Summary of the Key Points
This interview focuses on Li Tong, the founder of Qinglang Intelligence. The core message is that robots are not part of a science fiction tale; they represent a business solution to the issue of labor shortages. Li Tong insists on implementing a "job-specific" approach, meaning robots should first perform specific tasks (such as food delivery and cleaning) rather than striving for versatility. By leveraging 16 years of data collected from real-world scenarios, Qinglang has established a competitive advantage. The company uses the RaaS (Robot as a Service) model to lower the barriers for customers to adopt these technologies. Li Tong emphasizes that moving from early-stage technological experimentation to commercial success requires focusing on practical results and calculating return on investment (ROI). He also predicts that the robotics industry will have a longer cycle than the automotive industry, with large companies entering the market once the technology matures, but differentiated positioning will still allow for survival.
1. Job-Specificity: Robots First as "Workers," Not Trying to Be All-Arounders
Li Tong strongly opposes the notion of robots being all-around solutions. Human society is organized around specific roles—people are chefs or servers, and so should robots be too. For example, in a McDonald's collaboration, Qinglang deployed four types of robots: one for delivery, one for cleaning, and two humanoid robots for general tasks. The specialized robots handle fixed tasks, while the humanoid ones fill in flexible roles, all addressing the specific labor needs of each job. Customers care less about how impressive the robots are and more about whether using them is more cost-effective than hiring employees (ROI).
Why this approach? Home environments are too complex and non-standard, making it take more than five years for humanoid robots to be widely adopted; however, semi-enclosed settings like restaurants and factories have standardized tasks that can be implemented quickly. For instance, Qinglang's robots in Haidilao help free up staff to provide warmer, more personalized service—robots handle the mechanical tasks, while humans focus on higher-value activities.
2. Data as a Competitive Advantage: 16 Years of Real-World Experience
Some in the industry argue that the bottleneck for embodied intelligence lies in computing power and algorithms, but Li Tong believes that real-world data is the true challenge. For example, in a restaurant, robots need to know how to interact with people walking side by side—fast if they are serving staff, or slowly if they encounter children. Such data on human behavior cannot be replicated in laboratories. Qinglang's 100,000 robots worldwide generate daily data on traffic patterns and environmental interactions, which is a unique competitive advantage.
Data security is also crucial; overseas servers are physically isolated to prevent local data from being transmitted back, which has helped the company gain trust from European and American customers.
3. RaaS Model: Renting Robots as "Employees"
The traditional model of selling robots requires a one-time large investment, which service industry owners find unprofitable. Li Tong introduced a "robot employment" approach where customers sign a three-year contract and pay a monthly fee (e.g., $30-50 per robot, compared to the cost of hiring a human). This reduces the initial barrier to adoption. Essentially, Qinglang acts as a labor outsourcing company, responsible for maintaining and servicing the robots while customers focus on using them. This model requires establishing service systems overseas and collaborating with local partners; otherwise, broken robots become useless.
4. From Tech Expert to Business Leader: Learning from Mistakes
Initially, Li Tong believed in the power of technology and developed highly realistic robots that caused panic (even security guards were frightened). He also tried selling products for haunted houses and science museums but struggled financially. At his lowest point, two of his four partners left, and he was still working on business deals on New Year's Eve, taking three long trips home. The turning point came when Qinglang secured a large order from Haidilao. After many failed attempts, Qinglang's robots proved capable of operating in the complex restaurant environment, proving their practical value to the market.
This experience taught him that entrepreneurship is like building a barrel—success depends on all aspects (technology, sales, supply chain, legal compliance). A tech company that doesn't focus on commercialization is simply wasting its potential.
5. The Future of the Industry: Long Cycle but Huge Potential
Li Tong predicts that the robotics industry will have a longer cycle than the automotive industry, as robots operate in more complex environments with higher technical demands and larger market potential. When will large companies like Huawei and BYD enter the market? It won't happen until humanoid robot sales reach tens of millions per year (what he calls the "Huawei moment"). However, there's enough room for differentiation; companies like Qinglang, by focusing on specific use cases (e.g., service robots), can thrive. His main concern is ensuring a steady cash flow, as any company can fail due to lack of it.
Conclusion
Qinglang's story highlights that technology must be practical and address real-world problems. By starting with job-specific applications, accumulating data, and innovating business models, Chinese tech companies can gain a foothold globally. The key is to combine technical expertise with a commitment to the fundamental principles of business.