highlights a critical bottleneck in enterprise AI: the quality and structure of underlying data. This finding is broadly applicable to other industries with legacy data systems, providing a concrete example of how AI can be used to remediate data quality issues before deploying more complex agents.
The cost comparison between using external models and internal infrastructure provides a practical data point for CIOs and CTOs evaluating their AI strategies. While the absolute savings are not disclosed, the estimated external cost range helps contextualize the financial stakes involved in processing large volumes of proprietary data.
The distinction between individual task automation and department-wide integration is crucial for understanding the current state of AI maturity in enterprises. The report confirms that while individual productivity gains are evident, scaling these gains to the organizational level remains a significant technical and managerial challenge, particularly regarding data security and system interoperability.
Interactive Mechanism: How It Actually Works
Explore the underlying technology behind this development interactively.
crm_get_transaction(id='4092').An agent must create a draft calendar event for Tuesday at 2 p.m. Which evidence would establish the requested result?
What to watch next
Monitor SK Telecom's future disclosures on specific productivity metrics or cost savings resulting from its AI initiatives. Watch for the rollout of its department-wide AI integration projects and how it handles data security for sensitive corporate information. Observe whether other major telecom carriers in South Korea report similar adoption figures or face comparable technical challenges in integrating AI with legacy data systems.
Future earnings reports or investor presentations from SK Telecom may include specific KPIs related to AI-driven productivity, such as hours saved per employee or specific cost reductions in network operations. The absence of these figures in the current report suggests they are not yet standardized or ready for public disclosure.
The success of SK Telecom's department-wide integration projects will be a key indicator of whether the company can overcome the technical hurdles of data accessibility and security. Watch for announcements regarding new AI features that require cross-departmental data access.
The competitive landscape in South Korea's telecom sector may see similar disclosures from KT and LG Uplus. If SK Telecom's figures are an outlier, it may signal a strategic advantage in AI adoption; if they are typical, it suggests a broader industry trend toward internal AI tooling.
Regulatory developments regarding the use of AI for processing sensitive employee data, such as medical expenses, may impact how SK Telecom and other companies structure their AI systems. The company's current approach of keeping sensitive records separate from models may become a standard practice or face regulatory scrutiny.