[Edaily Reporter Kim Hyun-ah ] When a malfunction occurs in factory equipment, artificial intelligence (AI) analyzes the cause and finds ways to minimize production disruptions. It also suggests solutions such as activating other equipment to replace the malfunctioning unit or adjusting work routes. If necessary, it can even assign tasks to robots. AI is evolving beyond simply answering questions to a stage where it understands and executes corporate operations.

Yoo Seung-jae, CEO of Persona AI, identified “Hyper Automation”—the automation of all business processes—as the next challenge for enterprise AI agents at the Future Technology Strategy Forum held on the 5th (co-chaired by Song Sang-hoon, President of the TTA, and Choi Kyung-jin, President of the Korean Association of Artificial Intelligence Law). He emphasized that AI must go beyond merely reading corporate data to understanding an organization’s operational rules and context, and actually performing tasks.

PersonaAI is also participating in the “AI for All” consortium led by the SKTelecom(017670)to jointly develop public-sector agents. Kim Ji-hoon, Head of the AI Business Division at SKTelecom, mentioned in his presentation that PersonaAI is collaborating on the development of public-sector agents.

Ontology: Connecting Scattered Data

CEO Yoo cited data silos, work context, and execution capabilities as barriers that enterprise AI agents must overcome. He explained that since corporate data is distributed across different systems, and AI cannot grasp organizational permissions and work rules, it is difficult for AI to make accurate judgments.

He proposed ontology as the technology to solve this problem. Ontology is a technology that structures the relationships between data points. Rather than processing individual pieces of information separately, it helps the AI understand the entire workflow by identifying how inventory, orders, stores, and distribution centers are interconnected.

For example, if a supply disruption is anticipated, the system can identify affected stores and inventory levels, and determine where to prioritize the delivery of shortfall quantities. It is also possible to find alternative distribution centers and adjust delivery routes and schedules. This approach goes beyond simply searching for data to connect and process various business tasks.

AI Creates and Modifies Work Procedures as Well

The way AI creates workflows is also changing. Previously, when a service outage occurred or a work procedure changed, the person in charge had to analyze the cause and modify the workflow. CEO Yoo explained that this process took an average of 3 to 4 weeks.

PersonaAI is developing technology that generates business procedures using natural language and proposes execution flows. He noted that tasks that previously took 3–4 weeks can now be completed in just 20–30 seconds. The process involves the AI creating a draft, which is then reviewed and approved by a human.

Making Factory Operational Decisions and Controlling Robots

CEO Yoo envisions “Physical AI” as the next step. This involves AI agents connecting manufacturing equipment and robots, extending beyond a company’s information technology (IT) systems.

Currently, factory automation focuses primarily on having equipment repeat predetermined tasks. When a problem arises with equipment, identifying the cause, assessing the impact on production, and devising alternatives often rely on human judgment.

If AI agents analyze production, quality, energy, and safety data collectively, they can predict production disruptions caused by equipment malfunctions and propose countermeasures. For example, when a specific piece of equipment malfunctions, operations can be shifted to alternative equipment or production schedules can be adjusted. In this structure, humans set goals and operational principles and approve plans, while robots handle the actual work.

PersonaAI is developing a Factory Manager Agent (FMA) in which Agent AI specialists in quality, energy, safety, and other fields collaborate. CEO Yoo also presented a case study in which 20 robotic dogs were simultaneously controlled to transport goods. He noted that the company is also participating in physical AI projects in the Jeonbuk and Gyeongnam regions.

CEO Yoo stated, “In the factories of the future, humans will set goals and policies, while AI will monitor conditions to formulate plans and carry out approved tasks,” adding, “When physical labor is required, the system will evolve to deploy robots, such as humanoids.”