Over the past decade, the robotics industry has long been constrained by hardware costs, limited AI generalization, and the difficulty of deploying robots in real-world environments. Although industrial robots are already widely used in automotive manufacturing and warehouse logistics, these systems usually perform highly fixed, repetitive motions. They still struggle to adapt to complex environments the way humans can. With rapid breakthroughs in large language models（LLMs）, vision models, and multimodal AI, Humanoid Robots have once again become a major focus across the global technology sector.
Within this new wave of “Embodied AI,” Figure AI is seen as one of the most representative companies. Compared with traditional robotics firms that emphasize mechanical design and motion control, Figure AI places greater importance on deeply integrating large AI models with the robotic body. Its goal is to build a general-purpose robotics platform that can truly understand the physical world and reason autonomously.
Figure AI is a U.S. AI Robotics company founded in 2022 and headquartered in California. It was founded by serial entrepreneur Brett Adcock. The company’s goal is clear: to develop general-purpose Humanoid Robots capable of replacing humans in real-world work.
Unlike traditional industrial robots, Figure AI’s robots are designed as a “general-purpose labor platform.” This means they are not limited to moving items in factories. They could also perform complex tasks in warehouses, logistics, retail, home services, and even healthcare settings.
Figure AI’s current core products include:
Among them, Helix AI is widely regarded as the most important technical core of the entire Figure AI ecosystem.
Figure AI chose Humanoid Robots instead of wheeled robots or fixed robotic arms for one important reason: the real world itself is designed around the human form.
Whether it is factory tools, warehouse shelves, elevators, door handles, or home kitchens, most real-world facilities are built with human users in mind. A robot with two arms, two legs, and a humanlike structure can therefore adapt more easily to existing social infrastructure without requiring the environment to be completely rebuilt.
Figure AI believes humanoid robots could become a general-purpose computing platform in the physical world, much like smartphones became the core platform of the mobile internet.
This approach shares similarities with Tesla Optimus, Agility Robotics Digit, and other projects, but Figure AI places stronger emphasis on AI model driven capabilities rather than hardware engineering alone.
What Is Helix AI
Helix AI is Figure AI’s core robotic intelligence system and the key factor that sets it apart from traditional robotics companies.
Traditional robots usually rely on preset rules. For example, a robotic arm may only pick up objects along a fixed path. Once the environment changes, the robot may no longer function properly.
Helix AI, by contrast, uses a Vision-Language-Action（VLA）architecture. This means the robot can perceive the real world, understand language instructions, reason through tasks, and generate actions on its own.
For example, when a person tells the robot, “Put the apple on the table into the refrigerator,” the robot must identify the apple, understand the task, locate the refrigerator, plan a path, and execute the action. This process is already close to the ability of an AI Agent to operate in the physical world.
Figure 01 was Figure AI’s first publicly shown humanoid robot prototype. It was mainly used to demonstrate motion control, walking ability, and basic manipulation functions.
Figure 02 marks the point where Figure AI began moving toward commercialization. Compared with Figure 01, Figure 02 has stronger AI processing capabilities, more natural human-machine interaction, and more precise hand manipulation. It is also better suited for long-duration industrial deployment.
Figure AI has shown a demo in which a robot speaks with a human in real time, understands natural language, and responds immediately. In addition, Figure 02 has already begun taking part in real work scenario tests at BMW factories. This suggests that Figure AI is gradually moving from a laboratory project toward industrial deployment.
BMW is one of Figure AI’s most important commercial partners.
For the Humanoid Robot industry, the biggest challenge is not “whether a robot can move,” but “whether a robot can truly enter real production systems.”
The importance of the BMW factory partnership lies in the fact that Figure AI can access real industrial scenario data and allow its robots to keep training on real-world tasks. At the same time, its AI models can build a genuine real-world feedback loop.
Compared with demo style robot showcases, entering an actual automotive factory means Figure AI must solve issues such as stability, safety, long-duration operation, multitask execution, and human-robot collaboration. This is also a key dividing line between humanoid robots as a concept and humanoid robots as an industrial reality.
Figure AI previously worked with OpenAI to explore the use of large language models in robotics.
The two sides once presented a widely discussed demo in which a robot could talk with a human in real time and understand tasks in its physical surroundings. This example gave the outside world an early look at the potential of combining LLMs with robots.
Later, however, Figure AI gradually shifted toward an independent AI path, placing more emphasis on self-developed robotics models and embodied intelligence systems.
The reason is that robotic AI is very different from pure text based AI. Robots need more than language understanding. They also need spatial perception, motion planning, visual reasoning, real-time control, and multisensor fusion. For this reason, Figure AI aims to build a Robotics Foundation Model designed specifically for the physical world.
Figure AI has received backing from major technology investors, including Microsoft, NVIDIA, Jeff Bezos, and the OpenAI Startup Fund.
The capital market is paying close attention to Figure AI mainly because the Humanoid Robot market could become enormous. If robots can replace part of the human labor force, the potential market size could reach trillions of dollars.
At the same time, AI is beginning to move beyond the digital world and enter the physical world. Global labor shortages and rising demand for manufacturing automation are also accelerating the development of the Humanoid Robot industry.
Many investment institutions believe Embodied AI could become the next major AI platform after large language models.
Although Figure AI has attracted significant attention, the Humanoid Robot industry still faces major challenges.
The first is cost. High-performance robots remain very expensive, and mass adoption is still a long way off.
The second is battery life and endurance. Humanoid robots need to operate for long periods, but mobile robotic systems consume a great deal of energy.
In addition, AI generalization remains limited. Robots still fall far short of humans in terms of stability in complex real-world environments.
Safety is equally important. Once robots enter factories and homes, they must be able to operate without creating risks for humans.
Finally, large-scale manufacturing is another challenge the entire industry must solve. How to mass produce robots in the way cars are produced remains a major issue for the Humanoid Robot industry.
If Figure AI’s technical path succeeds, its impact could extend far beyond the traditional robotics industry.
The first sectors likely to change include Manufacturing, Warehouse, Logistics, Retail, Healthcare, and Home Assistant.
Over the long term, Humanoid Robots may even become “AI labor infrastructure” in the real world. This means robots would not simply be automation tools. They could become a major productivity system for future society.
Figure AI’s current development focus is mainly concentrated in three areas.
First, it aims to keep improving robotic AI capabilities. Through Helix AI, Figure AI hopes to give robots stronger reasoning ability and better task generalization.
Second, it wants to expand the scale of commercial deployment. Moving from factory testing into real commercial environments is one of Figure AI’s key goals at this stage.
Third, it needs to reduce robot production costs. Through manufacturing systems such as BotQ, Figure AI hopes to build a more mature capability for mass producing humanoid robots.
In the future, competition in Humanoid Robots may no longer be just about hardware. It may become a broader contest involving data scale, AI model capability, real-world scenario training, and manufacturing systems.
Conclusion
Figure AI is pushing Humanoid Robots from laboratory concepts toward real commercial deployment. Compared with traditional robotics companies, its defining feature is the deep integration of large AI models, visual understanding, and robotic control systems. Through Helix AI, it is building a general-purpose robotics platform with reasoning capabilities.
With its BMW factory partnership, breakthroughs in robotic AI, and continued global capital investment, Figure AI has become one of the most representative companies in the Embodied AI and Humanoid Robot wave.
FAQs
Figure AI was founded by Brett Adcock, who previously founded technology companies including Archer Aviation.
What Is Helix AI
Helix AI is Figure AI’s robotic AI system. It uses a Vision-Language-Action（VLA）architecture to help robots understand, reason, and execute actions.
Figure AI places greater emphasis on integrating AI models with robots, while Tesla Optimus relies more heavily on Tesla’s autonomous driving and manufacturing systems.
The two sides previously worked together to explore AI robotics, but Figure AI later began strengthening its own independent robotics AI path.
Figure AI has already begun deployment testing in real industrial environments, including BMW factories.
Why Are Humanoid Robots Important
Humanoid Robots can directly adapt to existing human social infrastructure, which is why they are considered a key direction for the development of general-purpose robots.
