[Edaily Reporter Yun Junghoon ] SKTelecom(017670)is accelerating its enterprise-wide AI transformation (AX), which goes beyond individual employees’ use of artificial intelligence (AI) to transform the way the entire organization works. With a culture of AI adoption spreading—as evidenced by the fact that more than 4,000 employees have created their own AI agents within the company—the company now plans to link the convenience of individual work to improved organizational productivity and business performance.

#KTCorporation announced on the 1st that it is fully launching “AX Innovation 2.0,” with the core direction of “From the Individual to the Organization, From Convenience to Practicality.” While AX Innovation 1.0, implemented in the first half of this year, focused on expanding individual employees’ use of AI and the creation of agents, AX 2.0—currently underway in the second half—focuses on redesigning the organization’s work processes and collaboration structures to be AI-centric. In particular, SKT has set the goal of AX 2.0 to demonstrate tangible results—such as actual cost savings or new revenue—that go beyond the costs and burdens associated with AI adoption and workflow redesign.

According to SKT, the number of employees who have created Agent AI in-house has now exceeded 4,000, and approximately 1,500 agents are in regular use. With voluntary adoption by employees having reached a certain level, the company plans to expand these individual experiments into organization-wide changes.

At the “AX Status Briefing” held on the 30th at the SKTelecom Talent Development Center in Icheon, Kim In-soo, Team Leader of the AI Board at SKTelecom, emphasized, “Significantly streamlining back-office operations can lead to substantial cost reductions and increased speed,” adding, “After that, we must move toward new business innovation.”

To this end, SKT is launching three implementation programs. First, the “AX Enabler Task Force (TF)” provides company-wide support for issues—such as data access, system integration, and security—that are difficult for individual employees to resolve on their own. While improving AI training and support systems, the task force is also responsible for redesigning management infrastructure, including assigning employee ID numbers to AI agents and establishing an evaluation system.

The “AX Sandbox” is an experimental space where existing workflows are redesigned from the ground up with a focus on AI. In the first half of the year, a service development team tested a model in which a single team member worked alongside an Agent AI to handle everything from planning to development and verification. In the second half, the company is expanding this initiative company-wide to explore a structure where multiple team members collaborate with Agents AI. The goal is not simply to add AI tools to existing workflows, but to transform the sequence of tasks, roles, and decision-making processes.

The company has also introduced the “AX Catalyst” program, in which frontline employees lead the expansion of AX. A total of 10 members were selected as the first cohort of catalysts last September. The program supports these frontline employees—who are not AI experts or developers—by covering the costs of using AI models and facilitating technical and data integration so they can directly solve problems they encounter on the job. The structure is such that these employees share the agents they’ve applied to their work and their experiences with other organizations, and when they identify common obstacles—such as issues with data, permissions, systems, or security—the Enabler Task Force steps in to resolve them.

The company is also building the technological infrastructure to support organization-wide AX. A prime example is “Z.E.Z. (Zero Exposure Zone),” which prevents sensitive raw data from being directly exposed to AI. Under this system, raw data—such as personal information—is managed in a separate area, and only the information necessary for task processing—such as metadata, comparison tokens, and calculated values—is provided to the AI. The goal is to simultaneously ensure data protection and enable AI utilization.

“Data Desk” helps AI understand the meaning and business context of internal data. Previously, data was managed primarily for systems and developers, making it difficult for AI to grasp the meaning of information even when it found what it needed. By providing explanations and business context to the data, the system enhances the Agent AI’s ability to search for and utilize information. SKT explained that after applying Data Desk to the Network Center’s database, the AI agent’s accuracy rate in answering questions rose from approximately 40% to over 80%.

The company is also promoting “Work Map,” a system designed to organize workflows so that AI can understand them at both the individual and organizational levels. Based on existing work materials and objectives, the AI creates a workflow map and refines work procedures and collaborative relationships through dialogue with employees. Through this, the company plans to identify tasks that agents can handle and establish standard procedures for humans and Agent AI to work together.

SKT’s vision is to focus on improving operational efficiency and reducing costs during the initial phase of AX 2.0, and then expand into new business innovations. The company aims to translate these efforts into business results across all sectors, including telecommunications, data, and AI data centers.

Team Leader Kim stated, “This is a process of creating a work environment where employees and AI agents work together to break through the trough of the AX J-curve,” adding, “We will organically link a culture of voluntary AI adoption among employees with a company-wide support system to spread the results of AX innovation throughout the entire organization.”