AI startups in Greater China have strong localization elements – including language, data, content, cloud services, customization, applications and ecosystems – to meet end-user needs that are very different from the rest of the world. However, the constant investment in customized services to meet customer needs has also made it difficult for these companies to achieve profitability. On September 27, Tracy Tsai, research vice president of Gartner, shared Gartner’s latest report, Market Guide for AI Startups, Greater China, and analyzed the AI in Greater China in depth. Market developments for startups.
Tracy Tsai, research vice president at Gartner
*Due to the broad scope of artificial intelligence technology, the report focuses more on the application areas of artificial intelligence software, and does not cover too much hardware such as autonomous driving, robots, drones or Chips.
How do you define a “startup”?
First let’s take a look at what exactly is a “startup”?
A “startup” in Gartner’s definition generally needs to meet three criteria:
1. The establishment time is about ten years;
2. An independent company, not a subsidiary company attached to a large group;
3. The company’s turnover reaches 100 million US dollars.
market trend
The report summarizes four market trends:
First, from hosted solutions to platform-based solutions.
“Take an intelligent robot as an example. When an enterprise has designed all the tools for developing intelligent robots, customers only need to set up the specific names of their necessary business scenarios, and then they can build their own intelligent customer service robots very simply. Reduce manual customization. The resource consumption is a trend, which can effectively accelerate the speed of business operations.” Cai Huifen explained.
Second, incorporating AI into hyper-automated business processes.
In the past, artificial intelligence was more like “perception”, that is, it could understand human voices, understand text, and understand graphics. In the future, more artificial intelligence will improve how it can be integrated into a part of cognition and can assist in decision support for critical or variable processes. “In the past, a business process may have been more regular, but the aid of text and computer vision in artificial intelligence can judge ‘what kind of situation is this and what kind of process should be followed’ according to the input content.” Cai Huifen Speaking of.
Note: “Hyperautomation” is a term created by Gartner, which means that any business and IT process can be automated as long as it can be automated, that is, innovation automation. Hyper-automation is not the automation of a single software, but the automation of handovers between processes, and may also be automation between departmental functions, or even an automation of internal and external connections within a company. For example, more and more artificial intelligence will be integrated into other different software, such as “machine process automation” or “intelligent business process management” or “platform integration system”, when more and more business process related software use When artificial intelligence assists the process to become more intelligent, we call it “hyperautomation”.
Third, there will be more and more localized AI solutions to meet market demands.
“The localization requirement is not only the content of the text, but also includes regulations and business processes. How can we improve the competitiveness of this enterprise? We will support the enterprise to develop more advanced application solutions as permitted by regulations. This It is the artificial intelligence that we see continues to make more in-depth localized artificial intelligence solutions to meet the needs of this market.” Cai Huifen said.
Fourth, there will be more and more low-code or no-code developer interfaces.
In the future, more and more people without professional ability will develop artificial intelligence or use artificial intelligence tools. “It could be a business person who can use a very simple no-code or low-code user interface to quickly create his own AI application with drag and drop or slightly simpler training, which is a very clear trend. ” Cai Huifen said.
Market situation analysis
In terms of market analysis, the report points out five points:
First, the artificial intelligence market is relatively fragmented, and it is difficult to be divided into a simple part.
“It is difficult to say that it is a pure technology, whether it is doing natural language or computer vision; it is also difficult to say that it is in the financial industry alone. Because the financial industry is very finely divided, is it doing insurance or credit? Is it doing Is fraud still doing content inspection, loan approval, or robot customer service?” Cai Huifen explained, “In fact, this market is very broad, and almost every one is fragmented.”
Second, many start-up companies will adopt a “competitive” cooperation method, that is, there will be cooperation and competition in an environment at the same time.
Cai Huifen believes that in the future, it is difficult to have a company that can meet all corporate needs, so cooperation with third parties is an inevitable trend. An AI startup might partner with or compete with a systems integrator for applications; it might also assist a service provider in a cloud platform architecture, but it might also be competing for the same customer to some extent.
Third, it is difficult to verify the level of startups through customer success stories provided by third parties.
“Sometimes a bank’s success story is used by many startups at the same time. Why? It’s because the bank itself has to find the right cooperation solution, so it may have tried many. Every one will say ‘I used to experiment with this bank. ‘, but what is the best? We actually have no way to see which one is the most successful. So in fact, the best way is that when companies use “success stories”, they should use their own existing Third-party partners to verify, or to verify directly with people who know this company.” Cai Huifen explained.
Fourth, start-up companies take a collaborative approach to jointly create solutions for a new application area through customers.
According to reports, this pattern of success is ongoing. Specifically, a client is willing to share their data and ask a third-party startup to help them successfully develop a model and then build an application. After success, the company can then develop the next customer based on its success.
Fifth, use a variety of pricing methods to meet customer needs.
Startups want to be able to attract customers and let them see the offer that best suits their needs.
report conclusion
Let’s take a second look at the report’s two main conclusions:
First, Chinese consumers are very willing to adopt new technologies and new products, which will motivate Chinese AI startups to continuously enhance innovation in AI applications. Cai Huifen explained that any artificial intelligence application requires a large amount of data. Unlike other countries or regions, Chinese consumers may be employees of ordinary enterprises, and they are willing to share their data, use some application carriers for convenience or Personalized services, etc. The strong will of consumers has boosted the innovation and development of Chinese AI companies.
Second, due to insufficient information provided on some key parts such as product differentiation, customer service, and the lack of an English interface on the company’s official website, it is difficult for foreign companies or some IT consulting leaders to evaluate AI startups to verify the company. Whether the description is consistent with its actual situation. “Gartner sees a lot of startups in China describing their products or services, maybe just describing the technology part they offer. But from the technology part, it’s hard to judge the company’s pros and cons and how it’s different from other competitors. What is the advantage of artificial intelligence?” Cai Huifen said, “There are still some global companies that may need to adopt a more localized artificial intelligence solution in China, but they find it very difficult to obtain such information, because the official websites of many artificial intelligence companies are related to The content information is mainly in Chinese, and the English interface is relatively lacking, so it is very difficult for foreign companies to obtain information about local Chinese AI startups.”
In this regard, Gartner also gave some suggestions:
First, when a global company, especially a local company, wants to produce a good localized application with a good user experience, companies should consider using local AI startups.
Second, when evaluating a local startup, it needs to be very clearly defined what the needs of its best-in-class business are. Then, according to its business needs, verify whether the support that the startup can provide (including language, understanding of business domain scenarios, and successful cases) meets its own needs.
Cai Huifen believes that if they want to operate in the Greater China region, artificial intelligence startups should consider using local solutions in China, and develop and design corresponding applications according to the needs of the local market. “China actually has great localization needs in terms of language, data, culture, or ecosystem. If you don’t develop a customized solution according to localization needs, companies may not be able to do it,” she said.
Technology field distribution of AI startups in Greater China
Gartner 2021 Hype Cycle for Artificial Intelligence (Hype Cycle for Artificial Intelligence, 2021)
As shown in the chart above, Shubhangi Vashisth, senior principal research analyst at Gartner, said: “The rate of innovation in artificial intelligence is rapid, and more than half of the technologies in the technology maturity curve will become mainstream within two to five years. Edge artificial intelligence, computer vision, decision-making Innovations such as intelligence and machine learning will both revolutionize the market in the coming years.”
In Greater China, AI startups are at the forefront of the computer vision field. The reason, Cai Huifen told us, because the government has many security needs, it is necessary to apply the solutions of computer vision companies. “AI startups in Greater China are particularly strong in face, computer vision, commodity recognition, object recognition or behavior, and videos,” she said, “because there is too much data in China to support these companies. Be able to make his model, and then change his model. The same is true for customer service robots. We can see that many robots are doing customer service whether it is a mobile APP or a phone call. Whether it is voice or text customer service, they do it all. are very mature, such as the 12345 government service convenience hotline.”
The Dilemma of AI Startups in Greater China – Needed, But Difficult to Profit
According to Cai Huifen, compared with the United States or Europe, the number of AI startups in Greater China is among the highest. Due to the strong appeal of artificial intelligence in the capital market and market demand, artificial intelligence startups have been able to continue to produce new ones. There are even some companies that did not do artificial intelligence in the past (for example, office system automation companies), in order to prove to their customers that they have the ability to provide artificial intelligence, they have called themselves startups to raise market capital and step into artificial intelligence. field. For many companies, whether they mainly sell artificial intelligence technology, or artificial intelligence technology supports the technology products they mainly sell, they will definitely be related to artificial intelligence. This shows that artificial intelligence is particularly popular in the Greater China market. However, it is also very difficult for such companies to make a profit. In Cai Huifen’s view, many startups have to invest a lot of capital costs and resources to develop, hire employees, and model, but when they complete the solution, they will find that their competitors will soon make the exact same solution. Therefore, we have to invest more and different customized content to meet the needs of customers. “It has been continuously investing in this kind of customization to meet the needs of customers, and it becomes difficult to make profits when new applications are created.” Cai Huifen said, “One of the profitable situations is that artificial intelligence startups already have enough The product capability can meet most of the needs of customers, and this situation is easier to profit. But the problem is that it has not been able to fully meet the requirements, because once it meets the requirements, many competitors will launch similar products, so it must always be Continuous investment, research and development of new scenarios or new applications makes it difficult to make profits.”
Development of standard tool platforms to increase profit opportunities
Customized services make artificial intelligence startups consume a lot of human and technical resources, so how should this drawback be overcome? Cai Huifen pointed out that in order to overcome the losses in this link, it is necessary to “do everything possible to create the concept of a standard tool platform”. That is to say, whether there is a way for startups to release their own capabilities and make them a standard development tool or development platform to help customers develop their own development more easily. “This is a long-term solution, otherwise I (artificial intelligence start-up company) have to help him (referring to customers) to customize every time and every thing, which is of course difficult.” Cai Huifen said. However, Cai Huifen also pointed out that for some scenes that have not been done before, although they still need to be customized, this situation will slow down. “In the beginning, because there wasn’t much standardization, startups would lose a lot. But now we can see a lot of technology mature, such as face recognition or robot customer service. So a mature application like this, You can turn to what I just said, using tools and platforms to help customers develop their own easily, so that the so-called loss can be alleviated and their profits can be increased,” she said.
The ‘must-haves’ for AI startup success
Finally, Cai Huifen also mentioned three basic factors that a successful AI startup needs to have:
1. A CEO with a clear strategy and direction
“I have seen many companies that accept a large number of customers in order to have success stories as soon as possible or want to give customers a lot of success stories, but this directly causes them to consume a lot of their own resources, which is a very dangerous approach. Because when the company’s direction begins to become When it is not clear, it is difficult to succeed and even more difficult to profit.” Cai Huifen said.
2. Possess excellent technology
3. Ability to effectively retain diverse talents
“I’m talking about talent, and it’s not just technical talent. Gartner found that successful AI startups have a high percentage of companies with excellent strategic marketing talent. Because if there’s no way to articulate where the value is, it’s better The products of the company cannot be successful, so the diversified talents not only refer to technical talents, but also include talents such as strategy and sales.” Cai Huifen explained.
AI startups in Greater China have strong localization elements – including language, data, content, cloud services, customization, applications and ecosystems – to meet end-user needs that are very different from the rest of the world. However, the constant investment in customized services to meet customer needs has also made it difficult for these companies to achieve profitability. On September 27, Tracy Tsai, research vice president of Gartner, shared Gartner’s latest report, Market Guide for AI Startups, Greater China, and analyzed the AI in Greater China in depth. Market developments for startups.
Tracy Tsai, research vice president at Gartner
*Due to the broad scope of artificial intelligence technology, the report focuses more on the application areas of artificial intelligence software, and does not cover too much hardware such as autonomous driving, robots, drones or chips.
How do you define a “startup”?
First let’s take a look at what exactly is a “startup”?
A “startup” in Gartner’s definition generally needs to meet three criteria:
1. The establishment time is about ten years;
2. An independent company, not a subsidiary company attached to a large group;
3. The company’s turnover reaches 100 million US dollars.
market trend
The report summarizes four market trends:
First, from hosted solutions to platform-based solutions.
“Take an intelligent robot as an example. When an enterprise has designed all the tools for developing intelligent robots, customers only need to set up the specific names of their necessary business scenarios, and then they can build their own intelligent customer service robots very simply. Reduce manual customization. The resource consumption is a trend, which can effectively accelerate the speed of business operations.” Cai Huifen explained.
Second, incorporating AI into hyper-automated business processes.
In the past, artificial intelligence was more like “perception”, that is, it could understand human voices, understand text, and understand graphics. In the future, more artificial intelligence will improve how it can be integrated into a part of cognition and can assist in decision support for critical or variable processes. “In the past, a business process may have been more regular, but the aid of text and computer vision in artificial intelligence can judge ‘what kind of situation is this and what kind of process should be followed’ according to the input content.” Cai Huifen Speaking of.
Note: “Hyperautomation” is a term created by Gartner, which means that any business and IT process can be automated as long as it can be automated, that is, innovation automation. Hyper-automation is not the automation of a single software, but the automation of handovers between processes, and may also be automation between departmental functions, or even an automation of internal and external connections within a company. For example, more and more artificial intelligence will be integrated into other different software, such as “machine process automation” or “intelligent business process management” or “platform integration system”, when more and more business process related software use When artificial intelligence assists the process to become more intelligent, we call it “hyperautomation”.
Third, there will be more and more localized AI solutions to meet market demands.
“The localization requirement is not only the content of the text, but also includes regulations and business processes. How can we improve the competitiveness of this enterprise? We will support the enterprise to develop more advanced application solutions as permitted by regulations. This It is the artificial intelligence that we see continues to make more in-depth localized artificial intelligence solutions to meet the needs of this market.” Cai Huifen said.
Fourth, there will be more and more low-code or no-code developer interfaces.
In the future, more and more people without professional ability will develop artificial intelligence or use artificial intelligence tools. “It could be a business person who can use a very simple no-code or low-code user interface to quickly create his own AI application with drag and drop or slightly simpler training, which is a very clear trend. ” Cai Huifen said.
Market situation analysis
In terms of market analysis, the report points out five points:
First, the artificial intelligence market is relatively fragmented, and it is difficult to be divided into a simple part.
“It is difficult to say that it is a pure technology, whether it is doing natural language or computer vision; it is also difficult to say that it is in the financial industry alone. Because the financial industry is very finely divided, is it doing insurance or credit? Is it doing Is fraud still doing content inspection, loan approval, or robot customer service?” Cai Huifen explained, “In fact, this market is very broad, and almost every one is fragmented.”
Second, many start-up companies will adopt a “competitive” cooperation method, that is, there will be cooperation and competition in an environment at the same time.
Cai Huifen believes that in the future, it is difficult to have a company that can meet all corporate needs, so cooperation with third parties is an inevitable trend. An AI startup might partner with or compete with a systems integrator for applications; it might also assist a service provider in a cloud platform architecture, but it might also be competing for the same customer to some extent.
Third, it is difficult to verify the level of startups through customer success stories provided by third parties.
“Sometimes a bank’s success story is used by many startups at the same time. Why? It’s because the bank itself has to find the right cooperation solution, so it may have tried many. Every one will say ‘I used to experiment with this bank. ‘, but what is the best? We actually have no way to see which one is the most successful. So in fact, the best way is that when companies use “success stories”, they should use their own existing Third-party partners to verify, or to verify directly with people who know this company.” Cai Huifen explained.
Fourth, start-up companies take a collaborative approach to jointly create solutions for a new application area through customers.
According to reports, this pattern of success is ongoing. Specifically, a client is willing to share their data and ask a third-party startup to help them successfully develop a model and then build an application. After success, the company can then develop the next customer based on its success.
Fifth, use a variety of pricing methods to meet customer needs.
Startups want to be able to attract customers and let them see the offer that best suits their needs.
report conclusion
Let’s take a second look at the report’s two main conclusions:
First, Chinese consumers are very willing to adopt new technologies and new products, which will motivate Chinese AI startups to continuously enhance innovation in AI applications. Cai Huifen explained that any artificial intelligence application requires a large amount of data. Unlike other countries or regions, Chinese consumers may be employees of ordinary enterprises, and they are willing to share their data, use some application carriers for convenience or Personalized services, etc. The strong will of consumers has boosted the innovation and development of Chinese AI companies.
Second, due to insufficient information provided on some key parts such as product differentiation, customer service, and the lack of an English interface on the company’s official website, it is difficult for foreign companies or some IT consulting leaders to evaluate AI startups to verify the company. Whether the description is consistent with its actual situation. “Gartner sees a lot of startups in China describing their products or services, maybe just describing the technology part they offer. But from the technology part, it’s hard to judge the company’s pros and cons and how it’s different from other competitors. What is the advantage of artificial intelligence?” Cai Huifen said, “There are still some global companies that may need to adopt a more localized artificial intelligence solution in China, but they find it very difficult to obtain such information, because the official websites of many artificial intelligence companies are related to The content information is mainly in Chinese, and the English interface is relatively lacking, so it is very difficult for foreign companies to obtain information about local Chinese AI startups.”
In this regard, Gartner also gave some suggestions:
First, when a global company, especially a local company, wants to produce a good localized application with a good user experience, companies should consider using local AI startups.
Second, when evaluating a local startup, it needs to be very clearly defined what the needs of its best-in-class business are. Then, according to its business needs, verify whether the support that the startup can provide (including language, understanding of business domain scenarios, and successful cases) meets its own needs.
Cai Huifen believes that if they want to operate in the Greater China region, artificial intelligence startups should consider using local solutions in China, and develop and design corresponding applications according to the needs of the local market. “China actually has great localization needs in terms of language, data, culture, or ecosystem. If you don’t develop a customized solution according to localization needs, companies may not be able to do it,” she said.
Technology field distribution of AI startups in Greater China
Gartner 2021 Hype Cycle for Artificial Intelligence (Hype Cycle for Artificial Intelligence, 2021)
As shown in the chart above, Shubhangi Vashisth, senior principal research analyst at Gartner, said: “The rate of innovation in artificial intelligence is rapid, and more than half of the technologies in the technology maturity curve will become mainstream within two to five years. Edge artificial intelligence, computer vision, decision-making Innovations such as intelligence and machine learning will both revolutionize the market in the coming years.”
In Greater China, AI startups are at the forefront of the computer vision field. The reason, Cai Huifen told us, because the government has many security needs, it is necessary to apply the solutions of computer vision companies. “AI startups in Greater China are particularly strong in face, computer vision, commodity recognition, object recognition or behavior, and videos,” she said, “because there is too much data in China to support these companies. Be able to make his model, and then change his model. The same is true for customer service robots. We can see that many robots are doing customer service whether it is a mobile APP or a phone call. Whether it is voice or text customer service, they do it all. are very mature, such as the 12345 government service convenience hotline.”
The Dilemma of AI Startups in Greater China – Needed, But Difficult to Profit
According to Cai Huifen, compared with the United States or Europe, the number of AI startups in Greater China is among the highest. Due to the strong appeal of artificial intelligence in the capital market and market demand, artificial intelligence startups have been able to continue to produce new ones. There are even some companies that did not do artificial intelligence in the past (for example, office system automation companies), in order to prove to their customers that they have the ability to provide artificial intelligence, they have called themselves startups to raise market capital and step into artificial intelligence. field. For many companies, whether they mainly sell artificial intelligence technology, or artificial intelligence technology supports the technology products they mainly sell, they will definitely be related to artificial intelligence. This shows that artificial intelligence is particularly popular in the Greater China market. However, it is also very difficult for such companies to make a profit. In Cai Huifen’s view, many startups have to invest a lot of capital costs and resources to develop, hire employees, and model, but when they complete the solution, they will find that their competitors will soon make the exact same solution. Therefore, we have to invest more and different customized content to meet the needs of customers. “It has been continuously investing in this kind of customization to meet the needs of customers, and it becomes difficult to make profits when new applications are created.” Cai Huifen said, “One of the profitable situations is that artificial intelligence startups already have enough The product capability can meet most of the needs of customers, and this situation is easier to profit. But the problem is that it has not been able to fully meet the requirements, because once it meets the requirements, many competitors will launch similar products, so it must always be Continuous investment, research and development of new scenarios or new applications makes it difficult to make profits.”
Development of standard tool platforms to increase profit opportunities
Customized services make artificial intelligence startups consume a lot of human and technical resources, so how should this drawback be overcome? Cai Huifen pointed out that in order to overcome the losses in this link, it is necessary to “do everything possible to create the concept of a standard tool platform”. That is to say, whether there is a way for startups to release their own capabilities and make them a standard development tool or development platform to help customers develop their own development more easily. “This is a long-term solution, otherwise I (artificial intelligence start-up company) have to help him (referring to customers) to customize every time and every thing, which is of course difficult.” Cai Huifen said. However, Cai Huifen also pointed out that for some scenes that have not been done before, although they still need to be customized, this situation will slow down. “In the beginning, because there wasn’t much standardization, startups would lose a lot. But now we can see a lot of technology mature, such as face recognition or robot customer service. So a mature application like this, You can turn to what I just said, using tools and platforms to help customers develop their own easily, so that the so-called loss can be alleviated and their profits can be increased,” she said.
The ‘must-haves’ for AI startup success
Finally, Cai Huifen also mentioned three basic factors that a successful AI startup needs to have:
1. A CEO with a clear strategy and direction
“I have seen many companies that accept a large number of customers in order to have success stories as soon as possible or want to give customers a lot of success stories, but this directly causes them to consume a lot of their own resources, which is a very dangerous approach. Because when the company’s direction begins to become When it is not clear, it is difficult to succeed and even more difficult to profit.” Cai Huifen said.
2. Possess excellent technology
3. Ability to effectively retain diverse talents
“I’m talking about talent, and it’s not just technical talent. Gartner found that successful AI startups have a high percentage of companies with excellent strategic marketing talent. Because if there’s no way to articulate where the value is, it’s better The products of the company cannot be successful, so the diversified talents not only refer to technical talents, but also include talents such as strategy and sales.” Cai Huifen explained.
The Links: KCS057QV1AJ-G39 6MBP30RH060 7MBR50SB120-50