"It is as if 10,000 smart AI interns had joined the company, but they were not given access badges, ERP accounts, or approval authority."
At the "KOLON BENIT AX BOOST SUMMIT 2026" held on the 13th at Seoul Dragon City in Yongsan-gu, Seoul, Jang Jae-yong, head of Kolon Benit's DX Division, used this analogy to describe companies' current AI adoption status. The point is that while AI models have become sufficiently advanced, organizations are not yet prepared to entrust them with actual work. Now, the question should not be "Have we adopted AI?" but rather "How has our company's work changed?" This was the theme raised by Kolon Benit on the day.
Kang Yi-gu, CEO of Kolon Benit, stated, "While there are tasks where AI performs much better than humans, there are also tasks where human experience and judgment are more important," adding, "It is more important to use AI properly in areas that yield results than simply using it extensively." He explained that merely layering the tool of AI onto existing work processes makes it difficult to generate significant change.
The head of the DX Division described the situation where proof-of-concept (PoC) projects are accumulating at each company as an "AI museum." He emphasized, "When there are too many PoCs, a company becomes an AI museum," and "Real AX begins after the PoC." This means that AI only starts doing actual work once system permissions are connected, business regulations are reflected, and operating costs and responsible organizations are defined.
He also stated that the criteria for evaluating performance must change. The head of the DX Division said, "It is more important whether month-end closing was completed three days earlier than a report stating that there are 3,000 AI users," adding, "Good AX should be a project that even the finance team welcomes." This means that instead of calculating the effectiveness of AI investment after a project ends, measurement criteria should be established from the initial stage.
To this end, Kolon Benit presented a six-stage AX methodology consisting of △task selection, △infrastructure configuration, △business process redesign, △organizational change, △performance measurement, and △operations. The head of the DX Division suggested starting by combining cloud and on-premises solutions as needed, noting that "buying GPUs first is similar to buying a gym building before even starting to exercise."
Hong Seong-hwa, head of Kolon Industries' AX Innovation Center, emphasized that the competitive landscape in the petrochemical and materials industries is changing due to cost competition, oversupply, and supply chain risks, stressing that "what value was created" is more important than the number of AI adoptions. Kolon Industries is promoting AX from six perspectives: value/KPIs, data, processes, architecture, talent, and organization.
The results are specific. By applying an automation agent for shipping documents, processing time per case was reduced from the previous four hours to five seconds, and a 99.8% accuracy rate was recorded in inspection tasks utilizing vision AI. These are cases that went beyond simply testing AI to actually changing work hours and quality metrics.
The event on the day consisted of 17 sessions across three tracks: △Data & Analytics, △AX Journey, and △Enterprise Service. Topics covered included natural language data analysis, autonomous manufacturing, vision AI, enterprise AI agent management, and SAP-based ERP/HR/finance innovation. Three themed booths were also set up at the venue, including "AXeed," which supports AX diagnostics, AI company matching, and review of government support programs.
Kang (CEO) stated, "While we said last year that we would become an 'AX Enabler' helping customers with their AX, this year we aim to take one step further," adding, "We will walk the entire AX journey together with customers, from the stage of identifying their concerns to the stage where actual results are achieved."
in TECH
