news

Header image | Screenshot of Fight Back to School

Produced by Huxiu Technology Group

Author | Wang Xin

Editor | Miao Zhengqing

"Please describe your past work experience in a manufacturing company and make the following gestures in front of the camera."

In order to apply for a screw-driving job in the factory, Liu Chuang followed the instructions on his mobile phone to make grasping movements with both hands and showed his arms to the camera.

The difference is that he is not facing a traditional human interviewer, but an AI interviewer, who is wearing a pink suit and is amiable. The key step is the finger flexibility and tattoo detection part of the AI ​​interview.

Job seekers are being screened by AI interview systems customized by companies. From JPMorgan Chase, Goldman Sachs, state-owned banks to SF Express, Walmart, Luckin Coffee... AI interviewers are accelerating their penetration into the recruitment processes of these multinational giants. Some AI interview companies have seen a five-fold increase in interviews this year.

Although AI interviews have only gradually entered the public eye in 2024, this track has taken shape ten years ago. In 2015, Chen Haoxin, former CHO of SEA Group, founded HRTPS and began to explore AI interviews in China. In 2020, the number of players entering the market exploded, reaching more than 30 companies at its peak.

According to Tianyancha, AI interviews started in 2014, and after ten years, more than 30 companies have survived. Now there are only a few left in China, including Haina AI, HRTPS, Avocado, etc. Haina AI is one of the few survivors in this field that focuses on blue-collar AI interviews.

"Some large factories will eliminate candidates with large tattoos or other possible violent tendencies when interviewing assembly line workers."

In an interview with Huxiu, Liang Gongjun, CEO of Haina AI, who founded Freshfruit.com, explained the factory's rules for selecting workers. This is because some social events in the past "made everyone very nervous", causing large factories to pay special attention to the psychological state of workers.

Because the screening rules are relatively clear and definite, Liang Gongjun began to explore the possibility of using AI to interview talents for basic positions. Therefore, Liang Gongjun decided to use the Blue Gray Collar AI interview as a starting point and established an AI interview company - Haina AI in 2019, targeting large-scale interview scenarios for basic positions such as couriers, shop assistants, waiters, salesmen, assembly line workers, etc.

So far, Haina AI's interview system has interviewed more than 10 million people. The recruitment process of SF Express couriers, Walmart store clerks, and Luckin Coffee baristas has been added to Haina AI's interview process. Interestingly, even Haina AI uses its own AI interview system when recruiting interns. Haina AI has also become one of the few companies in the AI ​​field to implement large-scale applications.

"Haina AI interviewer breaks down a talent into nearly 200 dimensions and 4,000 behavioral characteristics. An originally vague candidate can be quantified into an accurate data set after being judged by an AI interview system that works like an X-ray machine. This data set includes dimensions such as appearance, physical fitness, expression ability, psychological state, general qualities, job matching, and professional skills." Liang Gongjun introduced the method of their AI interview system to evaluate candidates.

Liang Gongjun believes that AI interviews are like college entrance exams, with unified standards. From this perspective, compared to human interviewers whose emotions and personal subjective ideas may affect the results, AI that has abandoned personal biases will be fairer.

With the entry of listed recruitment companies such as Boss Zhipin and Liepin, this track has entered a chaotic period, and the competition is becoming increasingly fierce. In this 10 billion track, he expects that at most three AI interview companies will remain in the future. The final situation is likely to be similar to that of the United States. As a neutral developer, Hirevue occupies 80% of the top 1000 groups with the largest number of employees in the United States. Other suppliers can only occupy part of the market of small and medium-sized enterprises due to their respective ecological positioning issues.

Liang Gongjun, who has experienced several industry cycles, will always "hold on". Looking through all his social media, I found that he has a kind of tough guy persistence, two of which are:

"Live like a cockroach, smiling when others want to cry."

King of the Kings

Liang Gongjun's marathon entrepreneurial history originated from the Internet 1.0 era when many heroes fought for supremacy. In 2005, he left Microsoft to start his own business with great enthusiasm, but the company ran out of funds after only two years. After the failure, he became more pragmatic. In 2007, he founded Xianguo.com, an online personalized information website with 30 million users at its peak, and obtained investment from Shanda Group and KPCB. In 2019, he finally found "the one" - he determined that Haina was the most important thing for him to do at this moment, "I have to do this, and this thing has to be me." After experiencing the previous storms, he became more indifferent. For the "small storm" of the products developed based on foreign large models being removed from the shelves in early 2023, it only took "a second to achieve inner peace."

One of the reasons why I have been able to persist from 20 years ago to now may be that "no one is more competitive than me."

Liang Gongjun, who started communicating with me at nine o'clock in the evening, described himself as the "king of the king of files". On that day, he still worked continuously from nine in the morning to nine in the evening. This is his daily routine of "making progress one step at a time", and it is also an important factor that he believes can enable him to persist until now.

Liang Gongjun, who was born as an engineer, was very enthusiastic and talkative. When he was excited, he talked non-stop. Our conversation was extended from the original one hour to three hours, but we still felt that it was not enough. He came from a pure technical background, but he also had a literary and philosophical side. In the end, we talked about Hegel's "Logic", Nietzsche's "Thus Spoke Zarathustra", and Jin Yong's "Flying Snow and Shooting White Deer, Laughing Book of Heroes Leaning on Green Mandarin Duck".

He cited the easy-to-understand party history to explain the company's strategic choices at some key nodes: he compared the acquisition of benchmark customer SF Express to the turning point of the "Zunyi Conference", and the first two or three years of entrepreneurship were like the darkest moment of "how long can the red flag be raised". He called the "small achievement" of interviewing 10 million people so far "the initial establishment of the Shaanxi-Gansu-Ningxia base".

As a marathon entrepreneur, he had figured out many of the real problems of this startup, and most of the developments were carried out according to his strategy. However, he had originally expected to "end the battle in three years" - to take over most of the domestic blue-grey collar market share. However, due to the complexity of the blue-grey collar market itself, Haina finally saw a substantial growth explosion in the fifth year of its establishment.

In the past six months, Liang Gongjun has clearly felt the change in the water temperature: in the past four years, it took Haina nearly a year to sign a contract with a Fortune 500 company, but in the past year, this cycle has been shortened to about two weeks.

All in Blue Collar

Liang Gongjun first came into contact with AI when he started investing in AI+SAAS in 2015. He systematically looked at 20 tracks, including the human resources track, which he believed had great potential.

By comparison, he discovered the hidden potential of this track: the market value of the top 20 US human resources listed companies is more than 5 billion US dollars. In 2016, there was not even a company in China with a market value of more than 5 billion US dollars. At the same time, China's entire labor market is nearly twice that of the United States. This means that in the next 20 years, China will definitely have about 19 human resources listed companies with a market value of more than 5 billion US dollars, and there will even be a giant company of the ecological level like ADP with a market value of 100 billion US dollars.

But at that time he found that some of the investment projects did not work. Years of investment experience gave him the judgment that "a duck knows first when the river water is warm in spring", so he decided to do it himself.

Different from some toVC startup projects, a large part of the nearly 100 million yuan of funds burned by Haina so far is Liang Gongjun's personal investment, and part of the funds come from his old friends - executives of several listed companies.

In February 2019, Liang Gongjun, Dr. Zeng Nanshan, Vice President of Azure AI Research at Microsoft, and Steve Jiang, former global partner of Korn Ferry International, jointly invested in Haina AI. The team also includes colleagues from Xianguo.com and Microsoft.

After studying the five-year financial reports of all human resources listed companies in the United States, he found three core points of the domestic HR industry: First, the industry scale is extremely large, reaching trillions. Second, it is fragmented, and there is no unified standard for recruiting personnel for various positions. Third, the efficiency of human resources is low, and the recruitment process is cumbersome and inefficient.

Since the first day of Haina's establishment, he hoped to solve the latter two pain points by standardizing talents in various positions. What really excites him is that if this thing is done, it will create huge value for the whole society. This is because of a cruel reality that few people disclose: "Team talents need to be selected more, and it is difficult to cultivate them by themselves." Therefore, it is particularly important to improve matching efficiency and find the right people.

The first major decision Liang Gongjun faced was the choice of the company's route. He believed that the barrier of this track did not lie in the underlying technology itself, but in the complex engineering advantages under the superposition of industry experts' know-how and massive data. To gain engineering advantages, we must continue to focus on a specific vertical scenario. In this case, the choice of the main route is particularly important.

AI interviews have several scenarios, including blue-collar recruitment, campus recruitment, and white-collar recruitment. At that time, HireVue, a giant American AI interview company, was gaining popularity, so it almost became a consensus in the domestic industry to follow the campus recruitment and white-collar recruitment routes that HireVue had first entered. Interestingly, among the more than 30 AI interview companies in China at that time, almost all of them chose the campus recruitment and white-collar interview route, just like HireVue.

However, Liang Gongjun believes that the situation in China and the United States is different, and the route of the industry's "big brother" cannot be directly copied. HireVue's approach is because the blue-collar market in the United States does not show the trend of "large-scale head chain concentration" like China, and there are many groups with more than 100,000 blue-collar workers, especially in China in recent years. Chain groups with thousands or tens of thousands of stores are everywhere. In addition, the number of people in each white-collar position is relatively small, and job hopping is far less frequent than that of blue-collar workers.

Giant groups that are rapidly expanding their recruitment need to use unified standards to recruit employees, and AI interviews naturally have the attribute of standardizing talent.

Based on these judgments, Liang Gongjun chose the blue-gray collar route of All in to surround the cities from the countryside.

Challenges and opportunities

Although Liang Gongjun had started a series of businesses before, when he really embarked on this path, problems beyond his expectations began to surface one after another.

The challenge is that in 2019, the large chain groups that Blue and Gray Collar Recruitment targets had almost zero acceptance of AI interviews, and there has never been an example of a super-large enterprise switching to AI interviews. In contrast, foreign companies, potential clients for campus recruitment and white-collar recruitment, are more accepting of AI interviews.

During this period, Liang Gongjun experienced a period of exploration in the darkest moments, and the bitterness of it was "not something that can be told to outsiders." He regarded the company as an independent life form and could see the way out and direction in the future, but for his companions who followed him, it was a tormenting and unclear road.

At the same time, we also encountered a lot of noise. Some competitors got hot money and competed at very low prices.

He quoted Clausewitz's words in On War to motivate himself: "Great generals are those who, in the midst of darkness, take out their own hearts, light them, and illuminate the troops with their tiny lights."

Until now, what he has done most is to build consensus, both internally and externally.

The biggest obstacle is: Is it really necessary for the company's decision-makers to do this? Is it necessary to do it? Because it will change the existing business processes of many departments. Promoting this in a large group with hundreds of thousands of employees is as challenging as "conducting institutional reforms in an empire."

But he insisted on taking on the most difficult task, and first dealt with the leading companies in each industry. Liang Gongjun believed that the CEOs of the leading companies in the industry were more likely to "dare to be the first in the world" to try revolutionary new things. At the same time, the selection of these lighthouse customers has a strong transmission effect. When choosing suppliers, companies in the industry tend to choose suppliers that have been verified by leading companies for many years. Therefore, the market share of the leading customers in each industry is another important barrier in this track.

In July 2020, many start-ups were hit hard by the epidemic, but Liang Gongjun ushered in a key turning point of the "Zunyi Conference".

During the epidemic, although transportation was inconvenient, the demand for courier recruitment increased. Couriers could no longer rely on the previous offline centralized interview method, which gave AI video interviews an opportunity to expand rapidly. Haina also won several industry TOP1 benchmark customers at this critical juncture.

He emphasized to me that AI should replace simple, repetitive, dangerous, cumbersome and dirty jobs, freeing people from these jobs. As for the whereabouts of these people, Liang Gongjun, who has always been rational, still gave his solution based on the first principles: "Upgrade these released laborers to mid-to-high-end service positions that require more human warmth, such as the elderly care industry, which will increasingly need a large number of service personnel in the future."

Currently, AI interviews are in a stage of rapid development, with many players entering the market. However, in the economic downturn cycle, competitors do not have the hot money support of the previous 10 years and will be relatively more rational. They will not fall into the endless and bottom-line low-price competition similar to the previous 10 years.

In Liang Gongjun's view, "the economic crisis is also the best window for the entire industry to return to professional collaboration. The upstream and downstream of the human resources industry should perform their respective duties and make the most professional links they focus on, which is a blessing for the entire industry."

The biggest problem in this field right now is not these potential competitors, but the overall shrinkage of the job market, which has led to a decline in demand from potential customers.

An employee of a large technology company revealed: "Last year, our department opened dozens of positions on the official website, but we actually didn't plan to recruit anyone."

Using AI to cut into the interview stage of the recruitment process is a major issue related to people's livelihood, but AI interviews are still unable to cut into the core scenarios of high-end recruitment. Therefore, the scale of this track is not large. Liang Gongjun conservatively estimates that the total scale of the domestic track is between 5 and 10 billion.

When I asked, "If you have finished the domestic market in one or two years, what will you do next?" He pondered for a long time and turned the topic to going overseas.

There are many labor-intensive countries along the Belt and Road Initiative. Vietnam, India, and Indonesia all have a labor force of over 100 million. Liang Gongjun believes that the king of the kings who emerged in the highly competitive environment of the domestic application layer has a complete talent supply chain and cost-effective advantages. In the future, his goal is to serve Fortune 500 companies overseas. "Now that the company is more mature, we can do some expansionary and aggressive actions."