Using real coders to pretend to be AI programmers, this Indian company is so imaginative!

Everyone knows that there is a bubble in AI financing. But how big can this bubble get?

A star AI startup calledEngineer.aijust broke this record. This company with AI as its domain name was founded by two Indian founders. It claims to use artificial intelligence programs to assist companies that lack engineers to automatically "assemble" new codes. Just like customizing pizza, you can automatically generate your own various websites and mobile apps for any person and company.

Low fees, fast speed, big market! This idea quickly gained support, attracting a total of US$29.5 million in venture capital from SoftBank’s artificial intelligence special fundDeepcore Inc.and several other well-known investment companies, making it the largest A in Europe last year. round of financing.

But just yesterday, the Wall Street Journal reported that an internal employee revealed thatEngineer.aidoes not have AI.

Most of the demo production and software generation it claims are currently done by manual engineers in India and other places. In fact, this company does not even have a decent team of artificial intelligence experts. The only thing that can be called technology internally is decision trees.

#How did this company deceive SoftBank and many investment institutions? Let’s take a look.

"Magic AI" that can automatically generate websites and apps

"We have built a software and artificial intelligence called Natasha that allows anyone to order a pizza , customize an app."Engineer.aifounder Sachin Dev Duggalsaid in an interview in India last year.

Engineer.aiPublicly claims that its "human-assisted AI" allows anyone to create mobile applications by clicking on the menu on its website.

Engineer.AI’s key product is its Builder tool, which is aimed at those who want to create an application or e-commerce website for their business, but don’t have the skills or budget to do so. to support development.Duggalgave an example of a French ski school with 200 instructors who wanted to make a website and application that connected different systems, but could not afford the $500,000 required to customize the software.

Engineer.AI

Duggalsays that since much of the code that powers popular applications is Similar, so the company's "human-assisted AI" can assist companies that lack engineers to automatically "assemble" new code.Duggalsaid at the time that approximately 82% of the apps the company had recently developed "were completed within an hour onEngineer.ai."

Engineer.aiFounderSachin Dev Duggalspeaks at a technology conference in Toronto. Image source: SPORTSFILE VIA GETTY IMAGES

Doesn’t it sound very tempting! We found this website and went to test it out.

Web address:Corresponding website

What’s interesting is that before creating, users can also choose existing ones that are similar to their ideas. There are apps likeUberorFacebook.Engineer.aican then begin to automatically create the basic parts of the application.

We randomly selected the type of "mobile application", and the similar product wasQuora(an overseas APP similar to "Zhihu").

After selecting, you will enter the work interface. It will first generate a design and then ask if it needs to be adjusted.

After each interface is selected, it will look like this.

At this time you will find a maximum estimated price in the upper right corner. It may be a bit emotional not to talk about money at this time. Sure enough, the next step was to ask me what platform I needed to build and the production speed.

The website will automatically provide price and time estimates for design and construction. The preliminary design of this software will take 2 months, and the final delivery estimate is February 2020.

What we currently choose is the design and implementation of the entire software. After selection, the payment interface will appear

The entire software The maximum cost is less than 25,000 US dollars, which is approximately less than 180,000 RMB, and can complete the entire process from design to construction of a software similar to Quara.

Compared with manual construction, this price does not seem to have a particularly big advantage for engineers in China and India, but for companies in the United States and Europe where labor costs are extremely expensive, it is very advantageous. It’s attractive, and in the process, the demo is generated in real time and the time and price are estimated, which is said to be based on machine learning.

On the company’s official website, the background of the company’s founders is also very bright. The two founders both graduated from Imperial College London and met there. After that, they successively founded a number of technology companies. , until Engineer.ai was established in 2017.

quickly attracted a large amount of investment and became the largest Series A investment in Europe

The idea of the star founder is in line with market expectations. This company AI companies have quickly gained the attention of a large number of investors, including Lakestar, which has invested in Facebook and Airbnb, Jungle Ventures, a well-known investment company in Singapore, and of course, SoftBank.

In the overseas investment projects on Jungle Ventures’ official website, Engineer.ai is described as “Customizing software through human-assisted AI.”

In November last year, Engineer.ai announced the completion of a $29.5 million Series A round of financing, led by Lakestar and Jungle Ventures, with participation from SoftBank’s DeepCore. This start-up company based in London and Los Angeles also achieved the largest Series A investment in Europe last year.

Last year, Japanese technology giant SoftBank Group Corp. launched an investment fund dedicated to artificial intelligence with expected capital of $108 billion. This investment also became the first overseas project invested by SoftBank DeepCore’s AI special fund. A spokesperson for

Deepcore said he has full confidence in Mr.Duggal’s vision and team. A spokesman for Jungle Ventures said he was proud to be an investor in Engineer.ai and its technology, adding that "the artificial intelligence landscape is diverse." A spokesman for Lakestar said he was also proud of Engineer.ai and its team. With confidence, "growth in the field of artificial intelligence is not going to happen overnight." He said Engineer.ai is being very cautious in showing off its technology to #Lakestarand other investors.

There is no NLP technology and no AI team. The actual technology used is decision tree

However, no matter how exquisite the packaging is, the bubble will still burst one day.

According to the Wall Street Journal, several people familiar with the company’s operations, including current and former employees, revealed that Engineer.ai does not use artificial intelligence to automatically generate the application’s code. The company relies on human engineers in India and elsewhere to do much of the work, they said.

"The company just started the research and development of this automated development program two months ago, and it will take at least a year before it is actually completed."

Some current and former employees said that currently, including a certain Some pricing and schedule calculations are generated by traditional software, not AI, and much of the work is done manually by employees.

These employees also revealed to the Wall Street Journal that the company does not have natural language processing technology. The technology actually used within the company is a technology that appeared in the 1950s - decision trees... and this Definitely not AI.

It uses decision trees internally, but externally claims that it is building deep learning AI automation software, and is selling it happily.

In the face of the Wall Street Journal's accusations, Engineer.ai also pointed to a statement it had published on its website, which stated that "on average about 60%" of the software was completed by machines, and the rest was completed by machines. Human generation. However, when asked again about how exactly which part of its product was produced using AI, the spokesperson refused to elaborate on the grounds that these details were core technology patents.

What’s even more shocking is that, in addition to having no core technology description, the company doesn’t even have a decent R&D team.

According to insiders, there is no advanced machine learning team in the Engineer.ai team. When first asked to identify senior employees with AI expertise, the company gave one.

And in a subsequent website statement, Engineer.ai said that artificial intelligence experts are difficult to hire, but it has recently recruited some machine learning and artificial intelligence researchers. In a separate statement, the company introduced three team members with some experience in data science and other disciplines, but did not elaborate.

Star founder, serial entrepreneur?

Sachin Dev Duggal (right) and Saurabh Pradeep Dhoot (left)

In fact, after carefully looking at the backgrounds of the two founders, Digest.com found that this was not their first "bubble" startup.

Engineer.ai’s two founders,Sachin Dev DuggalandSaurabh Pradeep Dhoot, after graduating from Imperial College London, founded a company called Nivio together in 2004 company. The company is committed to desktop virtualization services, and in 2007 announced testing of its first online desktop service based on Windows XP, which functioned similar to today's cloud computers and allowed customers to use computer applications without a personal computer. .

Indian media also reported extensively on this company and its two young founders.

However, as of July 2010, the company had estimated revenue of approximately $5 million but had not yet achieved profitability. Early Nivio Companion products were "tests" only. By November 2010,DuggalandDhootclaimed that their goal was education.

Although no constructive products were released, in February 2012, Nivio still attracted an investment of US$21 million. This money was led by AEC Partners led by Videocon, an international company valued at US$5.5 billion. The real reason behind this incomprehensible investment is that Videocon’s chairmanVenugopal Dhootis the uncle ofSaurabh Pradeep Dhoot.

However, one year after receiving the investment, Nivio’s official website was closed, its Twitter feed was also dormant, and the company seemed to have disappeared.

Artificial intelligence company or artificial company?

Not only are startups making various attempts to use the trend of AI to gain attention, but large companies have also done this.

Remember Facebook m, a text-based virtual assistant developed by Facebook? We are committed to developing a universal assistant that can chat with you and assist you in your work.

But there are agents hidden behind Facebook m’s artificial intelligence. When the chat content becomes too complex, these agents will take over. Of course, some people think that it is actually the agent who is chatting with you all the time.

Now Facebook m has shut down the service, claiming that it is just an experimental product.

There is also Magic, a company that became popular 4 years ago. It claims that as long as you send a text message, it can help you do anything you want to do, such as buying sushi on a yacht or sending a tiger to your door. Or make your parking ticket disappear.

#In just 48 hours, Magic has received more than 17,000 text messages. People think this is a highly intelligent butler-like artificial intelligence, but behind it are all real-person services hired by agents contracted by Magic.

The purpose of creating a gimmick is nothing more than to obtain a large amount of investment.

According to British investment firm MMC Ventures, startups with artificial intelligence can attract up to 50% more funding than other companies, and the company claimed in an interview with the Wall Street Journal that they suspect that 40% of them are even more Many companies are not using any form of artificial intelligence at all. Part of the reason is that AI is easy to test or preliminary, but much more difficult to deploy at scale.

The most interesting thing is that the name of Engineer.ai's product, Natasha, is the original name of Black Widow in Marvel, a female spy who is good at deception, which now seems quite meaningful.

Finally, I would like to remind everyone that the AI bubble is big and investment must be cautious.