Core Insights - The emergence of AI large models like DeepSeek has sparked interest in their potential applications across various industries, although a universal application has yet to be identified [1] - Experts suggest that focusing on B2B applications of large models may be a more viable approach in the short term [1] - The 2025 World Artificial Intelligence Conference showcased various AI technologies, including humanoid robots and digital humans, highlighting the advancements in AI [1] Group 1: AI Large Models and Digital Transformation - AI large models serve as a new digital foundation and intelligent application engine, enhancing operational efficiency and decision-making capabilities in enterprises [3] - Traditional digitalization focuses on speeding up processes, while AI large models enable clearer problem identification and more accurate decision-making [3] - The integration of AI large models into business processes is essential for achieving intelligent operations, moving beyond mere data analysis [3] Group 2: Enterprise Applications of AI - Inspur's Haiyue large model emphasizes deep integration with business scenarios, offering enterprise-level applications rather than standalone AI tools [3][5] - The Haiyue framework combines models, scenarios, processes, knowledge, and permissions to create a comprehensive enterprise agent system [5] - The focus is on embedding AI capabilities into business workflows, enhancing efficiency, risk control, and value creation [5] Group 3: Future of AI in Business - Companies need a hybrid AI capability that combines discriminative and generative AI, with high-quality data being crucial for maximizing AI application potential [6] - The shift from a competition of model capabilities to a focus on deepening scene value will determine the long-term success of AI in industry [6] - The essence of this transformation is to align technology with business needs, fostering a digital ecosystem that supports intelligent operations [6]

Group 1 - The "Xiaomiao" industrial model, developed by the Smart Building Materials Research Institute funded by China National Building Material Group, has been publicly launched, focusing on the cement sector as a testing ground [1] - The model integrates three core technologies: the fusion of time-series data with industrial mechanisms, multi-modal scenario collaboration, and decision-making fault tolerance, achieving over 1% reduction in cement batching costs [1] - After over two years of application, the model has established a mature engineering delivery capability, successfully implemented in nearly 100 cement enterprises, with data governance cycles reduced to as short as 14 days and model deployment within 7 days [1] Group 2 - China National Building Material Group's chairman believes AI will act as a "super accelerator" for new material research, significantly shortening development cycles and reducing trial-and-error costs [2] - The group is currently promoting AI's integration into strategic emerging industries for new materials, having built 231 scenario models covering the entire chain from core manufacturing to R&D and supply chain management [2] - In 2024, the State-owned Assets Supervision and Administration Commission will launch the "AI+" initiative for central enterprises, with several enterprises releasing industrial models, including China National Petroleum and State Grid [2]