Every morning, hundreds of millions of people around the world shop online. Over 100 million parcels are delivered daily in the United States alone. Behind this massive flow lies a logistics system of astonishing scale. Giant warehouses are lit up 24/7, with millions of workers carrying, sorting, and packaging boxes. The costs incurred in this process account for a significant portion of the global per capita GDP.

However, there is a problem. The pace of technological advancement in the logistics industry lags significantly behind other sectors. While automobiles have entered the era of electric vehicles and smartphones are equipped with artificial intelligence (AI) chips, logistics warehouses still heavily rely on human physical labor. It's a puzzling situation.

In 2017, a quiet change began. A Ph.D. student at Stanford University founded 'Dexterity.' As the company name, derived from the Latin word 'dexter' (right hand), suggests, his goal was to implement the agility of the hand with AI. Initially, no one paid attention. Robotics was already an established field, and logistics automation was an area several companies had attempted. However, this company was not just a simple robot manufacturer.


Dexterity's 'Physical AI' operates in the real world, not in digital space. The core is 'hand intelligence.' The human hand can instantly read the texture, center of gravity, and friction of an object without prior instruction and adjust its movements accordingly. By recreating this intuition with mathematical models and data learning, Dexterity's robots can stably grasp irregularly shaped objects, realign their posture in unexpected collisions, and choose various grip strategies. The goal was to create robots that recognize situations, learn independently, and adapt to unpredictable environments, rather than simply repeating programmed actions. Thus, Dexterity's robots excel in volatile warehouse and fulfillment environments rather than standardized lines.

Many were skeptical at the beginning. Logistics automation was a field that even giants like Amazon were slowly advancing through their own robotics departments. It seemed difficult for a startup to succeed. But coincidences overlapped. The COVID-19 pandemic hit, shaking the global logistics system. Paradoxically, the pandemic made everyone realize the necessity of logistics automation. The vulnerability of supply chains was revealed, and labor patterns changed rapidly. Most importantly, large corporations began seeking new technology partners. Dexterity seized that moment.

Global logistics companies like FedEx and UPS knocked on Dexterity's door. Samir Menon, CEO of Dexterity, sincerely listened to the voices of customers. He showed an attitude of solving real problems faced by customers together, rather than just making simple technology proposals. Complaints from warehouse managers about parcels getting wet in the rain, concerns from technicians about robots overheating in high altitudes, and the urgency of executives who couldn't operate without reliability—all these determined the direction of Dexterity's technology development.

The results were dramatic. In just 3 to 4 years, Dexterity evolved from existing humanoid robots to a completely different form. The two-armed robot named 'Mech' was born. It was a simple structure with two powerful arms mounted on a wheeled box, but it was a technological breakthrough that no one had achieved in the 50-year history of robotics. It realized the 'holy grail' of logistics, loading and unloading trucks.

This technological innovation also attracted attention in the capital market. Recently, Dexterity secured large-scale investments from renowned venture capitalists and strategic investors such as Lightspeed Partners, Kleiner Perkins, and Japan's Sumitomo. As a result, the company was valued as a 'unicorn' with a corporate value of $1.65 billion and is currently pursuing an IPO in the United States.

His greatest strength is the combination of high technical skills and humility. He prioritizes the customer's perspective in every management decision. He is known for respecting even trivial questions in the field, such as "Can your robot pick up wet parcels when it rains?" He often mentions that he realized the essence of technology only when he came down to the actual warehouse floor after looking at mathematical equations at Stanford University.

The author conducted an interview with CEO Menon at Dexterity's headquarters in Redwood City, California, in August.

― What was the background and process of founding Dexterity?

“I grew up dreaming of robots and AI from a young age. Watching sci-fi movies and works like 'Transformers,' I always pondered, 'How should I create this technology, and how can it contribute positively to the world?' While pursuing a Ph.D. in computer science at Stanford University, I focused on AI and robotics, particularly studying human movements deeply. I believed that understanding human movements properly could transfer that technology to robots, allowing them to perform human capabilities. During this process, I met excellent colleagues and encountered the long history of physical AI and robotics research, which strengthened my conviction.
As I approached graduation, I asked myself, 'Is now the right timing?' Physical AI is a technology that will inevitably become a reality someday. The question was timing. Looking at the market, the opportunity was enormous. NVIDIA's CEO Jensen Huang also predicts that physical AI could account for 40-50% of the global GDP. Even conservatively, it's hard for anyone to deny that it could exceed 10%. It was a sufficiently large market.
We confirmed our technical confidence, the ability to assemble a great team, and the market's readiness. Ultimately, timing was crucial. So, I founded Dexterity at the end of 2017 and started full-scale operations in early 2018. In the end, that choice was correct.
Coincidentally, macro changes like the spread of e-commerce opened up opportunities to commercialize the technology early. An environment that could excite customers was created, and we were able to start robot innovation by developing world-class physical AI. That was the starting point. Even now, we continue this journey with great expectations and excitement. The opportunities are growing, and the pace of technological advancement is beyond imagination. The achievements AI is showing now were unimaginable just five years ago.”

― Many robotics startups focusing on logistics automation emerged in the 2010s, but few succeeded. How did Dexterity grow rapidly?

“Our success strategy can be explained in three main pillars. First and foremost is 'customer-centricity.' Most founders of robotics startups come from a technical background, starting with the thought, 'This technology is amazing,' and then pondering, 'Where can we use it?' We completely reversed this order. We first asked, 'What are the problems of the customer and market?' and then thought, 'What technology should we apply to solve this problem?' This shift in perspective made our company operate entirely differently.
Physical AI robots can be applied in various fields such as dishwashing, mining, and warehouse logistics. However, we deeply analyzed the economic structure of each market and first identified where commercialization was actually possible. The biggest limitation of current robot technology is environmental uncertainty. Today's robots excel in precision and reliability in environments that are perfectly controlled, like automobile manufacturing plants.
However, building such an environment requires enormous costs. Even if you invest $100,000 in the robot itself, you need to invest over $1 million more in high-precision conveyor belts, flat concrete floors, and alignment devices to create a precisely controlled environment. Such environmental setup is only possible in manufacturing plants. Outside of manufacturing, robots become useless. The core problem we had to solve was 'creating robots that can operate amidst uncertainty,' in environments like logistics warehouses that are not perfectly controlled.
The second is the 'Full-Stack Approach.' We started with physical AI and software technology, but to completely solve customer problems, we needed to integrate not only software and AI but also hardware. So, we closely collaborated with hardware partners to build a complete solution. We had to identify customer problems, solve them, and simultaneously grow into a scalable business. Software alone is not enough to achieve all this.
The third is 'scalability.' The future development of physical AI is still uncertain. Looking at the evolution of digital AI helps understand why. Initially, chatbots appeared, then neural networks, and GPT transformers emerged. Dozens of approaches were tried before, and new technologies continue to emerge every year. In this process, transformers have established themselves as core technology.
Physical AI will undergo a similar process. It's currently unclear which technology will become the core. Therefore, we built a platform that can accommodate various AI approaches. Whether it's transformers, recurrent neural networks, graph neural networks, or physics-based simulations, we have a flexible structure that can apply any approach. Ultimately, I believe Dexterity succeeded in differentiation through three principles: solving customer problems, full-stack approach, and building a scalable platform.”

― Although Kiva Systems, acquired by Amazon, introduced logistics robots in the 2000s, the acceleration truly began after the COVID-19 pandemic. How has the logistics automation environment changed when comparing the 2010s to the post-COVID era? How did Dexterity respond to this change?

“When starting a company, four elements are needed: market, technology, team, and timing. We founded in 2018, and shortly after, the COVID-19 pandemic hit. It was a massive event and created three turning points for us.
The first is that the 'vulnerability of the logistics supply chain' became globally clear. It was revealed that a single major event could severely disrupt the entire supply chain, directly affecting our survival. As a result, tremendous attention was focused on logistics and supply chains, making it crucial to make them more robust and resilient. Naturally, enormous demand emerged.
We initially identified logistics as a core market. We saw logistics as the first market that could elevate the intelligence level of physical AI. Simultaneously, logistics was the largest initial market. The pandemic further reinforced this judgment and created a strong motivation for customers to strategically collaborate with us.
The second is the 'collapse of labor patterns.' As remote work spread, many people fundamentally changed their way of life, choosing more distributed forms of employment. This change is still affecting various parts of our economy. Notably, there was a large-scale withdrawal of female labor. Female labor force participation significantly dropped during the pandemic and has not yet fully recovered.
This change posed an important question: 'Do humans need to continue doing stressful, injury-prone, and physically demanding work?' This question led to a very positive societal perception of physical AI and robotics. Initially, there was significant societal resistance to physical AI entering people's daily lives. However, COVID-19 ironically created a consensus that 'we need the help of machines.' Social acceptance noticeably increased. As a result, customers began actively embracing us.
The third is 'capital inflow.' During the pandemic, enormous capital flowed into the market, and Dexterity focused that capital on solving customer problems. We believed that 'capital should be used to solve customer problems,' and we have indeed done so.
Personally, I think the most important aspect is the 'social response.' When I go out and show people the robots we've created, they smile when they see "this crazy Mech robot, this Transformer-like robot." I believe that smile is more important than anything else. That is the reason for our existence.”


― What was the biggest turning point in Dexterity's growth into the company it is today? Were there any 'aha' moments, failures, or successes?

“There have been countless turning points at Dexterity. It's really hard to pick just one. But there's one unchanging truth. The most significant turning point is when customers genuinely believe in our technology. That is more important than any other turning point.
Among them, the event that had a tremendous impact on us was when our key customer, FedEx, approached us with the problem of truck loading and unloading applications. Until then, we had solved problems using commercially available hardware. Our focus was thoroughly on building physical AI capabilities, and we didn't put much effort into robot arms or related infrastructure. We would look at the robot catalog of partners like Kawasaki and choose, 'We need version 7 robots this time,' or 'We need RL version 41 robots this time,' and combine them with our AI to succeed.
However, when FedEx proposed the truck loading and unloading problem, the situation was completely different. Now the robot had to enter the truck directly. This meant that the robot couldn't be fixed to the floor or have legs. In the large box space of a truck, a robot with legs would inevitably fall. Moreover, the required performance was almost superhuman.
Until then, our company's philosophy was clear. 'We must create robots that perform as well as humans.' For years, we aimed for 'human-level performance.' But in the face of the truck loading and unloading problem, that mindset crumbled. It was something even humans found difficult to accomplish. The conclusion was clear. 'We need superhuman robots, not human-level ones.'
So, over the past 3-4 years, we have made significant investments with FedEx to create 'superhuman' robots. The result was dramatic. The humanoid-like robot evolved into a completely different form. A powerful form of robot, reminiscent of Transformers or the Mechs from the movie 'Pacific Rim,' was born. This was a customer-driven major transformation, and we completely rebuilt the technology to meet that demand.
The 'Mech' robot we recently officially launched is the result of that effort. At first glance, it looks very simple. I often jokingly explain it as 'two powerful arms mounted on a wheeled box.' However, anyone with even a little knowledge of robotics knows how challenging this technology is. In the 50-year history of robotics, no one had successfully achieved truck loading and unloading with a 'two-armed box.' We made that technology a reality.
This Mech robot was created for the market and customers and is solving real problems today. We launched it in March this year, and we've already opened the fifth deployment site. It's currently operating in the United States and Japan and is rapidly expanding. One customer changed our strategy, and we responded to that demand by creating technology that no one had achieved in 50 years. This is the story of Dexterity's growth.”

― The logistics market is not only large but also dominated by companies like Amazon with their own robotics departments, making it extremely difficult for startups to enter. Dexterity succeeded in collaborating with major logistics companies like UPS and FedEx. How was that process? What was the biggest challenge in making large corporations trust a startup's robot solutions?

“That's right. This field is particularly challenging not just because the companies are large but because they bear enormous social responsibility. Companies like FedEx, Amazon, Walmart, Japan's Yosogawa, and UPS are key companies that society relies on. If their networks stop for even a single day, serious problems arise. So, I have great respect for these companies. Ultimately, the key is 'how to earn their trust.'
The first step to earning trust is a genuine effort to understand the customer. You must build respect for the customer and not overlook even seemingly trivial issues.
I earned a Ph.D. in computer science from Stanford University and spent a long time immersed in the world of theory. I lived only within mathematical equations and abstract concepts. But when you enter an actual warehouse, a completely different world unfolds. Warehouse managers ask, 'Can your robot pick up wet parcels when it rains?' There is a huge gap between the technologist in the theoretical world and the operator in the real world. On the surface, it may seem like a simple process of 'the parcel enters here, goes out there, and arrives at my house,' but in reality, it's very complex. You must maintain employee motivation, increase productivity, meet schedules, respond to weather, reflect changing purchasing patterns, and even prepare for breakdowns or power outages. Every day's situation is slightly different.
The reason Dexterity was able to earn trust is that we seriously considered and solved those 'small problems' together with our customers. We focused as much on the problem of 'parcels getting wet when it rains' as we did on mathematical equations.
We do

Popular News