KaiwuDB ranks among the leaders of IDC's China Industrial Large Model and Agent Solutions.

Recently, IDC, a globally authoritative consulting firm, released the "IDC MarketScape: Vendor Assessment of China's Industrial Large Models and Agents Solutions in 2025". The report selected 18 typical service providers of industrial large models and agent solutions in the Chinese market for in - depth research. It evaluated the vendors from two aspects: existing capabilities and future strategies, providing a reference for industrial enterprises to choose large model and agent services.

Relying on its technological exploration and strategic capabilities in the field of AI + Industrial Internet, COSMOPlat was positioned in the highest "Leader" category in the assessment. It ranked first in the strategic dimension of the report. Its leading exploration and practice in areas such as industrial large - model construction and agent development have been recognized by industry authorities.

Accelerate Technological Integration and Innovate the AI + Intelligent Manufacturing System

The report mentioned that the attention and investment in large models and agents in the industrial field have been continuously increasing. Industrial large models and agents are moving from the embryonic stage to the extensive exploration stage, and their application scenarios are also constantly deepening and expanding.

"After nearly two years of rapid evolution, industrial large models have moved from concept exploration to scenario implementation. Industry leaders and large - and medium - sized enterprises are accelerating their pilot projects, and the impetus from policies and the technological ecosystem is also constantly strengthening. Meanwhile, data governance, model accuracy, ROI evaluation, and human - machine integration remain key issues that enterprises need to solve urgently. Generally speaking, industrial large models and agents will become important driving forces for the digital transformation of manufacturing enterprises in the next 3 - 5 years. We suggest that enterprises choose appropriate service providers and application paths based on their own industry characteristics and IT capabilities, start from key scenarios, and gradually build a scalable and replicable practice system." said Du Yanze, a senior research manager in the manufacturing industry at IDC.

In response to the technological and ecological difficulties in the implementation of industrial large models, COSMOPlat, based on the industrial Internet platform, has connected the enterprise data link, strengthened data foundation and governance capabilities. With the Tianzhi industrial large model as the top - level technological support, industry agents and industrial software as the link, and intelligent terminals and industrial robots as the implementation means, it has built a hardware - software integrated "AI + Industrial Internet" technological system, realizing a complete closed - loop of perception, decision - making, and execution, and creating unique advantages for large models in data integration, technological integration, and ecological collaboration.

The report evaluated that COSMOPlat has gradually transformed its existing industrial data collection and processing capabilities into competitive advantages for large models and agents. It actively integrates AI capabilities into the existing application system, strengthens its advantageous links, makes up for its capability shortcomings, and enhances the value of the overall solution, achieving a good intelligent integration of enterprise operation management and production manufacturing.

From General to Specialized: Create New Practices for AI Implementation

The report mentioned that COSMOPlat has formed vertical large models with industry characteristics in industries such as petrochemicals, energy, home appliances, and machinery, and has precipitated production - process - optimization agents in core process links such as injection molding, sheet metal processing, gasoline blending, and air compressor stations, achieving a breakthrough from general assistants to in - depth scenario applications.

On the one hand, through the precipitation of industry data assets and the injection of professional knowledge, the Tianzhi industrial large model has broken through the generalization limitations of "general large models", built cognitive barriers in vertical fields, and provided services for building industry large models in vertical fields, offering customized model solutions for key industries such as home appliances, petrochemicals, and energy.

In the petrochemical industry, Yanchang Petroleum and COSMOPlat jointly built the Yanchang Yunxiang Industrial Internet Platform, constructing 38 agents in 6 categories. They have achieved practical results in multiple core links such as intelligent diagnosis of indicator diagrams, process optimization of aldehyde separation towers, and intelligent gasoline blending. The timely discovery rate of faults has increased by 90%, the quality control accuracy has doubled, and the one - time success rate of gasoline blending has increased by 60%.

On the other hand, agents have become the "bridge" for large models to be implemented in industrial scenarios. They have changed the complex and high - level human - machine interaction mode in traditional industrial scenarios. One end receives decision - making instructions from industrial large models, and the other end receives natural - language instructions. Through the self - developed agent development platform, the platform integrates a general agent framework and special agent development modules, which can quickly build intelligent systems with autonomous decision - making, task execution, and interactive learning capabilities, facilitating the implementation and application of industrial AI.

In the Haier Shanghai Washing Machine Connected Factory, which was recently selected as a Lighthouse Factory, for the high - energy - consumption injection - molding process, the factory and COSMOPlat jointly developed an injection - molding process agent. After technicians issue instructions, the agent will automatically identify the requirements, retrieve hundreds of process parameters, and command the injection - molding machine to adjust and produce, reducing the machine - adjustment time from 120 minutes to 10 minutes and greatly reducing the consumption of resources and energy.

IDC believes that as industry best practices gradually mature and spread, the market will enter the stage of small - scale replication and large - scale application. By building a solid data foundation with the industrial Internet platform, improving industrial precision with industry large models, and connecting scenario implementation and interaction links with agents, currently, the Tianzhi industrial large model of COSMOPlat has gathered a large - scale and high - quality industrial knowledge base, as well as ecological resources such as more than 4700 mechanism models, more than 200 expert algorithms, and more than 110 agent development tools. It has been implemented in more than 40 scenarios in 9 industries, achieving good enabling results.

In the future, COSMOPlat will continue to promote the exploration of key technologies for large models, continuously optimize and learn through practice and data accumulation, create more scenario agents and industry large models that meet the actual industrial needs, and boost the high - quality and sustainable development of the manufacturing industry.