(Yicai) Jan. 16 -- After two years of experimenting with large language models in areas such as translation, document processing and human resources, China’s manufacturing sector is now moving AI to the production line where it can have a far bigger impact on costs and productivity.

Midea Cloud, the digital arm of Chinese home appliance giant Midea Group, unveiled a matrix of industrial AI agents yesterday along with the launch of its Meiking AIGC 3.1 platform. Speaking at the event, Midea Cloud President Jin Jiang said Midea aims to slash costs by around CNY900 million (USD129 million) through AI-related applications this year.

The more than 40 industrial AI agents released to the public were selected from tens of thousands of agents already used internally at Midea, Jin said. The Foshan-based company has now rolled out large-scale AI applications across 158 core business scenarios, spanning research and development, manufacturing, supply chains, quality control, logistics, finance, human resources and overseas services.

For example, in production line management, Midea has deployed multiple AI agents to analyze losses. In the past, losses caused by equipment shutdowns or quality issues had to be manually calculated every few hours, making it difficult to quickly pinpoint the root cause. With AI agents, daily losses can now be broken down to specific processes and job roles, with targeted improvement suggestions provided.

In the supply chain, where the vast variety of suppliers and materials makes operations vulnerable to sudden disruptions such as typhoons, AI agents can quickly assess the scope of the impact and propose alternative sourcing plans.

Midea has also introduced AI agents into human resources and legal affairs to assist in resume screening, campus recruitment interviews and contract review processes.

In some areas, AI applications can trim costs by as much as 90 percent, Jin said. The savings generated by Midea’s implementation of AI have continued to grow, from roughly CNY40 million (USD574,000) in 2023 to about CNY180 million in 2024. The figure is expected to reach CNY600 million (USD86 million) in 2025 and climb to CNY900 million this year.

However, rolling out industrial AI agents more widely, especially to small and medium-sized enterprises, still faces real challenges.

“Companies need to shore up their digital foundations before talking about large-scale AI adoption,” Jin said. “The effectiveness of AI applications depends heavily on the level of business participation and the quality of basic data. If a firm’s digital infrastructure is weak or its data is incomplete, results will be compromised.”

In areas such as sales forecasting and production planning, companies may appear on the surface to be using the original systems, but behind the scenes, AI will increasingly provide decision-making support, with human intervention becoming less and less frequent, he added. Ultimately, it will be AI inside, not just another chat window.

Editor: Kim Taylor