Artificial intelligence is no longer limited to screens, chatbots and software. The next major frontier is physical AI—machines that can see, understand, move through the real world and perform useful tasks.
Few companies illustrate that shift more clearly than Figure AI, founded by entrepreneur Brett Adcock. The company is developing general-purpose humanoid robots designed to work in factories, warehouses and eventually homes.
What makes the story interesting is not simply that humanoid robots exist. It is that the industry is increasingly trying to prove they can perform useful work for extended periods, rather than simply complete impressive demonstrations.
Figure’s latest developments show just how quickly that transition is happening. Its Figure 03 robot has entered manufacturing environments, while its Helix AI system is being developed to help robots understand and perform tasks across increasingly unfamiliar environments.
What Are Humanoid Robots?
Humanoid robots are machines designed around a human-like physical structure, typically featuring a torso, arms, hands and legs.
The design is intentional.
Much of the world humans have built—from factories and warehouses to homes—is designed around the human body. Doors have handles, shelves have particular heights, tools are designed for hands, and workspaces are arranged for people.
A robot with a human-like body can potentially operate within those environments without requiring everything to be redesigned.
However, having a human-like body is only part of the challenge.
The bigger question is:
Can the robot understand what is happening around it and decide what to do next?
That is where AI becomes critical.
Figure AI’s Approach: Combining Robotics With AI
Figure is attempting to combine advanced humanoid hardware with an AI system called Helix.
Figure describes Helix as a Vision-Language-Action system. In simple terms, the system connects what a robot sees and understands with the physical actions it takes.
Instead of simply following a rigid sequence of programmed movements, the goal is for the robot to interpret an environment and respond accordingly.
Figure says Helix can control perception, movement and reasoning on the robot in real time.
That distinction could become extremely important.
A traditional industrial robot might be excellent at repeating the same movement thousands of times. A general-purpose humanoid, however, needs to deal with changing objects, people, obstacles and environments.
Figure 03 Is Designed for the Real World
Figure introduced Figure 03 in October 2025 as its third-generation humanoid robot.
The company designed the robot around three major goals:
- Working with Helix AI
- Operating in homes
- Supporting large-scale manufacturing
Figure says the robot received a redesigned sensory system and hand system, alongside improvements intended to make it more suitable for household environments and mass production.
This is an important shift.
Building one impressive prototype is one thing. Building thousands of robots that can be manufactured, maintained and deployed reliably is a much harder engineering problem.
Figure says its BotQ manufacturing operation had produced more than 350 Figure 03 robots by April 2026 and demonstrated a production rate of one robot per hour.
From Factory Experiments to Actual Production
One of the strongest tests for humanoid robots is whether companies can use them in real industrial environments.
Figure has already been working with BMW.
In June 2026, Figure announced that Figure 03 had arrived at BMW Group’s Spartanburg plant for logistics work. The company said its previous Figure 02 deployment had contributed to the assembly of 30,000 vehicles in 2025.
The newer Figure 03 deployment is focused on more complex logistics activity, including a workflow in which the robot can manipulate objects while walking and repositioning its body.
That matters because real factories introduce conditions that laboratory demonstrations cannot fully reproduce.
Workers move around.
Objects vary.
Tasks change.
Equipment occupies space.
The robot has to continuously respond to its surroundings.
Helix 02 Takes Physical AI Further
Figure’s Helix 02 system represents another important development.
The company says Helix 02 can control the robot’s entire body, combining walking, manipulation and balance through a unified neural system.
One demonstration involved a humanoid autonomously completing a multi-step dishwasher task across a kitchen without human intervention or resets.
Figure has also demonstrated two humanoid robots working together on household tasks, including making a bed, putting objects away and organizing a bedroom.
These demonstrations point toward an important idea:
The future of robotics may depend less on programming robots for individual tasks and more on teaching them general physical intelligence.
The Bigger Breakthrough Could Be Robots That Adapt
A robot that knows how to perform one task is useful.
A robot that can enter an unfamiliar environment and figure out how to perform several tasks could be far more valuable.
That is the problem Figure’s newer Helix 2.5 system is attempting to address.
In September 2026, Figure announced Helix 2.5, describing it as a system designed to generalize across unfamiliar homes rather than learning each environment independently.
This is significant because homes are among the most unpredictable environments a robot could encounter.
Unlike a factory, a home does not necessarily have standardized layouts, objects or workflows.
One kitchen could look completely different from another.
One bedroom could have furniture arranged differently from the next.
For a general-purpose robot, adaptability is therefore essential.
Why Robot Training Data Matters
There is another major challenge that receives less attention than the robots themselves: data.
AI models need enormous amounts of useful information to become capable. Physical AI has a particularly difficult problem because robots cannot simply learn everything from text on the internet.
They need to learn how objects behave physically.
They need to understand movement.
They need to deal with friction, weight, balance, obstacles and unpredictable human environments.
Figure has been building a large-scale data strategy around this problem.
Its Index initiative is designed to collect real-world human activity data that can help train robots. Figure says the system has been designed around collecting diverse examples of people performing tasks in different environments.
This could become one of the biggest competitive advantages in humanoid robotics.
The company with the best robot hardware may not necessarily win.
The company with the best combination of hardware, AI models, training data, manufacturing and real-world deployment could have the stronger position.
What Could Humanoid Robots Actually Do?
If the technology continues improving, humanoid robots could eventually be useful across several industries.
1. Manufacturing
Factories could use humanoids for repetitive physical tasks, material handling, inspection and logistics.
2. Warehousing
Robots could move packages, organize inventory and perform repetitive warehouse operations.
3. Healthcare
Humanoid robots could potentially assist with certain non-clinical tasks, although safety and regulation would be especially important.
4. Hospitality
Hotels, restaurants and other businesses could eventually use robots for selected repetitive tasks.
5. Home Assistance
This is arguably the most ambitious goal.
A sufficiently capable humanoid could potentially help with cleaning, organizing, carrying objects and other everyday household activities.
Figure is explicitly positioning Figure 03 toward household use.
But Humanoid Robots Still Have Major Challenges
The excitement around humanoid robotics should not obscure the difficult problems that remain.
Reliability
A robot may successfully complete a demonstration but still struggle with unexpected situations.
Safety
A powerful machine operating around humans must be extremely predictable and safe.
Cost
Humanoid robots require sophisticated sensors, actuators, batteries, computing hardware and manufacturing infrastructure.
Battery Life
Physical work consumes energy. Robots need enough battery capacity to perform useful work without constantly stopping to recharge.
Generalization
A robot that works perfectly in one environment may behave differently somewhere else.
Maintenance
Thousands or millions of physical machines create maintenance challenges that software companies do not face at the same scale.
These limitations are why real-world deployment matters so much.
The Difference Between a Robot Demo and a Useful Robot
A polished demonstration can make robotics look closer to the future than it actually is.
The more meaningful test is whether a robot can:
- Enter a real environment.
- Understand its surroundings.
- Perform a useful task.
- Handle unexpected changes.
- Work safely alongside people.
- Repeat the task consistently.
- Operate for long periods.
- Do all of this at an economically viable cost.
That is the standard the humanoid robotics industry will ultimately have to meet.
Will Humanoid Robots Replace Human Workers?
This is one of the biggest questions surrounding the technology.
The answer is not straightforward.
In the short term, humanoid robots are more likely to augment specific jobs and automate individual tasks than completely replace entire occupations.
For example, a warehouse worker may still be responsible for coordinating operations while robots handle repetitive movement of goods.
Over time, however, increasingly capable robots could automate a wider range of physical tasks.
That could change the nature of many jobs.
At the same time, new roles could emerge around robot maintenance, supervision, training, deployment, safety and AI systems.
The impact will ultimately depend on how quickly the technology improves, how affordable it becomes and how businesses choose to deploy it.
Why This Matters for Businesses
Businesses should pay attention even if they have no immediate plans to purchase a humanoid robot.
The technology represents a broader transformation in automation.
Companies are increasingly moving toward systems that combine:
- Artificial intelligence
- Computer vision
- Robotics
- Automation
- Real-time decision-making
- Physical data collection
For entrepreneurs, this creates opportunities beyond building robots themselves.
There could be demand for software, robot training data, specialized applications, maintenance services, safety systems and industry-specific automation.
What Comes Next?
Figure’s trajectory shows that humanoid robotics is moving beyond the prototype stage.
Its robots are being tested in manufacturing environments, its AI systems are becoming more capable, and the company is investing heavily in the data and infrastructure required to scale physical AI.
Still, the biggest question is not whether humanoid robots can walk, pick things up or perform impressive demonstrations.
The real question is whether they can become reliable, affordable and genuinely useful machines at scale.
If that happens, the impact could extend far beyond the robotics industry.
It could change how factories operate, how warehouses function and eventually how people interact with technology inside their own homes.
Frequently Asked Questions
What is a humanoid robot?
A humanoid robot is a robot designed with a human-like physical structure, usually including a torso, arms, hands and legs. The design allows it to potentially operate in environments built for humans.
What is Figure AI?
Figure AI is a robotics company developing general-purpose humanoid robots and AI systems designed to help those robots perform tasks in the physical world.
Who is Brett Adcock?
Brett Adcock is the founder and CEO of Figure AI, the company behind the Figure humanoid robot platform.
What is Figure 03?
Figure 03 is Figure AI’s third-generation humanoid robot. It was designed for applications including industrial work, household environments and large-scale manufacturing.
What is Helix AI?
Helix is Figure’s Vision-Language-Action AI system, designed to connect perception, reasoning and physical control so its humanoid robots can perform tasks in real-world environments.
Are humanoid robots ready to replace humans?
Not broadly. Although the technology is advancing rapidly, reliability, safety, cost, adaptability and real-world performance remain major challenges.
Conclusion: The Humanoid Robot Race Is Just Beginning
The rise of humanoid robots represents a major change in the direction of artificial intelligence.
For years, AI primarily lived inside computers and smartphones. Now, companies are attempting to give AI a physical body capable of interacting with the real world.
Figure AI is one of the companies pushing that transition forward.
Its progress—from factory deployments and increasingly capable Helix models to large-scale robot production and real-world training data—suggests that the humanoid robotics race is becoming less about science-fiction demonstrations and more about practical deployment.
The technology is not finished.
But the direction is clear: AI is learning not only to understand the world, but potentially to act inside it.
Follow TechBrass for more technology news, AI developments, robotics stories and simple explanations of the technologies shaping what comes next.
