Let companies use Excel to do artificial intelligence
It was learned that “Nanqi Xiance” announced that it has received tens of millions of yuan in PreA round of financing. The lead investors in this round are Hillhouse Venture Capital, a venture capital fund owned by Hillhouse, Ginkgo Valley Capital, and old shareholders Yonghua Investment and Evolutionary Game follow-up investment. Prior to this, the company had received tens of millions of angel round financing led by Yonghua Investment.
Founded in September 2018, Nanqi Xiance Company is a technology incubator of the Artificial Intelligence Innovation Research Institute of Nanjing University. The company focuses on the field of general intelligent decision-making, and is helping customers in the e-commerce, automotive, logistics, and online car-hailing industries to provide intelligent decision-making systems. The landing scenarios include improving the GMV of e-commerce companies, optimizing the car design process, and increasing the revenue of online car-hailing coupons. At present, the company has reached strategic cooperation with C&D Logistics and other customers.
Yu Yang, the founder of Nanqi Xiance, told that the preliminary research and development of new traditional industrial products face a lot of trial and error adjustments, which may take more than a year. Customers can reduce trial and error through Nanqi Xiance’s intelligent decision-making platform and can The cycle is shortened to half a year, or even only a few months. In terms of online car-hailing coupon marketing, at the same cost, Nanqi Xiance’s technology can help companies increase their net profit by 5-8% in a short period of time, and with the further optimization of the decision-making algorithm, this ratio can still be Continue to improve steadily.
Annual revenue growth by 300%
Before the establishment of Nanqi Xiance, Yu Yang’s team had established cooperation with companies such as Taobao, Didi, and Cainiao Warehouse on scientific research projects, and carried out algorithm polishing and technical verification. For example, in the warehousing and picking optimization scenario of Cainiao Network, countless different dispatch strategies can be simulated with technology, and finally the dispatch plan that maximizes the efficiency of workers can be obtained.
Currently, the Nanqi Xiance team has about 50 people, of which nearly 80% are R&D personnel. The company’s business is at a rapid development stage. In 2020, the company’s revenue will grow by 300%, and in 2021, revenue is expected to reach tens of millions.
In January of this year, Nanqi Xiance also released a new product of its intelligent decision-making system “Revive” (REVIVE). Users only need to upload data tables such as Excel to the system according to the template in the system, and describe the business logic. , Xianqi system can automatically train to get the optimal solution path of the problem.
Xianqi system application interface
Reinforcement learning: From games to general intelligent decision-making
The core technology of Nanqi Xiance is reinforcement learning (ReinfOrcement Learning), it is in the same line as the famous AlphaGo.
AlphaGo Zero, which defeated Ke Jie, has not undergone any professional chess training or knowledge of Go experts. The designer only set the basic rules of the Go world and the ultimate goal (to win chess) that it hopes to achieve. After countless trial and error in the virtual world, AI finds the best chess method, and finally becomes the world’s No. 1 Go “player”.
Yu Yang told that there are many decisions in business that are similar in nature to Go. An enterprise has an ultimate goal and at the same time interlocking decisions. Just like playing Go, you need to take one step at a time. “As long as there is data for decision-making interaction, our system can function.”
Therefore, unlike the general intelligent decision-making system, a set of Nanqi Xiance system can be applied in different industries, whether it is the product ranking of the e-commerce recommendation system or the design of industrial boilers, and users do not need to have deep knowledge. Industry knowledge accumulation.
The industry is still in its early stages of development
At present, intensive learning technology is still in the early stage of development, and industry awareness is still weak. Major universities and technology giants (such as Google, DeepMind, OpenAI, etc.) master the core technology. There are also very few startups in the world that have implemented them on a large scale in the commercial field.
We have seen that in a single industry, there are other companies that use non-reinforcement learning technology to provide customers with intelligent decision-making optimization, especially in the fields of e-commerce retail and content recommendation with rich data accumulation.
According to iResearch data, the market size of artificial intelligence technology-assisted decision-making applications will reach RMB 5.586 billion in 2022. Taking the manufacturing industry as an example, many industrial digital twin systems have emerged, but Yu Yang told that such systems require experts with senior industry backgrounds to build, and the development cycle is very long. And Nanqi Xiance’s system can not only help shorten the development cycle, but also lower the development threshold and application threshold, and reduce the cost of the enterprise.
As for the team, the founder and CEO of the company, Yu Yang, is a Ph.D. in the Department of Computer Science and Technology of Nanjing University, and has won the OpenAI International Reinforcement Learning Competition champion. COO Xu Liang graduated from the Department of Computer Science and Technology of Nanjing University. He has worked for Huawei, Beijing Bodian, and Schneider Electric. He has more than 15 years of industrial R&D and market experience. CTO Qin Rong is a Ph.D. from the Department of Computer Science and Technology of Nanjing University, focusing on reinforcement learning and gaming.