On October 27, the "2023 China Enterprise Digital Transformation Summit and SIE User Conference" hosted by SIE was grandly held in Wuhan. The theme of this conference is "AI driven digital innovation", bringing together top domestic industry professors, experts, scholars, 500+business leaders, and 1000+industry elites to deeply explore and solve many issues faced by enterprises in digital transformation, such as intelligent management, AI scenario applications, "dual cross" integrated innovation, value based domestic substitution (information and innovation), and Chinese enterprises going global.

At this conference, the latest release of SIE Information, the industry's first AIGC middle office focusing on the big model of enterprise services (Shanmou GPT), became the focus of the industry.

SIE AIGC Middle Platform (Shanmou GPT) Release Ceremony

Research firm Gartner predicts that by 2024, 40% of AI applications will be generative AI applications, and by 2026, 100 million people will have an AI assistant. Behind the massive outbreak of AIGC applications, major technology manufacturers are entering the market and continuously launching AI models to push forward.

The Research Report on the Map of China's Artificial Intelligence Large Models released by the New Generation Artificial Intelligence Development Research Center of the Ministry of Science and Technology shows that the number of large models developed in China ranks second in the world, second only to the United States. At present, 79 large models with a parameter scale of more than 1 billion have been released in China.

The large model market continues to be hot, currently mainly divided into two categories: one is general (basic) large models; The second is the vertical (domain) large model.

A general large model refers to a model that can handle multiple tasks and domains, such as BERT, GPT, etc. It can be applied in multiple fields such as natural language processing, computer vision, speech recognition, etc. ChatGPT and domestic Baidu ERNIE Bot and Zhipu ChatGLM are all generic models.

Vertical large-scale models are ultra large scale models trained with deep learning for specific domains or tasks. Compared to general large-scale models, vertical large-scale models have stronger domain specialization and task specificity, which can better solve specific domain problems and provide more accurate services.

During the implementation of AIGC technology, the industry gradually realized that the value of large models lies in their integration with business scenarios. It is about generalizing, industrializing, and verticalizing AI capabilities, and deeply integrating them with the specific business needs of different industries and enterprises. Therefore, "model application, scenario first" is becoming a consensus in the industry, academia, and research circles, that is, moving from generalization to verticality, and the market has an urgent demand for large models that understand enterprise services.

However, objectively speaking, the application of large-scale models is still in its early stages, and many industries have not yet found mature scenarios. So, how to make the big model have more industry knowledge accumulation, stronger professionalism, and more targeted application in business scenarios? How to explore more practical application scenarios to help enterprises gain broader business prospects? This is the direction that major companies are striving to explore in this big model competition.

In the upsurge of this big model competition, SIE Information first chose to be a big model in the enterprise service field. Based on SIE's "cross industry, cross domain" service advantages and professional knowledge accumulation for many years, it focused on industry scenario models, deeply supporting various fields of enterprise digital transformation, rather than being a big model in common language.

It is reported that SIE AIGC middle office (Shanmou GPT) deeply integrates the knowledge and best practices of SIE in the fields of finance and taxation, human resources, marketing, supply chain, research and development, production and manufacturing, and expands the storage memory, adaptation application, scheduling execution and domain expertise of the large model through context memory, knowledge/database table index, Prompt engineering, agent execution, general tool set, etc., forming a systematic enterprise service model. To provide enterprises with capabilities such as multi model docking, vector management, pre training and application of private models, etc., to help enterprises quickly implement AI and achieve intelligent process management, interaction, guidance and circulation.

SIE AIGC middle office (Shanmou GPT) architecture and core capabilities

SIE AIGC middle office (Shanmou GPT) architecture and core competence Shanmou GPT provides enterprises with a variety of application scenarios, including digital employees (intelligent guidance), intelligent documents (intelligent assistance), enterprise knowledge base (intelligent question and answer), AI automatic quotation (intelligent quotation), etc. By integrating with the existing systems and data of the enterprise, Shanmou GPT can provide more personalized and intelligent services to improve the efficiency of production and management.

SIE AIGC middle platform (Shanmou GPT) solution and core application scenarios

At the press conference, Yuan Haigang, chief architect of SIE Gushen Laboratory, said: "The launch of AIGC Zhongtai (Shanmou GPT) marks an important breakthrough of SIE in the field of artificial intelligence. We believe that through technological innovation, we can provide more intelligent and efficient solutions for enterprises, helping them achieve digital innovation and development. In the future, we will strengthen in-depth cooperation and research with customers from different industries, expand the application and landing of AIGC in different industries and fields, and further enhance the performance and applicability of Shanmou GPT

With the popularization and promotion of AIGC, AI is no longer simply executing preset rules and algorithms, but can learn and make autonomous decisions, automatically execute predetermined tasks according to different scenarios and needs, and achieve more accurate and efficient human-computer interaction. Based on various enterprise service experiences, having rich application scenarios is a major advantage of SIE training model. In this regard, Yuan Haigang, the chief architect of SIE Gushen Laboratory, analyzed that: if the big model is to be implemented in the enterprise, it must be supported by the scene; Only by clearly identifying the AI enabled scenarios in digital transformation can enterprises truly achieve intelligent management innovation. At present, SIE has fully integrated AI capabilities into agile development, and achieved automatic management of the whole process from demand planning, demand analysis, task allocation, progress tracking, iterative testing and online deployment, with efficiency improved by more than 40%.

According to the introduction, SIE Gushen Lab has analyzed the enabling value and future application potential of AIGC in 36 categories and 175 sub categories of business scenarios in combination with the actual scenarios and business characteristics of SIE Information in seven core areas, namely, smart manpower, smart marketing, smart finance and taxation, smart supply chain, smart manufacturing, data intelligence, and smart platform, as well as one office assistant, in order to provide useful reference and reference for the digital and intelligent transformation of traditional enterprises. The next step will continue to gather industry strength and promote traditional enterprises to embrace the era of artificial intelligence.

此次AIGC中台（善谋GPT）的发布，标志着赛意信息在战略技术创新方面迈出了新的一步。未来，赛意信息将携手更多优秀企业共同摸索，提炼头部优秀企业最佳实践，逐渐完善自己的产品体系，为企业提供更高效、智能化的解决方案，帮助企业高效应用AI，更好地迈入“数智化转型”时代，实现业务的持续创新和发展。