Korea Industrial Complex Corp. President Lee Sang-hoon, Wizcore CEO Park Seung-hoon and other attendees listen to a briefing on a domestic ontology-based manufacturing AI platform and thermal process autonomous-operation technology at Wizcore's Seongsu Saenggak Factory in Seongdong-gu, Seoul, on Thursday. Lee is sixth from left. [Korea Industrial Complex Corp.]
Korea Industrial Complex Corp. President Lee Sang-hoon, Wizcore CEO Park Seung-hoon and other attendees listen to a briefing on a domestic ontology-based manufacturing AI platform and thermal process autonomous-operation technology at Wizcore's Seongsu Saenggak Factory in Seongdong-gu, Seoul, on Thursday. Lee is sixth from left. [Korea Industrial Complex Corp.]

AI foundation model for thermal processes to be developed and tested by 2028

Performance to be verified at 13 or more firms in steel, cement and foundry sectors

Wizcore partnership to explore linking manufacturing data via domestic ontology

Korea Industrial Complex Corp. will invest 23 billion won ($16.7 million) in national funds through 2028 to build a manufacturing AI foundation model tailored to thermal processes in the steel, cement and foundry industries. The goal is to connect manufacturing data currently managed in silos across companies and production lines into a single knowledge framework, cutting the time and cost that small and midsize enterprises must spend developing their own AI models.

The state-run industrial complex operator said Friday it visited Wizcore's Seongsu Saenggak Factory in Seongdong-gu, Seoul, on Thursday to discuss ways to apply domestic ontology-based manufacturing AI technology across industrial complexes.

Ontology is a technology that defines the meaning of and relationships among disparate data — covering equipment, processes and quality — so that AI can understand them as a unified body of knowledge. It allows AI to grasp not only numerical readings such as temperature and pressure generated by production equipment, but also which process produced the data and how it relates to product quality.

In manufacturing, data accumulates separately across production management systems, equipment, design drawings and quality control systems. Differences in data formats and terminology between companies and machines have made it difficult to feed that information directly into a single AI model. Small and midsize enterprises looking to adopt AI have faced a heavy cost burden because each company must organize its own data and build its own model from scratch.

To address this, Korea Industrial Complex Corp. is pursuing a project to develop and demonstrate a thermal-process-specialized manufacturing AI foundation model. The plan is to use the 23 billion won in national funding to first build a common AI-based model that multiple manufacturers can share, then adapt it to the specific process characteristics of each company.

The initial target is the heating and heat-treatment processes in the steel industry. The scope will then expand to cover preheating and cooling processes in the cement industry, and melting and casting processes in the foundry sector. The corporation plans to apply the model to at least 13 companies across the three industry groups to verify its performance.

Thermal processes are core manufacturing operations that alter the physical properties of raw materials or products by applying heat — including steel heating and heat treatment, preheating and cooling in cement production, and metal melting in casting. Because product quality and energy consumption vary with process conditions, the sector is considered well suited for demonstrating the benefits of AI-driven process optimization.

The project is divided into an overall management component and three sub-tasks. The first sub-task covers development of a thermal-process-specialized AI foundation model and AI agents, while the second focuses on building a platform for manufacturing data collection, standardization and synthetic data generation.

The third sub-task, led by Wizcore, involves developing and demonstrating an "AI agent-based thermal process autonomous-operation platform" that connects the technology to equipment and systems on the factory floor. The aim is for AI to analyze process conditions, recommend optimal operating parameters to workers, and ultimately enable fully autonomous process operation.

Wizcore has developed NEXPOM, an integrated AI manufacturing platform that collects and analyzes data from factory floors, along with Widdy, a manufacturing-specialized AI agent. In 2021, the Ministry of SMEs and Startups recognized the company as a best-practice case in manufacturing AI development under its Korea AI Manufacturing Platform program. At STK 2026 held in June, Wizcore also showcased manufacturing AI technology that links data from design drawings through to the production floor.

"A great deal of data on equipment, processes and quality is accumulating on factory floors, but it is often managed in different formats that make it difficult for AI to make full use of," Korea Industrial Complex Corp. President Lee Sang-hoon said. "We will explore ways to use domestic ontology to connect the data of manufacturing companies in industrial complexes into meaningful manufacturing knowledge, and to translate that into AI transformation and productivity gains for those companies."


hong@heraldcorp.com