Many "starring actors" including CodeWave are reinterpreting NetEase Shufan. In NetEase Shufan's new story, making professional, low-level, and core tools is the core foundation of its growth to this day.
Author | Doudou
Editor|Pi Ye
Produced | Industrialist
"I hope to build a good ecosystem in the middle layer."This sentence from Wang Yuan of NetEase makes the listener think about the meaning behind it.
As the executive director of NetEase Hangzhou Research Institute, Wang Yuan is also the vice president of NetEase and the general manager of NetEase Shufan. Wearing multiple hats makes Wang Yuan more sensitive and clear than ordinary people about technology development trends and market trends, and he also has a stronger view of the overall situation.
In 2006, NetEase Hangzhou Research Institute was established and began to build an Internet technology system. In 2016, NetEase officially exported the general technical capabilities of the Hangzhou Research Institute to the outside world and established the "NetEase Cloud" brand. In July 2020, NetEase Cloud was upgraded and renamed NetEase Shufan. As digitization and intelligence become more and more in-depth, NetEase Shufan announced not long ago that it will focus more clearly on the field of "digital intelligence software production" to help enterprises create unique digital software. Intellectual competitiveness.
NetEase Shufan's business covers four major technical fields: cloud native, big data, low code, and artificial intelligence. Low code is a product line that has seen considerable action this year. In April this year, NetEase Shufan low-code platform upgraded the CodeWave intelligent development platform. At the same time, it released practical strategies for industries such as automobiles, finance, retail, manufacturing and government.
It is worth noting that the CodeWave intelligent development platform is positioned for the intelligent production of core software for medium and large enterprises.On the one hand, it helps enterprises improve development efficiency from the bottom up.IT personnel can easily implement full-stack low-code application construction from "intelligent generation" to "visual drag-and-drop adjustment", making complex application development more efficient and accelerating the process of enterprise digitalization and intelligence. On the other hand, it can attract more ISV service providers with industry know-how, to enhance its ability to serve various industries.
Behind the creation of this capability corresponds to NetEase Shufan's unique digital attitude and ecological methodology.
The story of NetEase TO B is being reinterpreted by many "starring actors" including CodeWave. And Wang Yuan is one of the people who wrote this new story.
1. The “maze” of enterprise digitalization
Efficient management software is an important tool for enterprise digital transformation. However, for enterprises, if they want to develop a complete system, not only the development threshold is high, but also it requires a huge amount of money and time.
In order to lower the threshold and difficulty of software development, the concept of "low code" was officially proposed in 2014. It became popular all over the country around 2019.
IDC survey data shows that in 2018, the domestic low-code market size was only 370 million yuan, but by 2022, this value was already 3.92 billion yuan, and it is expected to reach 11.84 billion yuan in 2025.
Enterprise demand for low-code is growing.
Large enterprises often face uneven distribution of IT resources. The CIO of a large domestic financial institution with an independent R&D center and an independent technology company that provides IT support for the entire group once said that 80% of its business needs cannot be scheduled for IT.
In addition, due to the differences in the work nature and thinking between IT personnel and business personnel in large enterprises, IT personnel hope that functional applications will be updated less frequently, while business personnel hope that applications will continue to innovate. This difference also existsIt forces large enterprises to move towards low-code development and application.
Because small and medium-sized enterprises do not have large-scale IT development teams and have poor payment capabilities, it is difficult to pay deployment costs that can easily range from hundreds of thousands to tens of millions. The SaaS model may be a good choice, but SaaS standardized products sometimes cannot meet all the needs of small and medium-sized enterprises.
Therefore,All types of enterprises, large, medium and small, have different needs for low-code.
However, many low-code platforms also have limitations. From the current point of view, large enterprises will more often use low-code platforms to solve long-tail requirements at the edge of the system, while small and medium-sized enterprises more often use low-code platforms to build basic core system functions.
“Most low-code platforms now have capabilities mainly concentrated on the front end.”Wang Yuan told industrialists. In short, it is difficult to build complex core systems for large enterprises. This has also brought some inconvenience to the digital transformation of enterprises.
For example, some capabilities are mainly concentrated on the front-end low-code platform, the underlying application layer is not unified, the quality and maintainability of the application are slightly lower, and the development cost will also increase. In fact, in the past, most companies believed that the low threshold of low code corresponded to simple software construction; while complex software construction corresponded to high threshold IT capabilities, which often caused enterprises to fall into the digitization brought about by low code. "maze". Domestically, low-code platforms such as Tritium Cloud, Movable Type Grid, and Jiandao Cloud are relatively well developed. Low-code platforms can basically only implement drag-and-drop development when developing simple functions. When implementing complex functions, technical personnel are still required to write code.
"Business staff resisted, IT staff objected",Wang Yuan still hears this voice today.
In comparison, the development of foreign low-code businesses is relatively benign.
Take the foreign low-code head manufacturer OutSystems as an example. It can basically meet the development of all complex functions.
OutSystems model, in foreign low-code markets, can attract ISV service providers to supplement their capabilities, and create a visual platform for enterprise IT personnel from the bottom up. Build good enterprise digitalization and intelligent paths and business formats.
“From a self-awareness perspective, our team actually doesn’t talk much about the term low-code.”Wang Yuan told industrialists, “NetEase Shufan’s CodeWave intelligent development platform , the benchmark is OutSystems."
Wang Yuan has different expectations for NetEase Shufan and CodeWave.
2. Redefine low-code
Tai Marketing APP is a business office software specially created by Taikang Life for insurance practitioners. It has always brought significant cost reduction and efficiency improvement to business management.
However, as its business develops, the demand for customized and personalized applications has intensified. Whether it is branches or sales teams, there is a strong demand for personalized marketing management applications. However, the traditional development method has the problem of insufficient developers and the inability to quickly respond to business needs.
After internal communication, Taikang Life Insurance decided to introduce the CodeWave intelligent development platform owned by NetEase Shufan. Through the Thai Marketing app-related system built on the low-code development platform, each branch can create its own branch area template in the app.
For example, based on actual business needs, the Shenzhen branch launched home page display, comprehensive health area, report display, personal data dashboard, and integrated functional modules such as Jianbaotong, Baibo Hospital, and customer health rights.
Taikang Life Insurance took full advantage of the "low threshold" of low-code development and trained 120 low-code developers internally within 3 weeks, which helped to balance the company's developers at all levels. The proportion reduces the pressure of developer shortage, while improving development efficiency to meet business changes and innovative needs.
And based on the CodeWave intelligent development platform, Taikang Life Insurance built more than 80 customized application systems within 2 months,precipitated more than ten application templates, help subsidiaries achieve online marketing and operations. Key points such as quick response to demands and reduced talent requirements helped it achieve its target requirements while reducing development costs by more than one million.
In the case of empowering Taikang Life Insurance’s digital transformation,CodeWave intelligent development platform is based on the low-code development capabilities of core software,not only meets Its personalized and customized needs also lower the development threshold and make the application closer to the scene.
“There are also some financial companies that started doing POC testing with us four months ago.”Wang Yuan told industrialists. In short, the "NetEase Shufan + AI large model" model has begun to be delivered,and has been implemented in the financial industry.
In terms of digital transformation in the financial industry, it is undeniable that this is a product worth seeing.
IDC survey data shows that about 1/5 of China's financial companies have deployed or applied low-code development platforms. Among companies that have not deployed or applied low-code development platforms, 65% plan to deploy or apply them in the next 2 to 5 years. Financial institutions are applying low-code development platforms to wider and deeper areas.
In the conversation with Wang Yuan, the industrialist also learned thatAmong the domestic low-code manufacturers, CodeWave has the ability to "from front-end to back-end" It seems that few people have the intelligent development platform capability of unified language.
The question was raised again, what is NetEase Shufan’s confidence?
3. The “addition” of full-stack and AI capabilities
When many companies purchase system software, they consider the data and business barriers between software, as well as the problem of opening up. I prefer to buy full-stack software. That is, the packaging of full-link capabilities from the upper layer to the lower layer. In the low-code track, there are very few manufacturers that have full-stack technical capabilities and can meet the needs of core application development.
The fact is that many startups do not have a technical foundation. Although some companies have strong base capabilities, they lack upper-level BI capabilities.
“Our technology stack is relatively comprehensive and complete from top to bottom.”Wang Yuan told industrialists that this full-stack capability also makes NetEase Shufan Able to build CCDA’s product system.
Specifically, NetEase Shufan's low-code platform (CodeWave) helps enterprises create tools for digital transformation; big data products (BigData) precipitate internal data of enterprises and clean, organize and manage enterprise data, turning data into Valuable resources; AI large models (AI) inherit, learn, and train data, enhance product intelligence capabilities, make intelligent decisions for enterprises, and respond promptly when problems arise in the enterprise; finally, NetEase Shufan leverages Cloud Native creates a stable and efficient technical base to protect business innovation.
This isone of the biggest advantages of NetEase Shufan. This advantage is more obvious when combined with large model capabilities.
In fact, since the wave of AI large models hit, industry discussions on their implementation have never stopped. But it's not an easy task.
In terms of large AI models, NetEase Shufan has two focuses.One is how to apply large models to software development, which is the CodeWave intelligent development platform mentioned above; the other is how to apply large models to software development; the other is how to apply large models to software development. One is data analysis based on large models, and ChatBI is the product of the latter.
"The reason why large AI models are difficult to implement is that they are 'nonsense.'" Wang Yuan believes that credibility and customization are the core of its implementation.
Specifically, it is first necessary to prevent the execution process from becoming a black box, that is, to explain to customers the technical path for training, learning, and decision-making of large models; secondly, it is necessary to ensure that customers can revise erroneous data in a timely manner.
“The accuracy rate is basically between 80% and 90% depending on the scenario.”According to Wang Yuan, NetEase Shufan can only do to this accuracy.
Wang Yuan revealed to an industrialist, "In the field of BI, it seems that there is no large model that can be deployed privately."
4. Behind technology and ecology, the new industrial pattern
“This year, the requirements of governments and enterprises are getting higher and higher.”Wang Yuan paused and told the industrialist.
As far as low-code is concerned, it has been popular for many years and is not a new thing. An obvious trend is that the capabilities of low-code manufacturers are rapidly aligning. Basic capabilities can no longer form a competitive barrier, and manufacturers need to find different paths.
The first is to focus on specific industries and scenarios and dig deep into the moat. Manufacturers that are deeply involved in components and development templates can better meet the needs of Party A and thereby enhance market competitiveness; The second is mainly for developers and ISV developers , precipitation industry Know-how.
In addition to finding the right path, we must also pay attention to building a developer ecosystem. For example, in response to the business needs of ecological partners such as ISVs, low-code vendors should provide them with a software release platform to achieve a win-win situation with ecological partners.
In Wang Yuan's vision, this is also the industrial structure he hopes for.
“I hope to build a good ecosystem. Ecological manufacturers mainly based on ISV, based on NetEase Shufan’s tools and products, achieve cost reduction and efficiency increase. NetEase Shufan passed The industry know-how accumulated by ecological manufacturers in various industries can meet the business digital needs of various industries."
More than 10,000 is the number of partners that NetEase Shufan will gather in the next five years, covering strategy, services, solutions, distribution, and talent partners. Its purpose It promotes low-code development to all walks of life, and even becomes a new career choice for more young people.
In addition, from the perspective of industrial development, large models will also provide new directions for technology and ecology by improving the low-code development experience.
On the one hand, based on large model capabilities, low-code development platforms can provide intelligent application building capabilities for natural language interaction and automatically generate simple applications or prototypes based on demand descriptions.Compared with visualization The “drag and drop” interaction method will further lower the development threshold.
On the other hand, in the application development process,the "development assistant" empowered by large models can assist in generating data models, forms, rule configurations, code snippets, etc., and improve Accuracy.
In the long run, with the optimization and application of large models and AIGC capabilities, especially the enhancement of code generation capabilities, AIGC can directly generate simple applications for medium and long-tail scenarios without relying on programming or low-code development platforms. This will It may have a subversive impact on low-code product forms, business models, competitive landscape, etc.
However, in Wang Yuan’s view, large models essentially bring efficiency improvements to low-code. “Large models are only used by developers when developing software. Auxiliary efficiency tools, because all codes generated by large models still need to be reviewed by developers."
Looking at the current situation, the context of NetEase Shufan's new story is becoming clearer. It is to build professional, bottom-level, and core tools, not to blindly follow the trend, not to follow what others say, and to follow the path it wants to take. This is also the growth of this company. By far the most core background color.