Artem Sokolov, Founder of Humanoid – Interview Series
New Post has been published on https://thedigitalinsider.com/artem-sokolov-founder-of-humanoid-interview-series/
Artem Sokolov, Founder of Humanoid – Interview Series
Artem Sokolov is the founder of Humanoid, as well as a global investor and entrepreneur. He successfully took over his family business and grew it to a $1 billion valuation. He later founded Humanoid to build safe and reliable humanoid robots that free humans from physically demanding work. Today, he leads a team of over 130 professionals from some of the world’s leading tech companies, bringing world-class technical depth to engineer the future of human–machine collaboration.
Founded by Sokolov in 2024, Humanoid is a UK-based robotics innovation company dedicated to developing advanced humanoid robots that enhance human capabilities. With offices in London, Boston, and Vancouver, the company is focused on creating commercially viable, scalable, and safe robotic solutions for real-world applications across industries.
What inspired you to launch Humanoid, and what led you to choose a humanoid form factor over other robotic designs?
My personal background played a big role in shaping what I’m doing today. My grandparents spent their entire lives working in jewelry manufacturing – with long hours, from early morning until late at night, they didn’t see the world at all. As I scaled my jewelry business, I saw how repetitive work negatively affected people.
That experience inspired me to start Humanoid: to build safe, reliable, and helpful humanoid robots that free people from hard and monotonous work.
Why humanoids? There’s a lot of debate on this topic, with many arguing that industrial robots are a more efficient, mature and deployment-ready technology.
During my years as an entrepreneur and investor, I had a realization while observing a warehouse design and process: every shelf, workstation, and piece of equipment was fundamentally designed around proportions of a human body and human capabilities.
The world is designed for humans and instead of reinventing our entire industrial infrastructure to accommodate robots, it’s the humanoid robots that should naturally fit into these human-scaled environments.
This simple idea evolved and became the catalyst for launching Humanoid, with the mission of empowering humanity by building the most reliable, safe and helpful humanoid robots.
Moreover, unlike industrial robots designed for specific tasks, the human form is incredibly versatile. We can walk, run, grasp objects, use different tools, and much more. Humanoid robots have the potential to perform a wide range of tasks and continuously learn. So, instead of purchasing a new robot for every task, a humanoid can simply be taught new skills. It makes them far more flexible.
With companies like Tesla, Agility Robotics, and Figure AI entering the humanoid robotics space, what differentiates Humanoid’s approach?
It’s true, there is no shortage of competitors, and some of them have already achieved impressive breakthroughs. But I believe that competition, in our case, is a great thing. It drives the entire market, shapes expectations, creates demand, and eventually drives the cost down as well. Right now, we’re all building this industry from scratch – there isn’t yet a successful product example to follow.
Global trends like labor shortages and aging population are making it clear that the demand is massive. I don’t think there will be just one ultimate winner in this humanoid race. Instead, multiple companies will emerge, collectively meeting the demand and shaping the future of the industry. This is a marathon, not a sprint.
We’re developing Humanoid with a focus on practical, market-ready solutions rather than just doing interesting robotics research. We’ve seen many videos of robots doing backflips, dancing, or playing sports on social media – but the question is, what’s the purpose? Warehouses or production lines don’t need robots that can dance.
Our robots are designed for mass commercialization, so our main priority is to bring them from labs to the real world as fast as we can. Our approach prioritizes making Humanoid robots both affordable and immediately deployable in real-world settings. We’re starting with pick-and-place use cases, for example, picking individual items from a shelf and placing them into a tote. Then, we’ll move to more complex tasks like visual checks or assembly in manufacturing.
Our time-to-market is incredibly fast compared to competitors. Founded in May 2024, we are already moving with commercial testing, just one year after launch.
To achieve this goal, in just one year we’ve built a team of more than 130 experts in humanoid robotics. We benefit from a second-mover advantage. Our team has worked at top robotics and AI companies, so we’ve seen what doesn’t work and now we’re able to focus on what does. That means we can move faster, skip early R&D, and avoid expensive mistakes. To be honest, I see my strength in attracting top talent and building effective teams capable of achieving extraordinary results.
You emphasize the ethical development of humanoid robots. What principles guide your team in ensuring responsible AI and robotics innovation?
The ethics of humanoid robotics is one of the most debated topics in the field today.
We all know Isaac Asimov’s Three Laws of Robotics, and they are still relevant. But they were formulated over 80 years ago, and since then, some entirely new challenges have emerged.
One key question is the balance between autonomy and control. How independent should robots be? Another major challenge is data privacy. By 2035, we expect humanoid robots to become a regular part of households. They will assist with daily chores, provide elder care, or even help combat loneliness. But if robots live with us, they will inevitably collect vast amounts of personal data. How much are we willing to share? Where do we draw the line?
There’s no correct answer because humanity has never lived and worked alongside robots before. These are the questions that the industry will address in the coming years as humanoid robots move from prototypes to everyday life.
At Humanoid, we are starting with industrial use cases, which are much more predictable. This allows us to test various scenarios, refine our technology, and guarantee that our robots are fully safe before introducing them into households.
Moreover, being headquartered in Europe is an advantage in many ways, given the region’s robotics expertise and strong industrial heritage, particularly from renowned technical universities and research institutions.
We maintain rigorous safety protocols and regular ethical audits, while actively collaborating with these experts and stakeholders to ensure our robots enhance workplace productivity without compromising human wellbeing or job security. This is crucial.
At the end of the day, we believe in transparency and the importance of maintaining clear documentation of our development processes to ensure our technology remains accountable and trustworthy.
What kind of AI powers Humanoid’s robots? Are you developing proprietary AI models, leveraging large language models (LLMs), or integrating third-party AI solutions?
We’re developing our own models, but initially, we’re using existing VLAs as a base and building our Application Layer Models on top. Over time, we’re progressing toward a fully proprietary autonomous generalization model.
Our robots can collect and synthesize knowledge related to many different applications, which makes them great for general purposes.
We don’t use traditional robotics stack. Instead, we put Vision-Language-Action (VLA) models and a reasoning system at the very core. This way, our robots don’t just perceive the world – they understand it.
Our AI development timeline is highly competitive. That’s because key breakthroughs in manipulation, scene understanding, reasoning, and vision-language models have only recently become viable, May-June last year. No company has a significant head start.
In the end, I believe that success in humanoid robotics won’t be about having the best foundation model alone. It will come from having the best application-specific data, hardware, and seamless real-world integration. The companies that win will be the ones that execute best – offering reliable products and great customer service. That’s exactly what we’re focused on.
As for the robots’ autonomy, we’re addressing it in three stages. The first one is shared autonomy, when robots can ask for human help when needed. We expect that by 2026-2027, our robots will perform 80% of industrial tasks with minimal human assistance. And by late 2027 they will reach full autonomy.
How does Humanoid’s AI approach loco-manipulation and environment perception to ensure safe and efficient operation in real-world environments?
At Humanoid, our AI approach to loco-manipulation and environment perception integrates advanced computer vision, adaptive locomotion algorithms, and intelligent manipulation systems.
Our robots are equipped with cameras and AI to recognize objects, people, and obstacles. They use cutting-edge sensor fusion and machine learning to build a semantic understanding of their surroundings, enabling safe navigation and interaction in complex, dynamic environments.
This is coupled with a continuous learning framework that allows our humanoid robots to improve over time, making them increasingly capable and efficient partners for humans in various real-world settings.
Modularity is a key feature of HMND 01. How does this flexibility improve real-world deployment compared to other humanoid robots?
We design our HMND 01 robots with modularity at the core, because it’s one of the key features for mass commercialization. HMND 01’s modular design offers flexibility in real-world deployment, allowing for quick reconfiguration to suit diverse tasks and environments.
We’re starting with a wheeled robot because it can reach the market faster – it’s a safer, more flexible solution. It’s also in higher demand right now, especially in logistics, where over 80% of use cases can be addressed with this type of platform.
With the wheeled robot, we can focus on solving a single challenge: manipulation – rather than tackling manipulation, locomotion, and safety all at once.
Both platforms also share the same upper body design – so once the wheeled robot masters certain manipulation tasks, those skills can be easily transferred to the bipedal robot.
Beyond convenience, modularity is also cost-effective. It enables easy upgrades, repairs, and customizations without replacing the entire robot, which reduces downtime and maintenance costs.
The modular architecture helps with rapid innovation and integration of new technologies, making it easier for HMND 01 to adapt to evolving industry needs compared to a traditional, monolithic humanoid design.
Additionally, many people noticed in our first product video that HMND 01 robots could wear different garments. Their primary function is not style, of course, – these garments protect both the robot’s systems and the surrounding environment by reducing contamination and minimizing collision effect. There’s also a strong branding component for clients – they can customize the robot with their brand colors, logos, or other visual elements.
Given that humanoid robots require advanced motion control, how does Humanoid’s robotics architecture ensure balance, agility, and adaptability in diverse environments?
Engineering bipedal robots is a demanding challenge, because walking on two legs is incredibly complex. Humans make it look very easy, but in reality, it requires advanced mechanics, balance control, and precise coordination.
Then, a lab floor is one thing, but real-world environments could be unpredictable – slippery surfaces, gravel, or obstacles all make motion control even harder.
As I’ve already mentioned, we are addressing the locomotion challenge by starting with the wheeled platform first. We also utilize a combination of model-predictive control (MPC) and whole-body control strategies, enabling our humanoid robots to make strategic locomotion decisions like walking based on data from the surrounding environment.
Our architecture incorporates learning-based control strategies that enable the robots to improve their balance capabilities over time. The robustness and precision of our control systems enable HMND 01 robots to navigate confined spaces with accuracy.
What do you think will be the biggest societal challenge as humanoid robots become more integrated into everyday life?
I see a lot of discussions, especially on social media, about humanoid robots taking over jobs, but the whole idea of ‘Humans versus Machines’ is fundamentally wrong. Humanoids won’t replace humans, they will work with us, fill labor shortages and create entirely new roles around their use.
Remember when people feared the internet would make workers obsolete? Instead, it triggered a major societal shift, transforming workplaces and making digital skills essential for most professions. I see the same future for humanoid robots. But shifting public perception will take time.
Additionally, as a society, we will need new frameworks and workplace regulations to define how we interact with robots as colleagues. We’ve never coexisted in this way before, so it will require entirely new rules and behaviors.
As humanoid robots become ubiquitous, I think it will also change a lot of things about how we perceive ourselves and our place in the world.
There may be some resistance or existential crises, but I’m certain it will lead us to explore new frontiers of human potential and creativity.
Scaling robotics hardware can be costly. What is Humanoid’s commercialization strategy to make humanoid robots more accessible and cost-effective?
Compared to industrial robots, humanoids may seem like a less cost-effective solution. However, as production scales up prices are already coming down: according to Goldman Sachs, the average price has dropped from $250,000 per unit to $150,000 per unit.
Of course, current pricings and costs reflect the early, non-mass manufacturing stage. As production grows and the supply chain improves, we can expect costs to further go down. Some companies are already talking about $20,000 robots, but that’s mostly hype — the market doesn’t really expect prices that low. For industrial and logistics use, a more realistic price range is around $50,000-70,000.
Let’s consider this cost in the context of labor expenses. According to PwC, with a conservative lifespan of 3 years, the estimated ownership cost of a humanoid robot would be around $5 per hour. According to our internal estimation, the number is higher, around $10 per hour. Still it is far lower than the average wages for workers in developed countries, even before the benefits, HR costs, and training expenses. This shift could be a game-changer for workforce economics.
It’s useful to look at BOM costs, but when it comes to robots, what really matters is the total cost of ownership (TCO). For Humanoid, low TCO is one of the key product features. We’re looking to enhance capabilities of our robots while keeping hardware costs down, making them more accessible for various applications in industrial settings.
We’re also exploring partnerships and collaborations with tech giants and targeting specific sectors like logistics and manufacturing to drive adoption and achieve economies of scale.
Looking ahead, what are Humanoid’s next major milestones in the development and deployment of your robots?
Humanoid’s next major milestone is the launch of our bipedal and wheeled alpha models later this year. They will feature advanced AI capabilities and modular design for versatility across various industries. Moreover, we are planning to launch pilot projects with partners in logistics and manufacturing. I’m confident that we’ll be announcing some exciting developments and partnerships soon.
We’ve recently launched a product video, offering a first look at our general-purpose humanoid robot HMND 01 in action. The audience’s feedback has been incredible, but showcasing a product in a video is one thing – bringing it to life is another. That is now our main focus, and Humanoid’s hardware and software teams are working tirelessly to make it a reality.
I’m very proud of the team we’ve assembled – they are some of the best professionals in the industry, and with our recently opened offices in the U.S. and Canada, we’re also focusing on attracting the best talent in these regions.
Thank you for the great interview, readers who wish to learn more should visit Humanoid.
