"Analyzing 1,103 mobile carrier dealerships in Seoul's districts, the top three districts suitable for opening new KT dealerships are Songpa District, Gangseo District, and Gwanak District. KT's dealership share in Songpa District is 28.6%, lower than the Seoul average share of 34.7%, so opening is recommended. Gangseo District has a stable market size and balanced competition, and Gwanak District has good potential for openings relative to population density."
On the 23rd, at the KT Innovation Hub inside KT's Gwanghwamun office in Seoul. When a KT official asked an artificial intelligence (AI) agent to "create a commercial district analysis system for opening new KT telecom dealerships," the AI asked users questions such as "What is the most difficult part of the current commercial district analysis process?" and "Where do you get commercial district data?" Then it built an AI tool that provides ▲ market status analysis ▲ analysis of KT's entry opportunities ▲ site suitability assessment ▲ investment proposals. It visualized the locations of competitors' and KT's dealerships on a map.
KT is focusing on the AX (AI transformation) B2B (business-to-business) business as a growth pillar for corporations. The Innovation Hub was created for this purpose. It is a space where B2B customers can view KT's AX technology and receive customized consulting. It opened in Oct. last year, and about 200 corporations have visited so far. Of these, about 30 corporations actually adopted AX in cooperation with KT.
Corporations that visit the Innovation Hub can work with KT to build customized AI. They can produce a prototype that can actually run to verify whether the technology is effective.
Jeon Seung-rok, head of KT's AX Division Strategy Headquarters, said, "While 88% globally have adopted AI and are all becoming 'AI companies,' only 14% of corporations have actually increased revenue and financial performance through AI," adding, "At the Innovation Hub, KT enables customers to experience AX firsthand and supports rapid AI commercialization on the industrial front lines."
Regarding what sets KT's AX business apart, Jeon said, "We support the customer's AX execution journey end-to-end, from task discovery to proof," and added, "Before spending about 50 million to 100 million won at the proof-of-concept (PoC) stage, we can minimize the risk of failure." He continued, "The Innovation Hub has more than 120 AI experts, and if a partnership with global big tech is needed, we provide a connection."
For corporations considering how to adopt AI, KT has operated the on-site 'KT AX Squad' since this year. Over six weeks at the customer's site, it develops AI agents, incorporates feedback, and verifies effectiveness. It goes beyond simple tech demos to verify whether there is a return on investment (ROI) in real work environments.
KT emphasized that it is advancing its in-house AX capabilities. KT concentrated its own technologies to develop the 'KT AX Harness' in-house. It designs tools, permissions, execution environments, tests, logs, and approval flows so AI can safely and repeatably perform real tasks. The core of this technology is internalizing the design configurations of agents, tools, and prompts that had relied on external infrastructure, enabling fine-tuning internally. It is not tied to a specific AI model, and based on a high-performance open-source framework, it allows the creation of any AI model needed to suit business situations at any time.
A KT official said, "Because we directly control the number of repetitions per agent and the model's importance rating internally, we can reduce unnecessary computation and finely tune expense and the number of attempts," adding, "With existing tools, when systems run for a long time, they may fail to grasp context properly or lose consistency in answers, but the Harness maintains stable quality without context loss even in research and development (R&D) and practical environments where infrastructure runs continuously over long periods."