Skip to content
Our Human Future

Robotics

Six AgiBot Robots Work a Shanghai Store

AgiBot says six wheeled humanoids completed 1,730 tasks during a 24-hour convenience-store trial.

By Our Human Future Research Desk

Evidence: Emerging· Early evidence; may change

Video: Shanghai Eye Youtube

AgiBot says six wheeled humanoids completed 1,730 tasks during a 24-hour convenience-store trial.

Six AgiBot G2 robots spent 24 hours working inside an operating Shanghai convenience store, completing a company-reported 1,730 tasks between November 2 and 3. The wheeled humanoids made coffee, restocked shelves, cleaned and assisted customers while a livestream drew more than two million viewers. [s1]

Four of the six robots passed AgiBot's final performance assessment. That detail separates the event from a flawless promotional spectacle: one-third of the machines did not meet the company's own threshold. AgiBot has not published the scoring rubric, intervention rate or the share of tasks completed without human recovery. [s1]

The trial matters because it moved embodied AI into a working retail environment for a full day, rather than a short stage demonstration. It was still a controlled pilot run by the manufacturer, not proof of durable commercial labour.

The human question begins with the work itself. If retailers move from trials to recurring deployment, employees should have a voice in task design, pace, staffing and the distribution of any productivity gains.

What we do not know

  • Q: Was this a commercial deployment? A: No. It was a 24-hour controlled pilot in an operating store, livestreamed and assessed by the manufacturer.
  • Q: Did the robots replace store workers? A: The public account does not report staffing changes, labour savings or autonomous completion rates.

Sources

Interests and disclosures

AgiBot Ownership
AgiBot has a direct institutional, commercial, employment or advocacy interest in the development described. Organisation: AgiBot

Share this story

For Instagram, use Share on your phone or copy the link.

Spotted a mistake? Report an error