Figure AI's humanoid robots have crossed a critical threshold: they can now work a full eight-hour factory shift completely autonomously, matching human performance speeds. On May 13, 2026, CEO Brett Adcock livestreamed the company's Figure 03 robots processing warehouse packages for eight straight hours without human assistance, handling roughly 2.6 seconds per item. This wasn't a carefully edited highlight reel; it was unedited, real-time footage that exposed both the robots' capabilities and occasional imperfections.

What Makes Helix-02 Different From Previous Robot Systems?

The breakthrough lies in how Figure redesigned the robot's brain. Traditional industrial robots rely on two separate controllers: one for motion and another for handling objects. Helix-02 merges both into a single neural network that processes all inputs simultaneously, including camera feeds from the robot's head and hands, touch sensors on its fingertips, and position data from its joints. This unified architecture allows the robot to adapt to real-world chaos in ways that hand-coded instructions cannot.

Figure also released System 0, an end-to-end neural controller trained on over 1,000 hours of human motion data. This single learned system replaces more than 109,000 lines of handwritten C++ code. The difference matters: traditional code is brittle and breaks when conditions change, while learned systems generalize and handle unpredictable environments.

How Are These Robots Actually Performing on the Factory Floor?

During the livestream, the robots demonstrated impressive fine motor control. In the first 10 minutes alone, they successfully oriented 230 packages, translating to roughly 2.6 seconds per item. The packages varied significantly, consisting mostly of deformable colored poly bags mixed with rigid cardboard boxes of different sizes, which highlights the system's ability to handle real-world variability.

Beyond warehouse logistics, Figure has demonstrated other capabilities in controlled settings. The robots have autonomously unloaded and loaded a dishwasher for four consecutive minutes without interruption, and two robots working together rearranged a bedroom by hanging clothes, making a bed, emptying waste bins, and arranging furniture. The robots also show fine motor precision: they can unscrew bottle caps, extract pills from organizers, dispense accurate syringe amounts, and manipulate metallic parts from bins using tactile feedback and in-hand vision.

Steps to Understanding Figure's Multi-Robot Coordination Strategy

• Visual Communication: The robots coordinate with each other through visual and physical cues, not verbal messaging. Head nods and body positioning allow them to synchronize actions without explicit communication protocols.
• Autonomous Failover: If one robot detects a problem, it self-diagnoses the issue and autonomously walks to maintenance to request a replacement from the fleet, keeping operations continuous with no human intervention.
• Fleet Redundancy: Multiple robots work together so that if one unit experiences a malfunction, others can continue the task, preventing the entire shift from halting.

"It's a multi-robot coordination with autonomous failover strategy. If a robot detects an issue, it will self-diagnose itself and if there's an issue it autonomously walks to maintenance and requests a replacement from the fleet, no humans in the loop," explained Brett Adcock, CEO of Figure AI.

Brett Adcock, CEO at Figure AI

Why Does This Matter for the Broader Robotics Industry?

The livestream was a direct response to a challenge from Dr. Scott Walter, a robotics expert who argued that humanoid robots would have limited utility until they could survive lengthy work hours like humans do. Adcock accepted the challenge and delivered proof. However, some industry observers have noted that the demonstration, while technically impressive, focuses on a single repetitive task rather than showcasing generalized autonomy across multiple job types.

The stakes are substantial. Figure is valued at $39 billion and has raised over $1 billion in funding. The company previously deployed robots at BMW's manufacturing plant in Spartanburg, South Carolina, where they handled over 10 hours per day, moving more than 90,000 parts daily and contributing to the production of over 30,000 cars. The livestream represents a public commitment to scaling this capability.

Looking ahead, Adcock has signaled plans to bring humanoid robots into homes under a subscription model costing roughly $600 per month, similar to a car lease. The robots would handle household tasks like laundry, dishes, and tidying multiple times daily. This consumer push, combined with the factory demonstrations, suggests Figure is positioning itself at the center of a broader shift toward autonomous labor across both industrial and domestic settings.

The race is intensifying. Tesla, Agility Robots, Apptronik, and other companies are competing aggressively in the humanoid robotics space. Figure's ability to prove sustained autonomous operation at human performance levels may give it a significant advantage in attracting both enterprise customers and consumer interest as the market begins to scale.