Figure AI demonstrated that its humanoid robots can operate autonomously for extended periods, sorting more than 28,000 packages over 24 hours without human intervention. The California-based robotics startup ran three robots, nicknamed Bob, Frank, and Gary by online viewers, through a continuous package-sorting task that was originally planned for just eight hours but kept running successfully.

What Makes This Test Different From Previous Robot Demos?

The livestreamed test revealed something that separates this demonstration from typical robot showcases: endurance without failure. The robots used Helix-02, Figure AI's in-house artificial intelligence (AI) system, to handle the complete workflow of picking up packages, detecting barcodes using onboard cameras, and placing items on a conveyor belt with the barcode facing down. Unlike remote-controlled demonstrations, the company emphasized that every action came directly from the AI system itself, not from human operators steering the robots from a distance.

The task itself sounds straightforward, but warehouse work demands steady movement, quick decision-making, and the ability to adapt when unexpected problems occur. The robots maintained speeds comparable to human workers throughout the 24-hour period, according to CEO Brett Adcock. What caught viewers' attention was not flashy backflips or entertainment value, but rather the mundane reality of robots grinding through repetitive labor hour after hour.

How Do These Robots Handle Real-World Warehouse Challenges?

Figure AI highlighted a critical capability that could determine whether these robots succeed in actual warehouses: automatic recovery. The company says Helix-02 can trigger an automatic reset when a robot gets stuck or encounters a situation outside its expected behavior patterns. This matters because a robot that needs human help every few minutes becomes a liability rather than an asset. A robot that can pause, diagnose its own problem, and resume work independently starts to look like a genuine productivity tool.

The system also includes a maintenance protocol where robots can leave the work floor if software or hardware issues emerge, allowing another robot to take over and keep operations moving without interruption. This redundancy approach suggests Figure AI is thinking about real workplace conditions, not just controlled lab environments.

Steps to Evaluate Humanoid Robot Readiness for Your Warehouse

• Test Endurance Under Real Conditions: Move beyond single-task demonstrations to multi-hour trials with varying package sizes, shapes, and label placements that reflect actual warehouse chaos.
• Assess Failure Rates and Maintenance Needs: Demand independent data on how often robots fail, what maintenance intervals they require, and whether they can handle messy conditions without slowing operations.
• Evaluate Cost-Benefit Analysis: Calculate whether the robots can justify their expense through labor savings, speed improvements, and reduced downtime compared to human workers or existing automation.
• Monitor Safety and Integration: Verify that robots can safely operate alongside human workers, handle unexpected obstacles, and integrate with existing conveyor systems and warehouse management software.

What Still Needs to Happen Before Warehouses Deploy These Robots?

The 24-hour test proved endurance, but real-world deployment demands far more evidence. Warehouse floors are chaotic environments where packages arrive in unexpected shapes, labels appear in odd locations, conveyor belts jam, and people walk through work areas unpredictably. A robot that handles one livestreamed task still must prove it can handle the messier version of actual warehouse work without slowing down the entire operation.

Companies considering these robots will want independent verification of performance claims, not just data from company-controlled demonstrations. They will need to understand maintenance costs, failure rates, and whether the robots can adapt to the variability that defines real warehouse environments. Figure AI faces competition from Tesla, Agility Robotics, and Apptronik, all developing humanoid robots for industrial spaces. The company has already tested its robots at BMW manufacturing facilities in South Carolina, suggesting these systems may appear first in controlled industrial settings before spreading to other warehouse operations.

The broader question extends beyond technical capability. If robots can work continuously without breaks, warehouses may restructure overnight shifts, reduce staffing levels, or redeploy workers to problem-solving roles that robots cannot yet handle. This does not necessarily mean every warehouse job disappears. Real workplaces require human judgment to solve problems that demonstrations rarely encounter. However, Figure AI's test signals that humanoid robots are moving from short promotional videos toward longer, more realistic workplace trials.

The impact of this technology may show up in familiar places: faster package handling could affect delivery times, warehouses may change how they staff overnight shifts, and companies may use robots to fill repetitive roles that are hard to staff or physically demanding. For workers, the message is clear: automation is moving deeper into everyday labor. For companies, the message is equally clear: the economics of warehouse automation are shifting in favor of robots that can work reliably for extended periods without human supervision.