Home > News content

Li Xiaoyan, General Manager of Netease AI: Half a year's income over 10 million AR landed at the right time

via:博客园     time:2019/2/4 16:33:51     readed:387


Li Xiaoyan, General Manager of Netease Artificial Intelligence Division

Wen/Peng Fangting

Ten million yuan level. This is the income scale of Netease AIR AR in content marketing in the second half of 2018.

At the time when AR was labeled as "winter" in the whole market, Li Xiaoyan and her Netease AI team still chose this as the primary entry point for AI product, commercialization and scenario. It is not easy to achieve such a level of income, but it also brings some dawn to the industry.

Apollo Box, an American e-commerce company, released data in 2017: AR technology increased its conversion rate by 25%; tests in Beijing and East also showed that AR technology order conversion rate was 10% higher than the average level; and AR trial version directly increased the sales of cosmetics App Play in Taiwan by two times, and even increased the sales of some brands to more than six times.

Another report predicts that the size of AR advertising content market in China will be about 5 billion yuan in 2020, while the global AR related market will be about 45 billion yuan in 2020.

Since entering Netease in 2011, Li Xiaoyan, who has a doctoral background in Computer Science in Zhejiang University, has set up a multimedia technology department to continuously explore the underlying technologies of computer vision, in-depth learning and human-computer interaction. When AlphaGo (Alpha Go Robot) triggered an AI boom in 2016, Li Xiaoyan's team was making AI and AR systems into products. A series of applications, such as Netease insight and Netease visibility, were launched in that period.

In 2017, Baidu Tencent and other companies added codes to AI, while Li Xiaoyan's team began to promote commercialization, upgraded the Multimedia Technology Department to Netease AI Business Department, and established three major businesses: AI Platform, Netease AR, Speech Recognition and Interaction.


Netease's insight into joining hands with Weilai Automobile: the first multi-person AR commercial landing in China

Content and technology are two business priorities of Netease AI at AR level.

Moreover, Netease AR also has a content development team to provide content template for content ecology by studying AR content interaction benchmarking cases.

Li Xiaoyan once said that many cases of Netease AR are done by content ecology. In 2019, NetEase AR's content eco-partners will continue to expand to 50-100.

At present, NetEase AR has cooperated with many well-known consumer brands such as Nongfu Mountain Spring, MeiMei, Fangtai, KFC, Pizza Hut, Dayson and so on.


Netease Insight X Dayson

In addition, relying on Netease's resources, Netease AR also has technology endowed many game products, and has enough internal practical application cases in AR playing. In 2017, Netease Game Dream was recommended to more than 150 countries by Apple as an Apple Store boutique AR game. In 2018, NetEase AR continued to expand the cooperation of the game end, accessing more than 10 kinds of high-traffic games.

Li Xiaoyan said that NetEase AR's advantages lie not only in the leading technology, but also in the priority consciousness and practical experience of flow layout, ecological establishment and experience of word-of-mouth. After all, the latter three can better help NetEase AR's business closed-loop landing.

So, what is the business closed-loop in Li Xiaoyan's mouth?

If technology enabling is the most basic service, content marketing is the most important breakthrough scenario of Netease AR. This is not only from the content gene of Netease, but also based on the general needs of Internet applications. Faced with the situation that the supply side is larger than the consumer side, better advertising, operational content and experience can stand out and interact with users more deeply.

In the future, Netease AR is also exploring more application scenarios. Through the construction of benchmarking case, the boundary of Netease AR is wider and wider.


Through Netease's insight into APP scanner Fangtai Smart Lifting Fume Machine, AR animation demonstrating workflow appears, clicking on the button spot on the screen, you can see different functional parameters. All Fangtai stores (about 6,000 in total) have supported guides to demonstrate product functions to consumers through NetEase's insight into APP.

The development experience of Blippar, a British veteran AR unicorn, also proves this model very well.

Founded in 2011, Blippar was valued at $10 billion. However, when the content scenario and technology architecture were not yet perfect, Blippar pushed itself to the C-end prematurely, consumed a lot of resources on visual search, and finally was acquired by a real estate company for $400 million.

In addition to AR, Netease AI team from its own game application advantages, will also give data intelligence capabilities to the game distribution business, and establish an AI model for the optimization of game distribution in the sea.

On the basis of deep learning model, Netease AI team provides optimized solutions for CPA (cost per action), LTV (total value of life cycle) and other key data of game applications. For example, in the CPA solution, the core algorithm team of Netease AI can predict more potential paying players through the model, and then submit the advertising platform before actual payment to shorten the revenue cycle.

Based on Netease's brand influence in the game industry, Netease AI team will not only implement more optimization schemes, but also reach strategic cooperation with the upstream and downstream head companies of the game industry chain, so that distribution optimization becomes a necessary part of China's game seagoing solutions.

With the optimization of game distribution in the sea as a starting point, Netease AI team hopes to establish cooperative relations with game publishers and content producers, and find a second growth point in the refined operation of AI in the game.

It is worth noting that Netease AI team has strong autonomy in decision-making. In addition to Li Xiaoyan, the general manager of Zhejiang Augmented Reality and Intelligent Interaction Engineering Center, a considerable number of members are also from AI-related majors of famous universities at home and abroad, and the proportion of graduate students exceeds 90%.

Following are some of the interviews that have been sorted out in depth:

Shen Sheng: AR / VR has been cool in the past two years. How do you think of the whole market environment and space? Where is the imagination of this track now?

Li Xiaoyan: First of all, we say AR has been cool in the past two years, indeed. But that's just one aspect of what we see.

AR and VR must be in the big direction. The research and technical circles are advancing along with Moore's law, so it depends on the development of the whole technology and the speed of its progress. This experience has developed very quickly.

In the past few years, it has been heated up from the aspect of arithmetic. People think that arithmetic has been greatly improved, and they think that many problems can be solved. This heat is just normal heat, or trend heat, but if you really look at the industry, the whole thing, I don't think the market is hot.

The development of AR depends on several technologies, such as perception technology, interaction technology, rendering technology, including computing power, and content production efficiency. These factors affect the development of AR market.

Three years ago, with this increase in computing power, we can now see the maturity of several key technologies, such as perception technology, GPU computing power, 5G and so on, which will push the development of AR into the fast track.

As for imagination space, AR is the place where vision intersects. Everything we can do in the physical world and in the digital world can be re-engineered by AR technology, so you know how big the market space is.

Shen Sheng: In terms of product technology alone, what position are we in?

Li Xiaoyan: We are confident to say that AR is the first echelon in the global AI industry if we just pick it out. This confidence is not only a technical advantage, but also a product advantage.

What is more? When you don't consider the development of the whole industry chain, it's a bit blind to simply talk about technology leadership, because your technology leadership should be in a whole, your technology occupies a part of the whole trend, and its role in the whole industry chain.

Because we are more concerned about how to play their own technology, product advantages and genes, including Netease in the AR industry chain can make the greatest contribution. This is an original intention that we had two or three years ago, and we have always adhered to it.

At present, what we do and the important players in the industry chain are all coupled. We are promoting different links, in fact, I would like to say that the chain goes hand in hand, AR industry can develop relatively fast.

If we simply say that our current leading edge, we are actually doing business, is to do the whole market.

Our current leading edge can be understood as: first, user access, second, content developer ecology, as well as content quality, is the final AR content experience of this piece of reputation.

In terms of user access, our current mechanism is a cloud AR architecture. Accessed APP can share our AR content experience mechanism, which is already leading in the industry. The second is content ecology.

And you know that the developers of 3D content are game ecology, so we have been leading in connection, training and accumulation of related ecology in the past two or three years.

Finally, the most important thing is the quality of content, that is, the experience of our whole AR content.

The AR you see is the quality of real-time rendering, such as Haagen-Dazs ice-cream balls, which are usually imitated very much like glass, but we will be more realistic. In addition to the content model, there is also the delivery game and logic.

So the quality of AR content is leading, and has gained a good reputation. That's why we connect some advertisers and so on behind us. They will stick to us and will continue to make cases for us.

Shen Sheng: In terms of business model, are there any overseas companies that are right with us?

Li Xiaoyan: When it comes to matching projects, there are important players in the industry who are close to each other, but there is no exactly the same development model as ours.

A British company is closer, called Blippar. It was valued at $10 billion, but it hasn't kept up with the latest on the whole architecture.

In addition, they did a wrong thing. He went to 2C too early to do visual search, because it consumed a lot of resources. But even so, he is now acquired by a real estate company and valued at more than $400 million.

But we are from B to C, such as advertising, marketing, operations and so on. We are the closest B to C.

We've had a content development team for two to three years, looking for a benchmark case, such as how to do this AR content? How does this interaction work?

This content team will gradually form some templates to give content ecology, how to cultivate content ecology to do a good AR content. Then a very small part of the team is exploring C-terminal explosions, such as those related to people, short videos and live broadcasts. Even now, we are focusing on the tendency to multi-player interaction and multi-player online.

Sound: We see that Netease's insightful partners cover many traditional and Internet enterprises, such as Dayson, Haagen-Dazs, Cade Mall, American League Comments, Fontainet, etc. Can you share with us Netease's insights into the main customer groups, customer demands, and our current mature solutions?

Li Xiaoyan: Now we can see many cases. We have no strict restrictions on our customers. As long as customers want to have deeper activities with their own users, B-end users will cooperate with their own users by interactive way to gain some insights from more users. He sees that we are more about quality content experience and development capabilities.

Operations, advertising and many other scenarios will be needed, you can look at the content-related supply side and distribution side, such as some of the current advertising, e-commerce, because it has a consumer side, as long as there is demand in production and consumption, supply and demand side can be increased, this way.

Take a concrete example. For example, Dayson is a brand. Dayson's customers also think AR is the closest way to show the physical world. That's the closest way, so he wants to do it. He wanted to show off the particle effect of his accelerator, because it was a feature of his product. Then when we showed him this, it met his needs and he wanted to do the case. Eventually, after the advertisement, everybody can click on it and see what Dayson is like anywhere, what the wind is like, it's very real.

Shen Sheng: What other scenarios do we have now besides marketing applications?

Li Xiaoyan: We will expand a lot of scenarios. At present, we focus on marketing because we tried to standardize it earlier, whether it is the portrait of customers, the production process with customers, the standards, the pain of customers, the places we care about.

Another is because of our user reach. You know most of the applications on the Internet now either do the network or do the operation. So it coincides with this customer. This is the most powerful one we have at present. But at the same time, we show off-line that there are customers and case input in the culture and travel industry, and we are doing benchmarking cases.

Shenzhen: In terms of revenue, there are media reports that the revenue in 2018 is good, but no specific data are disclosed. Is it convenient for us to disclose? When is the overall earnings expectation?

Li Xiaoyan: As mentioned just now, we are equivalent to exploring more typical cases of 2B in 2016, typical sites.

In 2017, it turned to content marketing and began to attach importance to user access. After 2018, especially after June 2018, many cases have begun to be scalable.

In the second half of 2018, revenue in the two quarters was 10 million yuan. In the latter half of the year, advertising alone operated the business.

We expect better running water in 2019.

Shen Sheng: The giants all have AR layout. What do you think of the competitors in the market?

Li Xiaoyan: I think so. The definition of competition is different in different time and dimension. There are many companies on the market are laying out AR, or all the global business giants are laying out AR, we are willing to see important players on this track, but the choice of access points, as well as the links in the industrial chain are different.

For example, Apple pushes ARKit, which is a boost to our entire industry. Many developers want to, but we can cross-platform, can help him solve many of his problems in development, these developers will use our SDK to do...

For example, we know that real-time rendering of AR content is implemented in Unity, but it is very difficult to achieve real-time rendering effect in general APP. We have done such a real-time rendering framework separately. So in the process of running AR content, you can have this game-level rendering experience just by accessing my SDK. This cannot be achieved under other frameworks. Because we don't feel like a pure technical solution, I think it's an end-to-end solution, so it links multiple relationships.

From content production to content access, we have an end-to-end solution. At the same time, our end-to-end engine is cooperating with the upstream and downstream of the industry chain. We have close cooperation with OEM manufacturers.

Shen Sheng: If we leave Netease's background, where are the threshold and moat of our project?

Li Xiaoyan: If we just talk about the project itself, I think the direction we choose, the accumulated Netease genes, such as your content production or the engine in the game, are all advantages. AR is ultimately an important core advantage in content. In addition, we are in the lead in technology, we have done a lot of Engineering work, cross-platform, and so on. But in this respect, whether it is the core technology, is the core AR technology and so on. Similarly, we are in the lead in engineering, that is, landing.

Second, I think the advantage is that we have a very close team that has worked together for nearly three years. The team was formed according to the entrepreneurial team at the beginning of its formation. So our team in addition to the core team, in the middle there are a lot of previous outside entrepreneurship, even founders, CTOs and so on joined. The cohesion and cooperation of this team are also good, which I believe is one of the advantages of this team.

When it comes to the threshold and moat, I don't think this is a single factor. Our in-depth understanding of the industry and the business model we build should say that we are technological first. In terms of the deep coupling relationship with the important players in the industry chain, the industry resources we uphold from Netease, 3D content design, games and so on can become a comprehensive barrier to it.

China IT News APP

Download China IT News APP

Please rate this news

The average score will be displayed after you score.

Post comment

Do not see clearly? Click for a new code.

User comments