[Edaily Reporter Shin Yeong-bin ] It has been pointed out that simply adopting generative AI models is insufficient for corporate AI transformation (AX) to yield tangible results. The explanation is that data, business logic, and actual execution processes scattered throughout an organization must be interconnected into a structure that AI can understand, and that “ontology” serves as the starting point for this.

Byeon Woo-cheol KTCorporation(030200), Head of Palantir’s Business Division, delivered a presentation titled “On-Site AX Realized by Palantir in the AI Era” at “Deloitte Connect Korea 2026,” held on the 24th at the JW Marriott Hotel Seoul in Seocho-gu, Seoul.

Palantir is a U.S.-based data analytics and AI software company. It provides a platform that integrates and analyzes data held by businesses and governments, linking it to actual decision-making and operational execution. Its flagship products include “Foundry,” an enterprise data operations platform, and “AIP,” an AI platform. KTCorporation formed a partnership with Palantir in March of last year and has been supporting the AX transformation of domestic companies.

Head Byeon explained Palantir’s competitive strengths in three areas: △solutions, △experience in solving problems in challenging environments, and △Chonbang Engineers (FDEs). The Chonbang Engineers model involves deploying engineers to customer sites to work alongside them, analyze real-world data, define problems, and assist in implementing solutions.

“We help define problems by reviewing the client’s domain data and propose the optimal model within the Palantir Foundry,” he said. “But it doesn’t end there—we also serve as change agents within the organization.” This means that rather than simply selling software, KT works alongside clients to define problems, implement systems, and drive organizational change.

Head Byeon also emphasized the difference between traditional IT project methods and the AI era. In the past, when implementing Enterprise Resource Planning (ERP) or Customer Relationship Management (CRM) systems, the goals were relatively clear, and projects could follow a predetermined path based on requirements specifications and implementation plans. However, he explained that in the AI era, where models and technologies change rapidly, the approach of mapping out a perfect roadmap from the very beginning is no longer appropriate.

“In the AI era, it is crucial to ask questions about fundamental issues, experiment, and identify failures early,” he said. “Rather than pushing through to the end using a waterfall approach, we must be able to adjust our direction flexibly.”

Ontology, presented as a key concept in the presentation, is a system that connects corporate data, business logic, and actual work actions, transforming them into a structure that AI can understand and execute. Byeon described ontology as “the output of DX and the input for AX.” The idea is that since it is difficult for AI to directly read and utilize internal corporate data as-is, it must be transformed into a structured format that includes work context and decision-making criteria.

For example, consider a situation where inventory at a specific dealership is running low. While traditional data analysis might simply show how many units remain in stock at that store, if ontology is used to link business logic—including optimal inventory levels, daily sales volume, order lead times, and whether urgent delivery is required—AI can make actual decisions.

Head Byeon categorized the levels of corporate AI adoption into three stages. The first is personal task assistance, such as summarizing meeting minutes or drafting documents. The second is the stage where AI performs some of a team member’s roles. The third is the stage where AI becomes a business partner.

“Ninety percent of companies are still at the level of personal task assistance,” he explained, adding, “Palantir focuses on Levels 2 and 3.” He noted that for a company’s AI adoption (AX) to deliver tangible results, AI must be connected to the company’s domain-specific operational data—such as ERP and CRM systems—and that data must be organized based on an ontology that AI can understand.

The industry views corporate demand for AI as shifting beyond chatbots and document drafting assistance toward the automation of actual business processes. KTCorporation’s emphasis on collaboration with Palantir aligns with this trend. A structure that enables AI to understand corporate data and translate that understanding into actual task execution is expected to become the core of future AX competition.