Innovation·Inquiry —— Interview with ZHAO Yue of SEER: From Building Robot Brains to One-Stop Platforms

What does an excellent enterprise look like, and what characteristics do successful outstanding individuals possess? On their journey of struggle, what "pitfalls" should they be aware of, and what is the most important change?

Chuang·Wen posed questions to some outstanding Huachuang entrepreneurs, investors, and industry experts, and also hopes to share their thoughts with you.

The protagonist of this issue is Zhao Yue, the founder & CEO of SEER Group. SEER Group is an intelligent robot company centered around control systems. Relying on its leading technology and market position in control systems, it has become the world's only large - scale open platform for intelligent robots, covering more than 20 industries such as automotive, new energy, 3C, semiconductor, engineering machinery, and medical. Its products and services are available in over 65 countries and regions around the world.

Huachuang Capital jointly led the Series A financing of SEER Group.

The most complex object in the world is probably the human brain, which consists of at least one hundred billion neurons and determines our consciousness and actions at every moment.

The controller is the "brain" of the robot, determining its perception, decision - making, and execution. Humans have created robot "brains" that can learn and think far faster than themselves, even though they still know very little about their own brains.

What Zhao Yue, the founder & CEO of SEER Group, does is to install "brains" in robots and enable them to understand each other's "languages".

However, he chose the most fundamental yet most difficult path. Zhao Yue described that in the early stage, this path involved a lot of dirty, arduous, and tiring work - for example, one had to have seen at least 1000 types of robots and know what problems they would encounter in tens of thousands of scenarios.

In the non - standard industrial market, only by creating standardized products can a company expand its scale. This is one of the reasons why Huachuang Capital invested in SEER Group in the early stage.

Since its establishment in 2020, SEER Group has become the world's largest intelligent robot company centered around control systems and has ranked first in global controller sales for two consecutive years.

Based on the controller, SEER Group has also built an open platform for intelligent robots. By providing a one - stop solution of "controller + software + robot", it lowers the industry threshold for intelligent robots.

Perceiving and judging in a complex environment is probably the biggest challenge that both robots and entrepreneurs face when making decisions. In Zhao Yue's view, stepping into pitfalls is inevitable, but it will turn into new knowledge, enabling them to seize more opportunities in the trillion - scale embodied intelligence market.

Recently, Zhao Yue was a guest on "Niubaiding" and had an in - depth conversation with Chen Huan, a partner at Huachuang Capital. They retraced the key decisions and the underlying thinking logic of SEER Group in the past few years, from the positioning of the controller, the challenges of standardization to the future space of platformization. From this, we can see how a startup company can grow into a new platform and a new species in the most hardcore and difficult direction.

The following is the content of the conversation from Huachuang Capital's podcast "Niubaiding", sorted out by the editorial department of "Chuang·Wen". The article has been abridged. You can listen to this episode for the full content.

Q: Chen Huan, Partner at Huachuang Capital

A: Zhao Yue, Founder & CEO of SEER Group

Enabling Robots to "Converse" with Each Other

Huachuang Capital: There are a large number of robot startups in the market now. How did SEER Group initially find its unique positioning in making controllers?

Zhao Yue: Actually, at the beginning of the startup, we also made robots ourselves. From equipment design, scenario adaptation to algorithm adjustment, we handled robot projects case by case. But as we went on, we found that customers in each industry had some specific requirements. At that time, the entire domestic automation industry was booming, and many companies were providing customized or industry - specific solutions.

We thought that if these companies had a set of universal controller products, could they build their own robots more efficiently? Through such exploration, SEER Group gradually found its positioning.

Huachuang Capital: At the startup stage, had you already determined that the industrial scenario was a more suitable entry point?

Zhao Yue: It took us about half a year to a year of exploration to focus on the industrial scenario.

Because the founding team of the company came from Zhejiang University. At that time, as the team leader, I led my classmates to participate in the RoboCup Small - Size Robot Soccer Tournament and won the world championship twice in a row. Out of interest and passion, we decided to start a business in the robot field.

To be honest, in the early stage of the startup, we were quite blind and had no clear plan or system. We took on any project that came our way and tried various types of robots such as food - delivery robots and inspection service robots. We encountered all kinds of ad - hoc requirements. For example, when making service robots for banks, some required the robot to greet customers with both hands, and some required voice dialogue. It was difficult to unify these requirements.

In contrast, industrial robots have definite functional requirements, can truly solve customers' problems, and allow us to control precision, efficiency, and safety. This is more suitable for us engineering guys, so we gradually shifted from making robots to focusing on making controllers in the industrial field.

Huachuang Capital: SEER Group is now the world's largest intelligent robot company centered around control systems. How should we understand the controller you make, and what problems of robots does it solve?

Zhao Yue: The controller we make is equivalent to the brain of the robot. It is responsible for three parts: First, perception. It can be connected to lidars, cameras, and surrounding sensors and is plug - and - play. Second, intelligent decision - making, which can execute and decompose tasks. Finally, execution of instruction issuance.

Before SEER Group launched its controller products, automation companies in the market had two options: either build their own algorithm teams, but the R & D cost was high; or directly buy off - the - shelf robots, but when customers had customization or modification requirements, these robot companies could not fully meet them.

We chose a third path - by creating relatively standard and universal controllers, we enabled everyone to build their own robots on this basis.

Huachuang Capital: The industrial market has diverse and trivial demands. How did you verify that your startup direction was correct?

Zhao Yue: We still need to make judgments from the perspective of the entire robot industry chain. The biggest pain point in the industrial robot industry is that factories need multiple types of robots to work together, but no single company can provide all types of robots. This supply - demand contradiction needs to be solved by platform - type companies.

80% of our customers are integrators. After buying our controllers, some are good at making intelligent forklifts, some at making cleaning robots, and some at making special machines for a certain industry. However, they found that factories have diverse and customized requirements and need robots for various scenarios, but no single company can meet all these needs.

So, on the basis of the controller, we built the world's first large - scale open platform for intelligent robots, providing a one - stop solution of "controller + software + robot" to downstream integrators and end - customers. We have accumulated thousands of different robot models. When customers want to build their own robots, they can use our controllers; when they don't want to build themselves, they can directly buy from SEER Group's platform.

Our role is to empower integrators or end - customers. Around the controller system, different robot companies form an ecosystem on this basis. They can have different combinations and provide various solutions to end - customers.

Huachuang Capital: It seems that you want to break down the barriers and thresholds between different end - customers?

Zhao Yue: Yes. When there are enough customers using our controller system, we found that although Customer A makes intelligent forklifts and Customer B makes cleaning robots, there will also be combined transactions between A and B. At this time, they hope that in the same scenario, the standards for right - of - way occupation, map format, API interfaces, etc. of the two types of robots are consistent; otherwise, they will definitely conflict.

Therefore, a unified standard is needed to build a platform. Otherwise, we would just be a "grocery store". Customers can buy the desired categories and models from SEER Group, but if the robots don't "recognize" each other, it may not make much sense.

Huachuang Capital: What are the benefits if robots can "converse" with each other?

Zhao Yue: If they use the same control system, it means that the controllers can communicate with each other. Even if the robots have different appearances and structures, it doesn't affect their communication. Just like a Chinese person and an American. Even if they are both very smart, if they don't speak the same language, there will be misunderstandings during the translation process.

So what we can do is to make different robots consistent at the brain - level, thereby reducing the development costs for downstream integrators and customers. As we gradually build up the ecosystem around a large number of controller applications, and finally become a real one - stop robot platform, all processes from robot development, acquisition, selection to final application can be completed on this platform, which will greatly lower the threshold for building robots.

Huachuang Capital: This was exactly what Huachuang Capital valued highly when investing in SEER Group in the early stage - finding relatively standardized products in the non - standard industrial market so that the company can expand its scale. However, this also brings huge challenges. How do you make your controller more compatible?

Zhao Yue: Actually, the hardest part is not coming up with this direction but being willing to take action because there is a lot of dirty, arduous, and tiring work on this path. This is the first hurdle.

If you only focus on making robots, you only need to ensure the safety of the supply chain and select a lidar that offers good cost - performance. You don't need to adapt to so many sensors. However, once you want to turn the controller into a product, you have to adapt to 80% of the lidars on the market. Because some customers pursue cost - performance, while others pursue high performance. Considering brand, protection level, cleanliness, and other requirements, you need at least dozens of types of lidars. So the most difficult part in this process is continuous adaptation, and you have to tackle each one.

The second hurdle is time. Many application parameters, processes, and configurations cannot be fully collected at the end - user level by ourselves alone. We have to rely on the feedback from a large number of integrators. Only when a customer has no other choice will he be willing to tell you all his requirements without reservation. At the time when we started our business, few companies in the market were willing to provide products or services to a large number of automation companies, so we had to provide the control system.

I remember when we first started cooperating with Foxconn, they asked me if I could open all the BOM lists. At that time, others were not willing to do so. But we gave our item usage lists and even sales contact information to downstream customers, and gradually gained the trust of integrators.

It was by overcoming these two hurdles that we built the barriers for SEER Group today.

Huachuang Capital: Making controllers is so difficult. Have there ever been moments when you were tempted, for example, when a customer put forward a big demand and you didn't have to work so hard anymore?

Zhao Yue: Whether it's temptation or opportunity, in fact, since the day we decided to make controllers, we have been facing these problems every day.

Such choices are bound to be difficult in the early stage of the startup. If you ask a salesperson, which is more fulfilling, selling something for 5000 yuan or 500,000 yuan? Other companies can sell a single robot for hundreds of thousands of yuan, while we only sell controllers for ten to twenty thousand yuan. Who would find it exciting?

The R & D colleagues also didn't understand this. So what if we adapted to 50 brands of lidars? Would it improve my R & D technology? Even if I became an expert in lidar adaptation, it didn't seem to make much sense.

We often received feedback from customers that there were problems with the products. When we went to the site, we found that it was just a wrong parameter setting. The project team members thought it would be better if we made the robots ourselves.

So in the early stage of the startup, the doubts often came from within the company rather than external temptations.

Huachuang Capital: How did you get through this stage?

Zhao Yue: It can only be achieved through determination. First, at all levels of the company, we made everyone feel that what they were doing was valuable and enhanced their sense of achievement. Second, since we were a student startup team, our understanding of the industry was not deep enough. We couldn't handle everything on our own at the beginning. Assisting downstream partners to do things well was the most suitable approach for us at that stage.

At the same time, we also set higher and more challenging goals within the company to motivate everyone. Once a closed - loop is formed, things can gradually get on track. At the beginning, making controllers was indeed a bit slow, but as the number of customers increased, the network effect began to show.

At first, to enhance our understanding of scenarios, we would work on projects with customers or directly assist them on - site. After about half a year, the first batch of users started to make stable repeat purchases, and this model gradually became viable.

To date, we support over 1500 customers, have deployed over 1000 types of robots in more than 20 industries such as 3C, automotive, and new energy, and have ranked first in global robot controller sales for two consecutive years.

The Ability to Abstract Customers' Requirements

Huachuang Capital: In the industrial field, many companies also want to make standardized products, but they may end up becoming companies providing non - standard solutions. In this process, how do you balance and make trade - offs regarding customers' requirements?

Zhao Yue: Even in the same scenario with the same requirements, different customers may have completely opposite demands.

Take the simplest example: What should a robot do after an emergency stop? Some customers, out of safety concerns, require the robot to be powered off; some think it's okay to just lock the brake; and some think that after an emergency stop, a person should at least be able to push the robot. For such a simple requirement, different customers consider it from different perspectives.

So our usual approach is to preset multiple possibilities and turn them into selectable parameters and configurations, allowing users to make their own choices.

Huachuang Capital: What requirements does this place on your own team?

Zhao Yue: Actually, it tests our ability to abstract customers' requirements.

Just like the Word software, we may use less than 5% of its functions every day. But why do we still need the other 95% of the functions? It doesn't mean there must be 100 requirements, but 5 or 10 methods can cover them.

We can't provide a customized version for each customer. Instead, we let them choose freely within the software, making our software both simple to use and feature - rich. As long as a customer thinks of a certain requirement, they can quickly obtain the corresponding function from a certain page. This is the ultimate solution.

In addition, for some overseas customers, we also offer low - code or no - code methods. Users can quickly generate execution scripts for robots through natural language and quickly make the robots work. In the future, with the iteration of new technologies, we will also upgrade our methods.

Huachuang Capital: In theory, for a controller to be very universal, everything should be configurable.