Imagine a manufacturing giant whose production line is filled with intense and orderly energy. At its decision-making center, however, the executive team is faced with the challenge of how to incorporate the latest advances in artificial intelligence — large model technology — into its complex supply chain management. They aspire to improve efficiency and reduce costs through AI, but they encounter obstacles in practice. Data silos hinder the flow of information, existing systems are a headache with compatibility with cutting-edge AI models, and employee acceptance and proficiency with new tools are uneven.
In this context, the Cosmic AI management assistant released by Kingdee Cloud Sky Summit provides new ideas for solving the above problems.
At the summit, Data Ape interviewed Zhao Yanxi, Executive Vice President of Kingdee China and President of Large Enterprise Division, and Li Fan, Vice President and General Manager of R&D Platform Department of Kingdee China, who shared Kingdee's thinking and practical experience. Next, we will discuss how to introduce AI technology and large models into enterprise business processes.
Li Fan is Vice President and General Manager of R&D Platform of Kingdee China
Integrate into business processes
It is the key for large models to enter B-end applications
In today's wave of digital transformation, the demand for large AI models in the B-end market is increasing. Compared with the C-end market, the attributes of the B-end business process are more significant, and its complexity and professionalism requirements are also more demanding. Enterprises not only have to process large amounts of data and information, but also ensure that this data flows efficiently and accurately across different business systems. Therefore, the implementation of large models on the B-side is not only a technical challenge, but also a process of deep understanding and reengineering of enterprise business processes.
Business processes play a central role in the operation of enterprises, and their complexity is directly related to the efficiency and competitiveness of enterprises. Compared with the requirements put forward by C-end users, the difficulty of processing B-end requirements is rooted in the characteristics of the business process itself.
Specifically, enterprise business processes are diverse, complex, and dynamic.
Diversity: Business processes vary significantly across industries. In the healthcare industry, for example, business processes need to meet stringent regulatory requirements and clinical standards, while business processes in the tech industry are more focused on innovation and rapid iteration. Even within the same industry, the business processes of different companies will vary according to their market positioning, corporate culture, and organizational structure. For example, a start-up technology company may focus more on agile development and rapid time to market, while an established manufacturing company may focus more on supply chain stability and cost control.
Complexity: Business processes within an enterprise often involve multiple departments, multiple hierarchies, and multiple resource types. There are complex interactions and dependencies between these processes. Taking the product development process as an example, it usually involves multiple links such as market research, design, R&D, testing, production, etc., and the output of each link is the input of the next link, and the delay of any link may lead to the delay of the entire project. In addition, business processes can be affected by a variety of factors, such as internal policies, employee behavior, and technical constraints, which interact to increase the complexity of process management.
Dynamics: Rapid changes in the market environment require companies to flexibly adapt their business processes to new market demands. For example, in the face of new consumer trends, businesses may need to quickly adjust their product design and marketing strategies; Faced with supply chain disruptions, companies may need to reimagine their procurement and logistics strategies. This dynamic nature requires businesses to be able to collect and analyze market information in a timely manner, make decisions quickly, and execute those decisions effectively.
It should be pointed out that the introduction of large model technology into the business process is not a combination with a certain business process, but an integration with the entire complex, diversified and dynamic business process system. Therefore, in the B-end market, the successful implementation of large model technology depends not only on the advancement of its algorithm, but also on its integration with the existing business processes of enterprises. This integration process faces multiple challenges, involving multiple aspects such as technology integration, process adaptation, Q&A professionalism, and user experience.
Data interface and integration: Large models need to be effectively connected with multiple business systems within the enterprise, which usually involves complex data integration work. Enterprise systems may be based on different technical architectures, using different data formats and interface standards, which requires a high degree of compatibility and scalability of large models. In the process of integration, it is also necessary to solve the problems of data consistency, real-time, and security to ensure that data flows accurately and efficiently between different systems.
Process adaptability: The large model must be able to adapt to different business processes, and the diversity of enterprise processes requires that the large model should not only be universal, but also be able to be customized and adjusted according to specific business scenarios. For example, a large model designed for the financial industry may need to integrate large amounts of economic data and risk assessment models, while a large model designed for manufacturing may be more focused on supply chain optimization and production scheduling.
Professionalism and accuracy of Q&A: Questions in enterprise business processes are often highly specialized and complex, which requires large models to provide answers that are not only accurate, but also professional. Large models need to have sufficient domain knowledge, be able to understand business terms, grasp business logic, and provide solutions that meet business needs. In addition, enterprise business processes are very fault-tolerant, and if the information provided by the large model is incorrect, it can lead to serious consequences.
User interaction design: The user interaction design of a large model is crucial to its successful implementation, and the user interaction design should not only take into account the user's usage habits, but also match the workflow of the enterprise. The design should be concise and intuitive, easy to understand and operate, while also providing timely and accurate feedback to help users make the right decisions. In addition, the design should also be flexible and able to adapt to the needs and preferences of different users.
Take, for example, the inventory management system of a large retail enterprise, which needs to process a large amount of sales data, inventory data, and supply chain data. The large model needs to be able to integrate with the enterprise's ERP system, CRM system, and supply chain management system to achieve seamless data exchange. At the same time, the large model also needs to be able to understand the business logic of inventory management and provide accurate inventory forecasting and replenishment recommendations. In addition, the user interface of the large model should be clear and intuitive, easy to operate, and able to provide real-time inventory status and early warning information.
There is another key ingredient to the successful introduction of large models into enterprise business processes – people.
On the one hand, management needs to recognize the strategic value of this matter. The level of awareness of management plays a crucial role, and they must fully understand the strategic value of large model technology to transform business processes. This involves not only grasping the potential of the technology, but also gaining insight into the organizational change and cultural impact that the implementation of the technology can bring.
On the other hand, employees need to keep an open mind, be able to adapt to this change, and slowly develop the ability to work together. Employees are the direct operators of new business processes, and their attitudes and skills directly affect the effectiveness of the application of large models. Companies need to improve employees' awareness of new technologies and enhance their ability to adapt to change through effective training and communication. At the same time, designing a reasonable incentive mechanism to encourage employees to actively participate in the learning and practice of new business processes is an important means to improve employees' operational ability.
In short, the integration and application of business processes and large models is a complex process, involving multiple aspects such as technology, business, and user experience. Enterprises need to deeply understand these challenges and adopt corresponding strategies and measures to effectively use large model technology to improve the efficiency and effectiveness of business processes.
How to introduce large models into business processes
Kingdee gave a solution
Through the above analysis, we know that it is not a simple matter to introduce AI large models into the business process system of enterprises. So, how should this problem be solved? Kingdee gave his own idea of how to solve the problem.
Kingdee's Cosmic AI management assistant released at the 2024 Kingdee Cloud Sky Summit marks an important progress in enterprise-level AI applications. The launch of Cosmic is not only the embodiment of Kingdee's technological innovation, but also a bold attempt to deeply understand and intelligently transform the business process of the B-end market.
In the view of Data Ape, the core breakthroughs of Kingdee Cosmic AI management assistant are mainly concentrated in two places:
First, Cosmic AI management assistants are able to understand and execute complex business logic.
The core capability of Cosmic AI management assistant is that it deeply integrates business logic and artificial intelligence technology, enabling it to understand and execute complex business processes in the enterprise. The foundation of this capability is Kingdee's years of deep cultivation in the field of enterprise management software, as well as the learning and simulation of 7.4 million customers' practical scenarios.
Cosmic's multi-modal interaction capabilities enable it to receive and process data from different sources and formats, including text, voice, images, etc., which provides rich information input for understanding complex business logic. On this basis, the large model behind Cosmic has been "trained" with trillions of training data, which cover various enterprise business scenarios, so that Cosmic has the ability to learn from data and extract business rules.
What's more, Cosmic is more than just a data processing tool, it is also capable of decision support and automated execution based on understood business logic. For example, in contract management, Cosmic is able to understand the content of contracts, automate the contract approval process, and provide intelligent monitoring and risk warning during contract fulfillment. Behind this capability is Cosmic's deep understanding and accurate grasp of all aspects of the business process.
Second, Cosmic can be customized and integrated with Kingdee's existing SaaS, finance and other business software to ensure seamless integration with the existing systems of the enterprise.
The customized integration of Cosmic AI management assistant with Kingdee's existing business software reflects Kingdee's in-depth thinking and forward-looking layout in the field of enterprise-level AI applications. This kind of integration is not a simple technical docking, but a deep business process integration. Through APIs, microservice architecture, or low-code platforms, Cosmic and SaaS, finance, and other systems can achieve data-level interoperability and functional complementarity, so as to ensure the continuity of business processes and data consistency.
At the business logic level, Cosmic can understand complex business scenarios such as financial rules and supply chain processes, and realize intelligent decision support and automated processing. For example, during the financial audit process, Cosmic can automatically extract and analyze relevant data to identify potential risk points, thereby improving the accuracy and efficiency of the audit. This customized integration also means that Cosmic is able to provide personalized features and services based on the specific needs of the enterprise, such as customized data analysis models, business process optimization recommendations, etc.
In addition, Cosmic's integration takes into account the coherence of the user experience. Through a unified interactive interface and operation logic, Cosmic reduces the learning cost of users, makes it easy for employees with non-technical backgrounds to use AI functions, and further improves the popularity and application value of AI technology. This user-centric integration strategy not only improves the operational efficiency of enterprises, but also lays a solid foundation for the digital transformation of enterprises. This integration strategy not only reduces the resistance of enterprises in the process of technology transformation, but also ensures business continuity and data consistency.
Since its release, Kingdee Cosmic AI management assistant has demonstrated its powerful capabilities and practical results in multiple enterprise-level business scenarios, and has been applied in many fields such as financial management, data analysis, and contract processing.
In the financial field, Cosmic AI management assistant can support functions such as business initiation, multi-modal intelligent review, and financial indicator query and analysis through its large model capabilities. For example, the contract middle office management system jointly built by C&D Real Estate and Kingdee uses the driving force of Cosmic AI to optimize the full life cycle management of contract templates, drafting, pre-approval, approval and performance, and significantly improves the intelligence level of business processes.
In the field of human resource management, the Cosmic AI management assistant also performs well. Hisense Group cooperated with Kingdee to create an employee living water platform and nearly 20 business scenarios through the application of Cosmic AI in the field of human resource management, realizing the intelligent experience of the whole journey of employees and the whole link of the talent supply chain. This cooperation not only increased the proportion of internal recruitment, but also greatly improved the efficiency of the cadre inspection process, which increased the proportion of internal recruitment by 120% and the efficiency of the cadre inspection process by 70%.
Large models are a bigger opportunity than cloud computing
With the continuous advancement of artificial intelligence technology, AI large models are becoming a new engine for enterprise business process transformation. Compared with cloud computing, the big model not only changes the way resources are used and the way enterprise services are delivered, but more importantly, it has a profound impact on the business process itself. This impact is reflected in multiple aspects such as cost savings, efficiency improvements, and process reengineering, which have brought unprecedented operating models and market competitiveness to enterprises.
In terms of cost savings and efficiency improvement, the AI large model reduces manual intervention through automation and intelligent processing, thereby reducing labor costs and error costs. For example, in the field of financial auditing, large models can quickly analyze large amounts of transaction data and automatically identify potential risks and anomalies, reducing the workload and error rate of auditors.
Furthermore, the application of AI large models makes it possible to recreate business processes. Traditional business processes tend to be linear and fixed, but the introduction of large models makes the process more flexible and dynamic. Through in-depth analysis of internal and external data, AI large models can reveal inefficiencies in business processes, and these insights enable enterprises to optimize processes in a targeted manner.
Moreover, the predictive power of large AI models enables companies to respond before market changes. By analyzing consumer behavior, market trends, and macroeconomic indicators, AI models can predict future market demand and guide companies to make more accurate decisions in product development, inventory management, and resource allocation. In addition, AI models can predict market trends and guide product innovation, thereby reconstructing traditional business models and realizing the transformation from product-oriented to market-oriented.
From a long-term perspective, the impact of large models on the operating model and market competitiveness of enterprises is far-reaching. It not only improves operational efficiency and reduces costs, but also drives innovation and transformation, improving market responsiveness and customer service. With the continuous development of technology and the deepening of applications, large models are expected to become a key factor to promote the sustainable development of enterprises and maintain competitive advantages.