Figure AI

A compact snapshot of Figure AI — from founding details and leadership to the three pillars of today's product lineup, partner relationships, and market focus.

Figure was founded in 2022 by entrepreneur Brett Adcock with the goal of building commercially viable general-purpose humanoid robots. The company is headquartered in San Jose, California.

Figure is building humanoid robots intended to operate in the same physical environments as humans.

Instead of designing a separate robot for every individual task, Figure believes a general-purpose humanoid can provide a common physical interface for many different jobs.

The concept is based on a simple observation.

Buildings, factories, warehouses, tools, shelves, workstations, vehicles, doors, and household environments were largely built around the human body.

A humanoid robot can therefore potentially use existing infrastructure without requiring an entirely new environment designed specifically for machines.

Figure's long-term objective is to combine this human-compatible physical form with advanced AI, allowing robots to learn and perform increasingly broad categories of tasks.

Figure's strategy is built around three interconnected challenges.

This creates a continuous development loop.

Figure's recent production efforts are designed around this principle. In April 2026, the company reported that its BotQ manufacturing facility had produced more than 350 third-generation robots and demonstrated a production rate of one Figure 03 per hour.

Figure's robot development has moved through several generations in only a few years.

Figure 01 was the company's first-generation humanoid robot.

Its significance was not necessarily its final capabilities, but what it represented for the company.

The robot demonstrated the basic capabilities required for a humanoid platform, including:

• Bipedal locomotion
• Manipulation
• Human-scale proportions
• Autonomous movement research
• Physical interaction with objects

Figure 01 laid the hardware and robotics foundation for the generations that followed.

Figure 02 marked Figure's transition from early humanoid development toward practical industrial applications.

The robot featured a more refined physical design, improved hand dexterity, and an integrated torso battery. Its development was closely connected to Figure's real-world deployment strategy.

In 2025, Figure 02 was deployed at BMW Group Plant Spartanburg, where it contributed to automotive production operations. Figure later reported that Figure 02 contributed to the production of approximately 30,000 vehicles at BMW during 2025.

Figure 03 is the company's current humanoid platform.

While earlier generations were heavily focused on proving humanoid capabilities in industrial environments, Figure 03 expands the platform toward a broader range of real-world applications.

Figure 03 is designed to navigate environments such as homes, including stairs, tight spaces, and changing layouts.

Figure 03 is not limited to household applications.

In June 2026, Figure announced that Figure 03 had returned to BMW Group Plant Spartanburg. The new deployment focuses on a more complex manufacturing logistics task known as sequencing.

Unlike simple pick-and-place operations, sequencing requires the robot to identify and organize parts that may arrive in different positions and orientations.

The robot can therefore adjust its body while manipulating parts rather than executing a fixed sequence of predetermined movements.

Figure's long-term vision extends beyond factories. After developing and testing humanoids in the workforce, the company reimagined Figure 03 for household environments.

The home presents a fundamentally different robotics challenge.

A household robot may encounter:

• Different furniture layouts
• Stairs
• Tight spaces
• Small objects
• Clutter
• Unpredictable human behavior
• Changing lighting
• Objects in unexpected locations

Figure 03 is designed around this more variable environment.

Potential household tasks include:

The company sees the home as one of the major long-term applications for general-purpose humanoids.

Figure's humanoid strategy depends on more than mechanical engineering.

The company is developing its own AI system called Helix. Figure describes Helix as a generalist Vision-Language-Action model designed to control humanoid robots.

The goal is to connect:

into a unified system capable of controlling the robot in real time.

Traditional industrial robots often rely on carefully programmed sequences. A human programmer defines:

1. 1Where the robot should move
2. 2What object it should pick up
3. 3How the arm should move
4. 4Where the object should be placed

This works well in controlled environments. But it becomes much harder when objects, environments, and conditions change.

Helix is designed around a different approach. The robot observes the environment, interprets what it sees, reasons about the task, and generates physical actions.

In January 2026, Figure introduced Helix 02, extending the company's AI architecture from upper-body control to full-body humanoid control.

Helix 02 integrates:

• Vision
• Touch
• Proprioception
• Locomotion
• Manipulation
• Balance
• Whole-body control

The system allows the robot to coordinate its hands, arms, torso, hips, and feet as part of a continuous behavior.

One of the hardest problems in humanoid robotics is loco-manipulation.

A robot may be able to walk. It may also be able to manipulate an object. But combining the two continuously is significantly harder.

Figure's Helix 02 architecture is designed to treat these behaviors as one continuous control problem. This is an important step toward robots capable of performing useful tasks in environments that were not engineered specifically for automation.

Figure is also expanding humanoid sensing beyond vision. Figure 03 incorporates embedded tactile sensing and palm cameras.

These capabilities can help the robot manipulate small or difficult objects where vision alone may not provide enough information.

Figure has demonstrated tasks involving:

• Small irregular objects
• Individual pills
• Syringe manipulation
• Cluttered objects
• Fine manipulation

Figure's latest technology initiative extends beyond the robot and AI model itself.

The company has introduced Index, an effort focused on building a large and diverse physical dataset.

The fundamental idea is straightforward.

AI models can learn from enormous amounts of digital information. Physical robots have far less training data.

Figure is therefore working to build datasets that capture how humans and robots interact with the physical world. The company's broader strategy is to use large-scale robot deployments to generate more data, improve AI models, and accelerate the development of capable humanoids.

Figure's strategy can be understood as a data flywheel.

This is one reason Figure is investing heavily in manufacturing capacity.

The company does not view production only as a way to sell robots. It also sees a larger robot fleet as an important source of real-world data for improving its AI systems.

A major challenge for humanoid robotics is moving from prototypes to production.

Figure established BotQ, a high-volume manufacturing facility designed to support this transition.

In April 2026, Figure reported that BotQ had produced more than 350 Figure 03 robots and had demonstrated a production rate of one robot per hour.

The company also reported more than 9,000 actuators produced across more than 10 distinct SKUs, along with dedicated testing and quality-control processes throughout production.

Manufacturing scale matters because a general-purpose humanoid requires more than an impressive prototype. It needs:

• Repeatable manufacturing
• Reliable components
• Quality control
• Battery production
• Actuator production
• Serviceability
• Fleet management
• Software updates

Humanoid robotics creates a unique manufacturing challenge.

A robot is not simply a mechanical product. Its performance depends on the interaction between:

As more robots enter the field, Figure can collect operational data and use it to improve both hardware and AI.

The company has also developed fleet-management infrastructure to monitor robot health, location, operational status, upgrades, and field performance.

Manufacturing is one of Figure's most important early markets.

The reason is straightforward. Factories contain many tasks that are:

• Repetitive
• Physically demanding
• Structured enough for automation
• Difficult to staff consistently

At the same time, factories already contain human-oriented workspaces.

BMW has been one of the most important real-world test environments for Figure's humanoid robots.

Figure 02 was deployed at BMW Group Plant Spartanburg in 2025. Figure reported that the robot contributed to the production of approximately 30,000 vehicles.

In 2026, Figure 03 returned to the same facility for a more complex logistics workflow involving sequencing and dynamic material handling.

Figure is also expanding into distribution and logistics.

In May 2026, Figure announced a commercial agreement with Catalyst Brands to deploy humanoid robots at its distribution and logistics network.

The initial deployment is focused on the company's Reno, Nevada Distribution Logistics Center and physically demanding supply-chain tasks.

This represents another important application for humanoids.

Warehouses contain constantly changing arrangements of boxes, carts, shelves, containers, and materials.

Figure's long-term vision is significantly broader than any individual industrial application.

The company envisions general-purpose humanoids operating across:

Figure's central argument for humanoid robots is compatibility.

Humans have already built the world around a particular physical form.

We use:

• Two-legged movement
• Two-handed manipulation
• Human-scale tools
• Human-height shelves
• Human-sized doors
• Human-oriented workstations

A general-purpose humanoid can potentially operate within this infrastructure.

This could allow one robotic platform to perform many different tasks without requiring a completely different machine for each application.

Figure's stated mission is:

The company sees humanoid robotics as a physical extension of artificial intelligence.

Instead of AI existing only inside computers, humanoids could allow AI systems to perceive and act in the physical world.

That means AI could potentially:

• Move objects
• Navigate spaces
• Manipulate tools
• Perform household tasks
• Work in factories
• Assist people

The ultimate objective is not simply to create a humanoid machine.

It is to create an intelligent physical system capable of performing useful work.

Figure's long-term vision is to deploy autonomous humanoid robots at global scale.

The company initially focused on workforce applications, particularly industries facing labor shortages and physically undesirable or unsafe tasks.

Over time, Figure envisions humanoids expanding into:

• Homes
• Elder care
• Consumer assistance
• Industrial automation
• Logistics
• Manufacturing
• Potentially off-world environments

The common thread is general-purpose physical intelligence.

Figure's strategy is unusual because it is attempting to control much of this technology stack internally rather than treating the humanoid as simply a hardware product.

These values reflect the scale and difficulty of the company's stated mission.

Figure's development cycle is based on rapid iteration.

A new robot generation provides:

The transition from Figure 01 to Figure 02 and then Figure 03 illustrates this approach. The company is effectively treating the robot itself as a continuously evolving AI-and-hardware platform.

The rise of large AI models has dramatically improved machine reasoning and perception.

But physical environments remain fundamentally different from digital environments.

A humanoid must:

• See an object
• Reach for it
• Estimate its position
• Move its body
• Apply the right amount of force
• Maintain balance
• Respond if something moves

This makes humanoid robotics one of the most demanding applications of AI.

Figure's Helix strategy is an attempt to close this gap. The objective is to turn AI intelligence into physical action.

Figure's commercial strategy has followed a logical progression.

1. 1Develop the humanoid platform.
2. 2Test structured tasks in industrial environments.
3. 3Deploy robots with commercial partners.
4. 4Increase manufacturing scale.
5. 5Use larger fleets to generate more physical-world data.
6. 6Expand the robot's capabilities through AI.
7. 7Move from industrial applications toward homes and broader environments.

This strategy allows Figure to begin with environments where the economic value of automation is easier to measure while gradually working toward more general-purpose applications.

Figure's progress is significant, but the underlying problem remains extraordinarily challenging.

A truly general-purpose humanoid must solve several difficult problems simultaneously:

Figure is attempting to solve all of these problems simultaneously.

Figure's long-term opportunity depends on whether humanoid robots can become economically useful at scale.

The company is pursuing a future in which humanoids can gradually move from:

This is a much larger ambition than building a successful robot product.

It is an attempt to create a new category of physical labor powered by AI.

Figure operates in one of the most competitive areas of modern robotics.

The company is developing alongside other major humanoid robotics programs, but its strategy has a distinctive emphasis on combining:

This combination is central to Figure's attempt to become a vertically integrated humanoid robotics company.

Dive deeper into the broader humanoid and robotics landscape — from competing humanoid platforms and industrial quadruped programs to entertainment and emotional robotics.