Kingdee shortlisted in the top ten of China’s generative AI model market

Recently, Gartner published “Market Share: Market Data Report for “Enterprise Software, Worldwide, 2024.” The report shows that Kingdee ranks sixth in the revenue of China’s generative AI model market in 2024. As the唯一 manufacturer of enterprise management software among the top ten in this field, this ranking fully confirms Kingdee’s strategic foresight and technological leadership in the implementation of AI technology commercialization. This breakthrough not only showcases Kingdee’s innovative practice results in deeply integrating AI <x1>capacity</x1> into enterprise-level service scenarios, but also reflects the accelerating development trend of the Chinese enterprise management software market moving towards <x2>intelligent</x2> and <x3>digital</x3> depth.

Currently, generative AI technology is rapidly reshaping the global business landscape. From the explosive growth of AI products like DeepSeek to the iterative innovation of cost-effective algorithm models, technological evolution is continuously expanding the application boundaries of “AI + Management.”

As a leader in enterprise management software in China, Kingdee began its layout for AI innovation in 2015 and has increased its investment in AI research and development in recent years, helping enterprises accelerate AI application in the intelligent era. In 2024, Kingdee published the Sky Agent platform and the AI management assistant based on the Kingdee LLM. At the recent 2025 Kingdee Cloud Sky AI Summit, Kingdee launched the fully upgraded Sky Agent Platform 2.0 and five major Agents, including the Golden Key Financial Report (Financial Report Analysis Agent), ChatBI (Enterprise Questioning Agent), Recruitment Agent, Travel Agent, and Enterprise Knowledge Agent. These cover multiple scenarios where AI is most widely applied in enterprises, are ready to use and easy to implement, truly making AI accessible to enterprise managers, business personnel, and all employees.

Kingdee’s AI technology is not just stuck in the lab stage, but has deeply integrated into the real scenarios of enterprises. By launching the AI popularization initiative, fully supporting the MCP protocol, and engaging in deep cooperation with leading companies in the industry, Kingdee is accelerating the transition of AI technology from a “tool” to “productivity,” continuously advancing the process of technological inclusivity, and establishing a practical paradigm for the application of AI in the B-end. Among them, WZ Group leverages intelligent Data Insight to enhance the efficiency of financial management processes; Tongwei Co., Ltd. has launched an HR AI assistant to enable smart inquiries, seamless order placement, and one-click navigation; China Shipbuilding Jiangjin relies on intelligent contract management to improve contract management and review efficiency, reducing contract risks.

Currently, AI technology is reshaping the business ecosystem with unprecedented depth, and the <x1> capacity </x1> for digital intelligence in <x2> enterprises </x2> has evolved from a competitive advantage to a survival necessity. Looking to the future, Kingdee will adhere to the “AI First” strategy, continuously increasing R&D investment, and fully transforming into an enterprise management AI company. The related AI products will also follow the five product concepts of “human creativity as the foundation,” “delivering results,” “safe and trustworthy,” “adaptive experience,” and “ecological openness,” continuously innovating in contextual scenarios to empower every enterprise and individual globally to achieve excellence.

Gartner, “Market Share: 《Enterprise Software, Worldwide, 2024》, April 2025

Gartner does not support any vendors, products, or services in its research reports, nor does it recommend that technology users only select those vendors that receive the highest scores or other titles. The Gartner research <x1>Reports</x1> contain the opinions of the Gartner research and advisory <x2>Organization</x2>, and these opinions should not be considered as statements of fact. As for the research report, Gartner disclaims all express or implied warranties, including any warranties of merchantability or fitness for a particular purpose.

GARTNER is a trademark and service <Indicator> of Gartner, Inc. and/or its affiliates in the United States and internationally, and is used here under license. Keep all rights.

 01

From MRP to EMAI (Enterprise Management AI)

AI has entered the era of universal accessibility, impacting all aspects, including enterprise management. As the Harvard Business Review suggests: AI is reshaping the business models, operations, and competitiveness of enterprises. In the future, every enterprise will be composed of <Agent>, forming an <Agent> decision-making engineering. According to a Gartner investigation, 93% of companies believe that revenue growth will be promoted by 2025, and 66% of companies think that if the opportunity of generative AI is not seized, it will pose a significant challenge in the long-term. The Massachusetts Sloan School of Management has an analysis that believes AI cannot solve all problems. AI excels in repetitive, data-intensive, or data-driven business scenarios. Therefore, it suggests that human-machine collaboration needs to select appropriate application scenarios, design processes well, and undergo management transformation.

In the past three months, Kingdee conducted a survey covering over 20 provinces and more than 80 cities. Our research shows that 82% of enterprises are willing to invest in AI, 75% of enterprises have already taken action, and 42% of enterprises have connected to large models. Since the release of DeepSeek at the beginning of this year, many enterprises have been exploring and discussing how to use large models to improve and transform their management practices. As a result, many large enterprises have already deployed the full version of DeepSeek within their organizations.

In the era of AI, the enterprise’s intelligent digital platform will evolve into Enterprise Management AI (EMAI). The development of enterprise management software has a long history, starting from the 1960s with MRP, which is Material Requirements Planning, then in the 1980s MRP II, which is Manufacturing Resource Planning, followed by ERP in the 1990s, which is the internal integration of resources for enterprise management of personnel, finance, materials, production, supply, and sales. Then came EBC in the early 21st century, which not only manages the internal aspects of enterprises but also extends to upstream and downstream, requiring collaboration with suppliers and customers. In the era of AI, it has undergone a core change. The previous four stages were all tools. Enterprise management AI, because the Agent can handle complex tasks, can execute all capabilities and tasks in a closed-loop scenario and deliver results. Therefore, we believe that enterprise management AI is a fusion of AI and the practice of enterprise management theory, creating a new generation of enterprise management platform centered around its core functions such as strategy, ecosystem, products, and operations.

In this era, as a leading vendor of enterprise management software, Kingdee has been undergoing a cloud transformation for the past decade. From 2014 to 2024, Kingdee has become the number one enterprise management cloud vendor in China, with cloud service revenue accounting for 81.6% in 2024. At the beginning of this year, Chairman Xu Shaochun proposed that Kingdee should transform from a cloud manufacturer to an enterprise management AI manufacturer, aiming to become the number one in China and a world-class enterprise management AI manufacturer within 3-5 years.

This is our product blueprint. The foundation is a unified platform, which is a data source. On top of that is a series of SaaS applications, catering to small, medium, and large SaaS applications. Above that is the agent, which is an agent we create through AI technology around certain scenarios, forming a closed-loop scenario. It can solve some work tasks within closed-loop scenarios. In the coming years, SaaS applications and <Agent> will merge, creating a coexistence state.

02

Five Core Scenarios for the Implementation of Enterprise Management AI

The enterprise management AI aims to solve operational management issues. What is the direction of operational management? Gary Hamel, the master of enterprise management, advocates for a transformation in business management. He believes that the current state of management in our enterprises is based on 19th-century management philosophy, specifically Taylor’s scientific management. Taylor represents the processes of the 20th century and the technology of the 21st century. Therefore, he argues for a systematic transformation in management, where the focus is not on improving efficiency but on unleashing human creativity. He emphasizes the need to establish a self-driven organization to achieve self-driven growth.

What kind of problems does enterprise management AI solve? We believe it should be viewed from five aspects. The first layer is operational innovation, which focuses on end-to-end processes, and this is where most of the capabilities of current enterprise management software lie. The second is product service innovation, which needs to support product innovation. For example, in the AI era, ChatGPT is a revolutionary product that brings competitiveness to our enterprises, along with the transformation of business models. In the past, Kingdee has been transforming to the cloud, which is also a transformation of the business model. This transformation has greatly improved our operational quality and competitiveness. Of course, there is also ecological innovation, creating an open ecosystem. For instance, DeepSeek has built a diverse and open ecosystem of large models in the AI era through an open-source model. Cultural innovation is the most important for every enterprise; it is about building your organizational capabilities and forming a culture. Therefore, we need to focus on organization, talent, culture, and leadership, and this part is particularly difficult to imitate. This is why many people find it challenging to learn from Huawei, as it has its corporate genes, and its management, culture, and leadership capabilities require a long time to accumulate.

Operational innovation requires a shift from static processes to dynamic human-machine collaboration. In the past, it was humans + tools, humans + digital and informational tools, focusing on a human-centered approach with tools as support. The future is about agents + humans, as agents can not only handle repetitive tasks but also manage knowledge-based work. Therefore, the future will involve humans setting goals, directing, and evaluating. We see that Walmart has created an Agent for long-tail suppliers, negotiating with them to reduce costs by 1.5 points, which is significant for retail businesses since their profits are only 2-3 points. Why achieve dynamic human-machine collaboration? This is because our current <Agent> can dynamically schedule related <Agent>s or tools based on intent understanding, and interact with people through feedback to dynamically optimize results. For example, the business trip demonstration by Song Kai earlier only required a single instruction to coordinate a series of <Agent>s, including booking tickets, arranging hotels, and planning itineraries. This way, it can achieve dynamic human-machine collaboration.

What are the core scenarios in enterprise management? We found that there are five core scenarios in enterprises that are AI core scenarios. One is data analysis, AI querying; the second is repetitive workflows, which many enterprises have already applied; there is also optimization of the production supply chain, which needs to be optimized through algorithms; customer insights and marketing, as well as financial risk and compliance. Different industries have different priorities. For example, the Manufacturing sector is more concerned with transactional work and improving efficiency, while for the service industry, customer insights and marketing, as well as replication capabilities, are more important.

For a specific example, in smart recruitment, if we break down the underlying principle, it involves giving it an instruction, such as a development <x1>manager</x1> wanting to hire an AI product <x2>manager</x2> with an HR background. Through a central <x3>Agent</x3>, it will call upon an intelligent agent to create a hierarchy. This hierarchy then calls upon the intelligent agent to publish the job posting, followed by resume screening. After that, it can proceed with job-person matching. Once the matching is successful, interviews are conducted, and the results are organized and fed back to the development <x1>manager</x1>. The development <x1>manager</x1> will then determine whether to proceed to the next stage based on the interview report provided by the AI. Therefore, AI has become the orchestrator of processes, not just an executor. Of course, it also executes based on the information provided by previous <x1>agents</x1> or humans, dynamically scheduling the corresponding <x1>agents</x1> or tools to achieve dynamic human-machine collaboration and realize a new human-machine symbiosis.

The second innovation is in Product and Service innovation. A well-known example is AlphaFold, which is from DeepMind. The project team consists of three people: one is the founder of DeepMind, along with a scientist and a biologist, who together won the 2024 Nobel Prize in Chemistry. An individual with an AI background received the Nobel Prize in Chemistry because AlphaFold can predict a total of 20,000 proteins invented by humans, and it can predict all of them within a few months, as well as innovate composite proteins. Companies like Novartis and Eli Lilly, which are among the Fortune 500, use it for drug development. We are also importing AI in our research and development at Kingdee, currently with over <x1>1000 people</x1> using it, covering <x2>positions</x2> in development and <x3>testing</x3>. Why can these positions and these AIs promote product innovation? Because it is a research and development paradigm, AI is built on data and computing power. Data is replicable, while computing power and algorithms exhibit exponential growth, which can break the constraints of physical resources. Therefore, whether it is software companies like Kingdee or pharmaceutical research and development companies, they can innovate products based on data and models. However, this kind of innovation is not limited to just these two industries; it is also applicable to <x1>Manufacturing</x1>. For example, when producing insights and doing CAD design, AI algorithms can be used for innovation.

The third is the innovation of the business model. In the AI era, the charging method will shift from charging for features to charging based on results. Many industries are already applying this. For example, Douyin used to charge for traffic advertising, but now it charges based on the results facilitated by AI recommendations, as it can <x1> intelligently match <x2> corresponding products and traffic, allowing for intelligent and precise recommendations. SIERRA is a customer service software provider. In the past, they charged based on the number of sites and seats, but now they charge based on the number of issues resolved. There is also Michelin, which used to sell tires but now charges based on the mileage of the tires. They have added chips that can monitor the mileage in real-time, and through these chips, they can understand the condition of the tires for preventive maintenance, achieving a win-win situation between manufacturers and users. Of course, our industry is also transforming, shifting from SaaS to RaaS. SaaS is a subscription model, while RaaS stands for Result as a Service, which charges based on results. The travel <Agent> we are demonstrating today will charge based on the number of business trip <line> entries, which is also a model that charges based on outcomes.

Why can we charge based on results? This is because the AI Agent has the ability to autonomously execute complex tasks, allowing it to complete a closed-loop scenario that can generate value. This value is measurable, so we can charge based on this measurable value, which is referred to as charging by results. So I believe that in the future, more and more products will be charged based on results.

The fourth is ecological innovation, which is a transformation of the automotive industry ecosystem. In the past, the automotive industry was dominated by the leading manufacturers, with the main factory at the core of the chain-based ecosystem. Now that electric vehicles have emerged, this ecosystem has changed, as seen in the collaboration between Huawei and Seres. In the electric vehicle industry, first and foremost, the intelligent driving system is very core because it controls the algorithms; secondly, we need complete vehicle manufacturers, with Seres being a complete vehicle manufacturer and Huawei being the intelligent driving system manufacturer, both of which are core manufacturers; there are also battery manufacturers, for example, they might use CATL; and there are chip manufacturers, and even cloud service providers, because it combines cloud and edge. Therefore, Huawei and Seres have created a diverse ecosystem in the electric vehicle industry. For example, Seres has achieved leapfrog development, with revenue surpassing 145.2 billion in 2024, a year-on-year increase of 305%. This new ecological model has enabled Seres to achieve a leap in growth. If we use a vivid metaphor, the past ecosystem is like The Lion King, while the future open and diverse ecosystem is like Avatar, where all living beings on Pandora are interconnected, forming a symbiotic ecosystem.

Why can we achieve coexistence in the AI era? First, it is because of ecology; each person has their strongest side, and they are business-oriented, forming a community of value. Second, the interconnected data; Huawei has redefined the automobile with software, so its intelligent driving system can access the vehicle’s operational data, making the data between the two systems interconnected. Of course, there is also the capability of AI, which excels at handling data-driven tasks. In this case, data combined with computing power and algorithms can achieve a new growth in value, enhancing the competitiveness of the entire ecosystem and allowing for better returns.

There is also one about cultural innovation. Microsoft CEO Satya Nadella said, “At the core of every technological change is the fundamental reshaping of culture.” After releasing chatGPT in 2023, Kingdee published an enterprise-level AI platform in 2024. This year, General Manager Xu proposed that we transition towards becoming an AI management manufacturer, so we have prioritized AI as one of our strategies. Each month, we will award the AI Innovation Award and organize an AI speech competition for our 10,000 employees. Our experience is that cultural innovation in AI transformation must have “three ‘has'”:

1.You need to have a sense of <x1>. It is important to dare to try, to use AI, to encourage innovation, and to allow for mistakes.

2.There should be an understanding of AI. First, understand what AI excels at and what it does not; as we saw earlier, the MIT Sloan School believes that “AI is better at data-driven, repetitive, and large data volume tasks.” Second, we need to establish an AI-first mindset, prioritizing the use of AI for problems that can be solved by it. Of course, when planning the steps, we must be value-driven and promote practice through use.

3.Have faith. Any transformation is not achieved overnight; various difficulties may arise along the way. At the beginning of the year, many of our clients piloted our AI applications using the knowledge base, but later found that the data in our knowledge base was not precise, so the results were not very good. However, I believe that AI, as a new productive force, will change the management of our enterprises, and this direction is clear. Therefore, we need to choose the appropriate scenarios, formulate the corresponding strategies, and persist.

Therefore, the AI transformation is a revolution in thinking, requiring <x1> curiosity, <x2> AI awareness, and <x3> AI faith.

03

Company Settings

(LLM and OCR)

How can enterprises implement the five management transformations I just mentioned? Kingdee has an AI transformation methodology that we call AIGO. A stands for doing a good job in assessment and architectural planning. Our AI transformation is not just a technological shift; it is also a management revolution. Therefore, we need to conduct vector analysis and perform top-level design, aligning with the company’s strategy to clarify the AI strategy and vision, and then establish our business processes and develop corresponding business frameworks. For example, the intelligent recruitment demonstrated earlier has changed the traditional recruitment process. In the past, a recruitment officer would present a resume to the business supervisor, but now that is no longer necessary. The business supervisor directly collaborates with the recruitment agent for human-machine cooperation, which has altered the original process. Secondly, implementation: AI is an innovative entity, so we need to adopt an agile iterative approach, selecting appropriate pilot scenarios for continuous iteration and optimization. Thirdly, governance: governance requires organization, talent, and culture, ensuring that the corresponding organization and talent are aligned. Fourthly, we need to focus on operation and optimization.

In the first half of this year, we communicated with many enterprises, and every chairman and executive was very concerned about AI, but they were all seeking breakthrough scenarios. The Gartner Speed Layer Model divides the <System> into three layers: One layer is the recording system, such as the Finance System, which has a relatively slow iteration speed, taking about 6-8 years; the differentiation system, which may be our core business system, has a relatively faster iteration speed, possibly 1-3 years; the innovation system is a new business, with an iteration speed of one year.

In different systems, its AI construction strategy varies. In core technology systems, it is often enhanced, with the original SaaS application remaining unchanged, stacking AI technology. For example, during financial document review, the past model used financial auditing based on rules for automation to improve efficiency. Now, generative AI is stacked on top to address the review of contract attachments and formal reviews, solving the last mile problem. Therefore, it is stacked on the original SaaS application. Differentiated systems often start with enhanced layering, such as intelligent purchasing and intelligent scheduling. The original SaaS applications remain unchanged, but AI capabilities are added. However, as the business evolves, for example with intelligent recruitment, it will eventually form an independent agent that creates a closed loop, potentially becoming AI-native. In the innovative system, we encourage everyone to use AI-native suites or to build their own using AI-native technologies, such as AI recruitment, AI live streaming, and AI marketing planning. Therefore, different methods are used in different <x1>domains</x1> and different <x2>systems</x2>.

Of course, to effectively manage enterprise AI, we need a collaborative effort from various ecosystems. We must embrace the LLM ecosystem, including open-source models and commercial models, and build a developer ecosystem that includes ISVs. Therefore, we have invited ISVs to participate in our meeting today, along with interactive partners. Additionally, we need to manage the research ecosystem, as enterprise management AI represents both a technological transformation and a management transformation. Institutions like the Shanghai National Accounting Institute and other research institutes and universities are studying intelligent finance, intelligent HR, and intelligent operations, which helps us understand the management changes behind these application scenario transformations. We are co-creating AI laboratories with many companies. This morning, 6 companies signed the co-construction cooperation agreement for the AI laboratory with us. The companies provide the scenarios, while Kingdee provides the technology and products.

This is the enterprise management AI tree. We hope that our enterprise management AI ecosystem grows robustly like a sapling, forming a forest.

In today’s era of transformation, I believe that enterprise management AI represents both a technological revolution and a management revolution. Kingdee has been committed to creating world-class enterprise management AI products and providing top-notch enterprise management AI services. We hope to collaborate with everyone to jointly discuss, build, and share a new world of enterprise management AI, empowering businesses and achieving excellence. Thank you all!