In May of this year, Lenovo announced that its "supply chain control tower" intelligent agent helped managers shorten decision-making time by 50% to 60%, improve order delivery timeliness by 5%, and reduce manufacturing and logistics costs by 20%; At the end of July, China Telecom announced that its Star Textile intelligent agent has achieved intelligent control of process parameters and fabric quality edge weaving and inspection, with a on-time delivery rate of 99%, a production efficiency improvement of 20%, and a defect detection rate of over 99%... In recent times, leading enterprises have intensified their layout of "industry+AI", promoting deep coupling between large models, intelligent agents, and industry experience, and driving the advancement of industrial intelligence. Currently, artificial intelligence technology is driving profound changes in the production methods and development models of the manufacturing industry. The industry believes that industrial intelligence, as an important direction for the transformation and upgrading of China's manufacturing industry, has advanced from single point technology application to systematization, networking, and self-reliance. However, it still faces key challenges in multiple aspects such as technology, industry, and ecology. In recent years, with the leap and continuous deep application of artificial intelligence technology, more and more industrial enterprises have begun to explore the application of industrial large-scale models and industrial intelligent agents in industrial manufacturing scenarios. Many leading enterprises have increased their layout of "industry+AI", from research and development design, production and manufacturing to business management, and the intelligent application scenarios are becoming increasingly rich. In May of this year, Lenovo released intelligent agents in the manufacturing field, with deep coupling of AI technology and industry experience as the core, to achieve data closed-loop and intelligent driving in key scenarios such as research and development design, production optimization, supply chain collaboration, and customer service. With the assistance of the "supply chain control tower" intelligent agent, Lenovo's global supply chain has achieved end-to-end full value chain coverage. This has reduced decision-making time by 50% to 60%, improved work efficiency by 10% to 20%, increased order delivery timeliness by 5%, and reduced manufacturing and logistics costs by 20%. On June 21, Huawei released the FusionPlant 3.0 intelligent industrial Internet platform strategy, and joined hands with the industry to create a one-stop incubation platform of "AI Agent+intelligent application+hardware innovation", helping partners to efficiently develop vertical exclusive AI Agent and industrial intelligent applications, and accelerating intelligent hardware innovation and entrepreneurship. In terms of Baidu, it is reported that its Wenxin model has been implemented in dozens of industries and hundreds of scenarios, including energy, power, manufacturing, finance, transportation, government affairs, the Internet, education, e-commerce, and more than half of central enterprises are cooperating with Baidu AI Cloud to carry out AI innovation. In addition, at the recent 2025 World Artificial Intelligence Conference, several companies including China Telecom, China Mobile, and China Unicom also showcased their latest achievements in empowering industrial development with AI. The proportion of exploring intelligent agents in industrial enterprises is also significantly increasing. The research results of IDC 2025 on Chinese industrial enterprises show that the proportion of industrial enterprises that have already applied large models and intelligent agents has increased from 9.6% in 2024 to 47.5% in 2025. Among them, the proportion of enterprises that have already implemented applications in multiple links has increased from 1.7% to 35%. In recent years, China's new industrialization has accelerated its progress, and the policy side has also been continuously promoting technological innovation to drive the digital transformation and intelligent upgrading of the manufacturing industry. On June 6th, the Leading Group Meeting for the Integration of Informatization and Industrialization of the Ministry of Industry and Information Technology reviewed the "Key Points for the Integration of Informatization and Industrialization of the Ministry of Industry and Information Technology in 2025". The meeting proposed to deepen the industrial application of artificial intelligence with industrial intelligent agents as the starting point, and drive the innovation and iteration of industrial datasets and industrial large-scale models. Cui Can, Senior Research Manager at IDC China, stated that in the near future, there will be more high-value scenario intelligent agent applications implemented and replicated. With the application of more intelligent agents, it is expected that by 2028, the AI expenditure of Chinese industrial enterprises will reach 90 billion yuan. At present, it is unlikely for enterprises to directly manipulate the production process using big language models. But in work processes that require knowledge and data interaction, such as equipment operation and maintenance, equipment management, and decision analysis, the ability of big language models can be integrated. ”Cao Kai, director of Baidu AI Cloud intelligent industrial solutions, said that the big model can quickly conduct knowledge retrieval, accurate analysis, and generate reports, which greatly improves efficiency. The reporter learned that now listed companies in the industrial chain are actively laying out industrial intelligence, constantly making efforts in sub fields such as research and development design, production optimization, supply chain collaboration, and customer service. Recently, Wiston stated on an interactive platform that the company has invested in the research and development of AI intelligent agent applications in the industrial field, and is forming an industrial information digital application form driven by the combination of industrial AI intelligent agents and large models, and is trying a new model of AI application with multi-mode collaboration. Dongfang Seiko disclosed that its subsidiary, Dongfang Hezhi, will continue to strengthen its research and development investment in this field, lead the intelligent transformation of the industry, and truly achieve AI+empowerment in the packaging industry by strengthening collaboration with the corrugated packaging equipment business unit of Dongfang Seiko, providing industrial Internet platforms and customized system software for industry customers, and providing digital and AI based upgrading solutions for the packaging industry. Henghua Technology stated that the company actively plans to integrate and innovate artificial intelligence technology with energy business scenarios, accelerating the research and development iteration speed of underlying platforms and products. In terms of core technology platforms, we will iteratively upgrade the artificial intelligence engine and intelligent streaming media platform to empower the research and development of multi scenario industry information products and projects, and improve research and development efficiency. In terms of application expansion, we actively promote the use of AI+BIM (Building Information Modeling) for intelligent design and construction services, AI+drone services for surveying, monitoring, and inspection services, AI+energy big data analysis and value-added services, and other multi scenario application services. IFlytek has disclosed that its Tuling Industrial Cloud Platform will actively promote the deep integration of artificial intelligence and the entire industrial process, empower the industrial sector with AI, achieve deep coverage of AI applications in the industrial industry, help industrial enterprises achieve digital and intelligent upgrades, and empower new industrial development with AI. The industry believes that with the continuous advancement of industrial intelligence, there will be more high-value large-scale model application scenarios emerging in the future. But for industrial chain companies, they also need to find a balance between technological breakthroughs, commercial closed loops, and social governance. The future competition is not only about the advantages and disadvantages of individual technologies, but also about the competition of ecological synergy capabilities. In the future of China's manufacturing industry, the application scenarios of big language models will be more diversified. In the field of AI, the focus will be on the mixed application and implementation of discriminative and generative models. ”Song Tao, General Manager of Manufacturing Industry of Lenovo China Government Enterprise Business Group, said. Cui Can also stated that in the coming period, more high-value big model application scenarios are expected to emerge. It is recommended that industrial enterprises establish correct expectations when carrying out big model applications, lay a solid data foundation, and deepen engineering optimization. However, some industry insiders have pointed out that the current industrial intelligence still faces multidimensional challenges such as data, technology, business models, policies, etc., such as data silos and fragmentation, uneven data quality, insufficient generalization ability of AI big models, continuous investment by leading enterprises, and difficulty in large-scale application by small and medium-sized enterprises due to factors such as funding and technology. Therefore, it is necessary to achieve breakthroughs through the coordinated efforts of the entire industry chain. There are still some difficulties that need to be solved urgently for the industrial landing of large models, such as the tight computing power, which has always been a key factor restricting the development of large models. At the same time, in the training and inference process of large models, the utilization efficiency of computing power is generally not high. ”Cao Kai stated. (New Society)