POSCO DX is advancing its transformation into an 'AI Native Company' by focusing on 'Physical AI' in manufacturing and 'AI Agents' in administrative and management areas. The company aims to enhance existing employees' roles through active use of AI rather than replacing them.
On July 29, POSCO DX held a media briefing at the Lotte Hotel in Jung-gu, Seoul, where it unveiled specific developments regarding Physical AI and AI workforce initiatives.
Choi Tae-hwan, head of the Management Planning Office at POSCO DX, stated, "An AI Native Company is not merely about integrating AI into existing tasks but redesigning processes with AI as the main actor in operations." POSCO DX plans to transition both manufacturing and administrative sectors to be AI-centric, leveraging its capabilities across operational technology (OT), enterprise resource planning (ERP), and supply chain management (SCM).
In manufacturing, the company aims to evolve its existing smart factories into 'Intelligent Factories' where humans, AI, and robots collaborate. Jo Seok-joo, head of the Convergence Research Institute at POSCO DX, presented Physical AI as a foundational technology to address the pressing issue of declining skilled labor in manufacturing.
Jo noted, "Manufacturing AI (m-AI) focuses on domain expertise to solve specific process issues, unlike general-purpose AI. We are developing systems that allow equipment to autonomously perform repetitive tasks and recognize anomalies." The company plans to implement 'AI Operators' that divide tasks among cognitive, judgment, and control agents, while combining industrial robots with m-AI for high-intensity and high-risk operations.
Yoon Seok-jun, head of the Robotics Automation Center at POSCO DX, emphasized, "Our goal is to establish an intelligent autonomous manufacturing system that can respond to unstructured situations through collaboration among humans, AI, and robots." Currently, robots are being applied in high-temperature tasks, heavy material handling, and sampling at steel mills and group companies. The company is also testing a 'Sim-to-Real' approach, which involves learning and validating in virtual environments before applying it in real factories, along with edge AI based on neural processing units (NPU), quadrupedal robots, and humanoids.
During the briefing, POSCO DX also introduced its domain-specific 'AI Employee' system based on its self-developed AI workforce platform, 'Agentie.' The company believes that merely increasing the number of individual agents has limitations in enhancing organizational productivity, so it aims to implement AI employees that work at the job level using company regulations and knowledge.
Agentie integrates the entire lifecycle of AI employees, including hiring, deployment, performance evaluation, retraining, and management of permissions, costs, and versions. It manages the work permissions, costs, and performance of each AI employee and supports integration with existing internal systems and other agents.
Currently, POSCO DX operates 113 types of internal AI employees across finance, human resources, labor, and project management. The entire POSCO Group is working on over 200 AI agents and more than 100 projects, with user numbers exceeding 500.
Kim Byeong-seok, head of the AI Workforce Task Force at POSCO DX, stated, "We have reduced the pre-closing risk assessment task from about three hours to around one minute, and if expanded to overall accounting closing, productivity could increase by at least 30%."
POSCO DX is also pursuing AI transformation (AX) consulting to analyze and redesign corporate workflows using Agentie. Additionally, the company aims to expand its capabilities in Physical AI and AI workforce, validated in group operations, to external markets, targeting a 30% increase in orders and revenue within the next three years.
Kim added, "While many companies, including competitors, focus on creating AI agents and platforms, our core is establishing a system that operates AI like actual employees. The ability to integrate AI agents into corporate operations and maintain a sustainable AI workforce system is our competitive edge."
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.