"Zhonge Kewen Ge" Completes Over 100 Million Yuan Strategic Financing to Promote the Implementation of Large-scale Models and Decision-making Intelligent Technology | 36Kr Exclusive

Text | Wang Fangyu

Editor | Su Jianxun 

36Kr has learned that recently, the decision intelligence technology and application service provider "Zhongke Wenge" has completed a strategic financing of over 100 million yuan, invested by the Beijing Artificial Intelligence Industry Investment Fund. This round of funds will be mainly used for R & D investment in the field of decision intelligence to promote the development of various businesses.

Zhongke Wenge was established in 2017 and is an enterprise for the transformation of scientific and technological achievements of the Chinese Academy of Sciences. Its founding team and core members are from the Institute of Automation, Chinese Academy of Sciences, and have more than ten years of accumulation in theoretical research, technical R & D and application practice of big data and artificial intelligence technology. Currently, the company has nearly 600 employees and its business is distributed nationwide.

After seven years of development, Zhongke Wenge has formed core capabilities such as deep semantic understanding, domain social computing, and AI platform engineering. Its products and technologies have been widely applied to nearly 1,000 customers in the fields of media, finance, governance, etc., with high revenue growth. It is one of the few AI companies in the industry with a scalable income.

Up to now, Zhongke Wenge has completed a cumulative financing of over 1 billion yuan, and the investors include Zhongwangtou, Shenzhen Capital Group, CAS Capital, CCTV Rongmeiti Fund, and China Development Bank Finance, etc.

In terms of business, Zhongke Wenge not only does the underlying AI large model and data technology, but also the upper-level industry applications. The company has continuously launched the Yayi domestic independently developed model, the Youya multimodal video model, and the Decision Intelligence Operating System DIOS, all of which can realize the productization to empower industry decision-making applications.

Among them, the Yayi model is mainly applied in the fields of intelligent question answering and text generation with semantic understanding and knowledge reasoning capabilities; the Youya multimodal video model integrates multiple information of videos, providing core technical driving for video content understanding, multimodal content generation, and digital asset management; the DIOS base provides customers with systematic supporting technical services from data, model to assisted decision-making, reducing the application threshold of artificial intelligence.

Wang Lei, the chairman of Zhongke Wenge, said that the positioning of the company's large model has two characteristics. One is the reliable and secure base large model Yayi, which is built from zero to one from data accumulation, algorithm training to AI application, with domestic origin and independent intellectual property rights. The second is to specifically serve government and enterprise To B customers, launch industry-level AI standardized products, and conduct independent deployment and continuous iteration in combination with customer data.

For the landing of large models and AI applications, the company has summarized through more than 100 successful practices and can be divided into three stages: data engineering, model engineering, and domain engineering.

Data engineering and model engineering focus on customer data understanding, industry data cognition, and model training and fine-tuning. The landing of Zhongke Wenge's large model mainly combines three data sets: a large amount of independent and secure high-quality data sets, domain fine-tuning instruction sets, and customer and industry data.

Domain engineering is the value embodiment of the large model landing in the industry, abstracting domain problems to realize the transformation from the industry to productivity. In Wang Lei's view, the current mainstream To C large models mainly learn the public knowledge of the Internet, often lacking in-depth understanding of the industry, unable to solve the decomposition and reasoning of complex problems, and insufficient to meet the needs of government and enterprise customers, urgently requiring a breakthrough in decision intelligence technology.

He believes that the landing of large models must go deep into various industries to create accurate industry models, making the models "smaller, more refined, and more accurate". This vertical industry model not only has higher task execution efficiency, but also can help enterprises save computing power and deployment costs. More importantly, it realizes the scientific cognition and decision-making of scene-level problems, and realizes the value remodeling of AI empowering the industry.

The Yayi large model of Zhongke Wenge focuses on vertical fields and has constructed multiple domain-specific "small" models including finance, government affairs, finance and taxation, public opinion, traditional Chinese medicine, etc., which are more in line with the vertical industry needs of government and enterprise customers. Currently, this model has supported the actual landing applications of many large domestic government and enterprise customers, helping customers realize the business system process reengineering and efficiency improvement.

According to the introduction, the highlights of Zhongke Wenge's large model business accumulation include: the all-media intelligent topic selection and production of Xinhua News Agency, the complex knowledge question answering of Beijing Bank in the financial field, the traditional Chinese medicine assisted medication of the Academy of Traditional Chinese Medicine, the AI intelligent government service system in Zhongguancun, the experts in the field of finance and taxation of Air China TravelSky Holding Company, the financial and legal experts in Hong Kong, the investment research and analysis of securities companies, the credit analysis of small and medium-sized enterprises, and the annual report review of science and technology supervision, etc.

Based on the experience of the landing of AI large model applications, Zhongke Wenge is actively investing in the R & D in the field of decision intelligence and will further focus on the construction of an intelligent decision-making system.

"Intelligent decision-making" is considered the next-generation artificial intelligence technology. AI can not only handle massive data, but also extract valuable insights from it, assisting decision-makers to understand the complex market dynamics and consumer behavior, making quick and efficient decisions, and helping enterprises gain an advantage in the fierce competition.

"Zhongke Wenge is committed to achieving the leap from a single-task AI to an intelligent agent with autonomous learning and decision-making capabilities. It plans to conduct in-depth layouts in the fields of intelligent decision-making, automated workflow, and cross-modal learning to promote the early commercialization of AGI technology," Wang Lei said.