Food & Beverage

CJ CheilJedang Joins Korea's AI Factory Program to Build F&B-Specific Manufacturing AI

CJ CheilJedang will develop F&B-specific autonomous manufacturing AI under Korea's AI Factory program, partnering with OnePredict, Ibex and KITECH to deploy PINN-based process control.

CJ제일제당, F&B 특화 '제조AI' 표준모델 개발
CJ제일제당, F&B 특화 '제조AI' 표준모델 개발 — Arno Senoner / unsplash

CJ CheilJedang will join Korea's Ministry of Trade, Industry and Energy "AI Factory" support program as the representative company for the food sector, the company announced on the 11th. The project aims to put Physical AI in charge of predicting and controlling equipment and processes on real production lines.

The program targets one outcome above all: AI that runs the plant, not merely watches it. CJ says it will move beyond applying individual AI tools toward a system where recognition, prediction, judgment and control operate as one loop on the factory floor.

Who builds it, and what replaces?

CJ CheilJedang will pursue the project — titled "Development of F&B Autonomous Manufacturing and Integrated Predictive Maintenance Technology Applying PINN Models and Physical AI" — with three partners: OnePredict, Ibex Labs and the Korea Institute of Industrial Technology (KITECH). The system builds on the digital transformation base CJ established through earlier smart factory and production automation investments.

In practice, the technology replaces static process settings and human inspection with AI-driven control. The core technical element is a Physics-Informed Neural Network, or PINN, which analyzes process characteristics and sensor data together to predict optimal operating conditions and autonomously control production. The company will first build an AI-Ready data foundation that connects and analyzes data from different equipment on site in real time for AI training.

What does the AI actually control?

The project covers four concrete application areas:

  • Autonomous process control. PINN-based AI predicts optimal operating conditions and adjusts production processes without operator intervention.
  • Quality inspection. Machine vision, robots and edge AI combined into Physical AI will detect foreign matter, damage and packaging defects in real time, linked directly to equipment control.
  • Predictive maintenance. AI-based anomaly detection and diagnosis will flag failure signs early and reduce unplanned equipment stoppages through an intelligent maintenance system.
  • Worker safety. AI CCTV will detect workers collapsing or entering dangerous zones in real time; on detection, the system will automatically slow or halt production equipment, part of an effort to prevent serious workplace accidents.

The safety application carries regulatory weight in Korea, where major industrial accident liability rules have tightened pressure on manufacturers. CJ did not disclose the program's budget, timeline milestones or how costs split between the ministry and the company — a gap worth noting for any operator benchmarking the model.

Why F&B needs physics-aware AI

Kang Min-soo, CJ CheilJedang's executive vice president heading production, framed the rationale in raw-material terms. "Food manufacturing is a sector where raw material characteristics and changes in the production environment directly affect product quality, so AI technology that understands field data and process characteristics together is essential," he said.

He added that the company will "secure Physical AI technology optimized for F&B manufacturing sites to raise productivity, quality and safety competitiveness, and contribute to strengthening the global competitiveness of K-food." That claim ties the factory-floor program to CJ's export ambitions, though the company offered no quantified targets for yield gains, downtime reduction or defect rates.

The ministry backing matters for scope. State support under the AI Factory program signals Seoul intends to spread autonomous manufacturing across Korean industry, with CJ as the food-sector test case. If the PINN approach works at CJ's scale, it could become a template for other F&B manufacturers facing the same variable raw materials and tight margins — but until CJ publishes results from a live line, the productivity promises remain unverified.

food-manufacturingartificial-intelligencepredictive-maintenancesmart-factoryworker-safety

More from Marcus Bennett

Marcus Bennett

Show full bio

Market editor covering technology and product news in Restaurant and hotel operations at The Pass Brief.

712 articles

Pairings

« Previous articleNext article »