At the 2026 Lean Digital Innovation Conference, held from September 5-7 at the Tianjin National Exhibition and Convention Center, Zhao Yu, President of IRC Group and CEO of Yinhe Intelligent Technology (Tianjin) Co., Ltd., delivered a keynote address on the transformative potential of industrial AI.

Zhao articulated an "AI economic five-layer cake" theory, spanning from the foundational energy layer, through chips, infrastructure, and models, to the application layer. He emphasized that while the first four layers determine what AI can technically achieve, it is the application layer that ultimately decides whether AI delivers practical value. "AI's value accumulates from the bottom up," he noted. "In the application layer, industrial and manufacturing scenarios are the most abundant, covering R&D design, production, logistics and supply chain, and marketing. Therefore, industrial AI is the ultimate value realization layer of AI."

According to Zhao, there are three key indicators that industrial AI has reached a scale where it can be successfully deployed within an enterprise. First, AI must evolve from a single-point tool used by individuals to becoming an integral part of production and operations. Second, it must expand from single-point applications to the entire value chain of operations. Third, AI must move from scenario-specific innovations to enhancing the overall organizational capabilities of the enterprise.

To ensure the large-scale implementation of industrial AI, Zhao proposed a core framework built around industrial knowledge engineering. This system is supported by two complementary "wings": safety engineering and talent development. Additionally, a five-tier technical governance structure provides assurance, starting from the underlying computational infrastructure and hardware, moving up through the model layer and the AI platform development layer, to the industrial knowledge and data layer, and culminating in the FDE (Field Data Engineering) framework which enables real-world industrial application and value monetization.

Zhao stressed that true "scale" is not defined by the number of intelligent agents an enterprise can create today. "The real marker is whether an enterprise can continuously and repeatedly create intelligent agents and successfully translate their value into tangible outcomes," he explained. Only this sustained capability should be considered as achieving scale.

In his closing remarks, Zhao issued three major calls to action for the industry. First, he urged all enterprises to be open to sharing their industrial scenarios, enabling a mutually beneficial "two-way journey" between cutting-edge AI technologies and real-world industrial environments. Second, he called for the engineering-based preservation and conversion of the invaluable industrial wisdom accumulated by companies over decades into core assets for the AI era. Third, he advocated for transforming AI capabilities from tools used by a few employees into a comprehensive organizational strength. He concluded by emphasizing that this evolution is essential for Chinese manufacturing to remain competitive in the global AI-driven landscape.

All conference records are transcribed on-site without review by the speaker.