“The transition to AI agents and multi-agents will open up opportunities for increased productivity and safer work environments in industry, as well as new innovations in manufacturing.”

Lee Geon-bok, Microsoft Korea Manager, gave a presentation on the topic of ‘Utilization of Cloud and AI for Manufacturing’ at the ‘2025 e4ds IIoT Innovation Day_Part2’ event held at COEX in Samseong-dong on March 12.

According to Manager Lee Geon-bok, recent advancements in AI technology are heralding a shift from simple generative AI to AI agents and multi-agents. He said that while generative AI primarily played a role in providing answers to users' questions, AI agents are going one step further and positioning themselves as entities that perform tasks.

One of the major changes in AI agents is their ability to perform. While AI in the past was limited to answering questions, today's AI agents understand users' instructions, make plans, and even execute them.

For example, while existing chatbots only answered users' questions, AI agents automatically perform tasks instructed by users, such as creating and sending reports.

Here is the multi-agent system Multiple agents with different roles collaborate to handle more complex tasks.

For example, agents that monitor the defect rate of a production line and agents that manage the raw materials supply chain each perform their own roles and are interconnected to produce optimal results.

Such multi-agent systems have already been introduced in some companies and are producing successful cases.

Toyota has successfully increased productivity by introducing AI agents through a project called 'Obeya' since December last year.

The Obeya system is a large room where people with different expertise come together and collaborate, and AI agents support this collaboration.

The Small Language Model (SLM) is also attracting attention.

SLM has the advantage of being able to operate in the field without a network connection, allowing for the safe processing of sensitive personal information.

SLM has a small footprint, so it can be used in edge devices or embedded devices.

In particular, combining SLM and LLM can provide even more powerful functions.

The introduction of such AI agents and multi-agent systems is driving innovation in various industries, including manufacturing and service industries.

In particular, new forms of robotics solutions are being developed in combination with humanoid robots, which are expected to bring about major changes in fields such as factory automation.

In addition, the development of such AI technologies requires social consensus and legal systems.

For example, in Germany, a law has been passed banning delivery robots from entering confined spaces, a measure designed to ensure that robots can coexist safely with people.

“AI agents and multi-agent systems will bring about great changes to our daily lives and industries,” said Executive Director Lee Geon-bok. “These changes are expected to provide a more efficient and safer work environment and open up opportunities for new innovation.”