As a network engineer, Heeyoung Oh’s mission is to ensure that services for Korean telco KT’s customers are not disrupted. If the network monitoring system flags a router issue in northern Seoul, she analyzes the situation and determines its cause, ideally in the first 10 minutes. This gives her time to decide how to fix it, before the fault becomes a major failure and complaints start flooding in about dropped calls or slow internet speeds.
Just investigating a single alarm used to take Oh some time, as she toggled between data housed in different systems. But since August, she’s been using AIO-Net, a continuous learning agentic AI platform trained by “Meisters”—network experts like her—and developed with forward deployed engineers from Microsoft. Now all that data can be viewed on a single plane, and AI agents take an average of just five minutes to identify and diagnose critical issues for her to make decisions.
“Because the analysis and root-cause identification have become faster, we can devote more attention and time to deciding how to respond,” says Oh. “Previously, the analysis took longer, which inevitably reduced the stability of the response and the operator’s ability to concentrate on it.
“Most importantly, AIO-Net significantly accelerates our assessment of whether customer services are being affected”.
Network reliability is front and center for KT, South Korea’s major telecommunications provider offering mobile, broadband, TV and telephone services and serving 29 million mobile subscribers.
“Network reliability takes precedence over operational efficiency,” says Seonil Jeong, KT’s head of network operations innovation.
Maintaining reliability is still largely a manual process. And with KT’s networks becoming more complex, its operators needed help with this manual load.
“As a wider variety of services have been introduced, analyzing alarms and warnings has also become increasingly time-consuming,” says Jeong. “We began considering AI as a way to make these processes more efficient, while enabling people to focus on critical decisions and actions.”
A co-development model
AIO-Net’s continuous learning agentic platform is a result of the close working ties between KT and Microsoft, with Microsoft engineers embedded in the telco’s network operations.
“One of the most significant aspects of this collaboration with Microsoft is that it has evolved into a co-development model, with Microsoft Frontier Company and the KT development team jointly defining problems in the field and developing solutions together,” says Jeong.
Microsoft’s forward-deployed engineers worked from KT’s offices alongside the development team, writing code and jointly building the processes for collecting, applying, and rolling back feedback. “AI development is constantly evolving, which makes it difficult to approach through traditional outsourcing,” says Hosong Lee, head of the KT AIO-Net development team. “The Microsoft team understood what we were trying to achieve and worked alongside us through implementation. They also brought deep experience in how best to use Azure.”
Learning from the Meister
For AIO-Net to analyze a network incident like an experienced operator, it was trained on the company’s standard operating procedures and by the Meisters themselves. Experts like Oh, for example, have intrinsic knowledge gained from over two decades in the field.
“Network operations involve a wide range of manuals and standards, but simply having access to more data does not necessarily make it easy to reach an accurate decision,” Jeong explains. “We believed that allowing AI to learn from the expertise of KT’s existing Meisters would help improve confidence in the system.”
The system, powered by Microsoft Azure, is not just a repository for KT’s institutional knowledge. It’s also designed to create a continuous learning loop, a process known as hill-climbing in AI, improving processes for both human and machine.
Every minute, its AI agents—built using multiple large language models in Microsoft Foundry—analyze anomalies in large volumes of real-time data generated by KT’s network equipment across the country. The agents’ conclusions and processes are all documented, enabling the network experts to evaluate them and hone their accuracy. The experts’ feedback informs how the system handles future cases, and new types of network incidents.
“Building a perfect AI system from the outset is extremely difficult,” says development team head Lee. “The system therefore needs to be designed to improve continuously.”
Meanwhile, operators who use AIO-Net are contributing to a feedback loop, just by using it.
“Previously, we had to request feedback, and typically one or two regional representatives would respond,” says Seonghyun Lee, KT’s development team manager. “The current system captures feedback naturally as people use it, even if we do not explicitly ask them to provide it. They simply need to continue using the system.”
The results so far? According to KT, it now takes 2 to 3 minutes for the system to detect a network issue, from over 10 minutes previously. On average, agents can investigate an incident and produce a report in five minutes, about half the time they used to take manually.
Human in the loop
“If AI helps us save time, our people can focus on areas they previously did not have sufficient capacity to address,” says Jeong. “Their work will shift toward higher-value areas such as telecommunications facility management and quality management.”
“AI’s authority extends to analysis and the presentation of supporting evidence. Decisions and actions remain with people,” he adds.
Strict limits are placed on what AIO-Net can do. It can flag incidents, query and inspect equipment as part of its analysis and suggest a solution. But it’s not allowed to take action. Agents can use only read-only commands included on an allow list.
“An incorrect command sent to network equipment could itself cause an incident,” says Lee. “This restriction is enforced within the application, not merely through a prompt, so the AI cannot perform the action even if it attempts to do so.”
The next step is to scale. KT plans to expand AIO-Net use from 70 units of critical equipment currently to 280, and eventually into access networks closer to subscribers.
“This is still only the beginning,” says Jeong. “We believe that continuing to prepare for this journey together with Microsoft will be valuable.”
Top image: KT Meister and network engineer Heeyoung Oh. She was one of the experts who trained AIO-Net, an agentic AI platform that helps spot and diagnose network issues. Photo by Taehoon Kang for Microsoft.
