The task was deliberately simple and repetitive to isolate a single measurable benchmark: how quickly each could process packages over an extended shift.

Final Results

By the end of the 10‑hour run, the totals were extremely close:

• Aime (human intern): 12,924 packages — about 2.79 seconds per package
• Figure F.03 robot: 12,732 packages — about 2.83 seconds per package

That meant the human won by 192 packages, roughly a 1.5% lead after ten hours of continuous work.

The small difference is notable because industrial warehouse work often operates within narrow throughput margins—meaning the robot’s performance was already close to typical human shift speeds.

Breaks vs. Nonstop Operation

One detail that made the contest interesting was how each competitor handled endurance.

• The human took occasional breaks during the long shift.
• The robot operated continuously without breaks.

Even with those breaks, the human still finished slightly ahead in total packages sorted.

However, the robot’s ability to keep working without fatigue points to one of the main advantages companies hope to achieve with humanoid automation: continuous operation over very long periods.

Separate demonstrations from Figure AI have shown humanoid robots sorting packages for more than 24 hours continuously, processing tens of thousands of parcels during extended livestream tests.

Brett Adcock’s Take: “Human Parity”

After the contest, Figure AI founder and CEO Brett Adcock emphasized that the result should be interpreted as progress rather than a loss for the robot.

According to Adcock, human workers performing similar warehouse sorting tasks typically average around three seconds per package across a full shift. He argued that the F.03 robot’s performance—just under that level—suggests it is approaching “human parity” in this narrow workflow.

In other words, the robot is now operating within the same throughput range as a human worker, even if peak human performance can still be faster.

Why the Event Matters

The contest gained attention because it offered a rare quantitative, real‑time comparison between a humanoid robot and a human worker performing the same physical task.

For observers tracking robotics progress, it provided three useful signals:

1. Robots are getting close to human speed
Processing a package roughly every three seconds is already within the throughput range of many warehouse roles. The robot finishing only slightly behind suggests meaningful advances in manipulation, perception, and coordination.

2. Endurance may become a robotic advantage
While the human narrowly won the contest, robots can theoretically operate continuously across multiple shifts without fatigue—something Figure has already attempted to demonstrate with multi‑day livestream sorting runs.

3. Controlled demos are not real warehouses
Experts note that such tests are highly constrained. Real logistics centers involve irregular objects, unexpected errors, safety concerns, and coordination with human workers—factors that remain challenging for autonomous robots.

The Real Takeaway

The livestream didn’t prove that humanoid robots are ready to replace warehouse workers. But it did show something important: robots are no longer dramatically slower than humans at certain repetitive physical tasks.

For now, the fairest conclusion is that humanoid robots like Figure’s F.03 have reached near‑human performance in a tightly controlled sorting task. Whether that translates into reliable, economical deployment in real warehouses remains an open question.

What the contest demonstrated most clearly is how narrow the gap has become—and how quickly it may close if robotics hardware and AI control systems keep improving.

The task was deliberately simple and repetitive to isolate a single measurable benchmark: how quickly each could process packages over an extended shift.

Final Results

By the end of the 10‑hour run, the totals were extremely close:

• Aime (human intern): 12,924 packages — about 2.79 seconds per package
• Figure F.03 robot: 12,732 packages — about 2.83 seconds per package

That meant the human won by 192 packages, roughly a 1.5% lead after ten hours of continuous work.

The small difference is notable because industrial warehouse work often operates within narrow throughput margins—meaning the robot’s performance was already close to typical human shift speeds.

Breaks vs. Nonstop Operation

One detail that made the contest interesting was how each competitor handled endurance.

• The human took occasional breaks during the long shift.
• The robot operated continuously without breaks.

Even with those breaks, the human still finished slightly ahead in total packages sorted.

However, the robot’s ability to keep working without fatigue points to one of the main advantages companies hope to achieve with humanoid automation: continuous operation over very long periods.

Separate demonstrations from Figure AI have shown humanoid robots sorting packages for more than 24 hours continuously, processing tens of thousands of parcels during extended livestream tests.

Brett Adcock’s Take: “Human Parity”

After the contest, Figure AI founder and CEO Brett Adcock emphasized that the result should be interpreted as progress rather than a loss for the robot.

According to Adcock, human workers performing similar warehouse sorting tasks typically average around three seconds per package across a full shift. He argued that the F.03 robot’s performance—just under that level—suggests it is approaching “human parity” in this narrow workflow.

In other words, the robot is now operating within the same throughput range as a human worker, even if peak human performance can still be faster.

Why the Event Matters

The contest gained attention because it offered a rare quantitative, real‑time comparison between a humanoid robot and a human worker performing the same physical task.

For observers tracking robotics progress, it provided three useful signals:

1. Robots are getting close to human speed
Processing a package roughly every three seconds is already within the throughput range of many warehouse roles. The robot finishing only slightly behind suggests meaningful advances in manipulation, perception, and coordination.

2. Endurance may become a robotic advantage
While the human narrowly won the contest, robots can theoretically operate continuously across multiple shifts without fatigue—something Figure has already attempted to demonstrate with multi‑day livestream sorting runs.

3. Controlled demos are not real warehouses
Experts note that such tests are highly constrained. Real logistics centers involve irregular objects, unexpected errors, safety concerns, and coordination with human workers—factors that remain challenging for autonomous robots.

The Real Takeaway

The livestream didn’t prove that humanoid robots are ready to replace warehouse workers. But it did show something important: robots are no longer dramatically slower than humans at certain repetitive physical tasks.

For now, the fairest conclusion is that humanoid robots like Figure’s F.03 have reached near‑human performance in a tightly controlled sorting task. Whether that translates into reliable, economical deployment in real warehouses remains an open question.

What the contest demonstrated most clearly is how narrow the gap has become—and how quickly it may close if robotics hardware and AI control systems keep improving.