In a video, the Chinese company Agibot showcases its new humanoid performing tasks that typically require direct human interaction. In a hotel setting, it greets guests, hands out access cards, and escorts visitors; in commercial establishments, it assists customers, presents products, and interacts with other robots; and at a car dealership, it explains vehicle features and helps attend to buyers.

The company aims to demonstrate that humanoids can begin taking on roles outside of factories—in environments where they must navigate spaces and interact with people. Some of these tasks appear simple only when performed by a human; handing over a card requires identifying the person, locating the object, and coordinating hand movements. Escorting someone involves traversing corridors, finding the right path, and navigating obstacles, while retail service requires interpreting requests that vary from one customer to the next. Agibot claims the A3 Ultra can perform various tasks autonomously and collaborate with other machines; however, promotional videos alone do not reveal how long it can operate without human intervention or which scenes involved supervision or remote control.

Now, the company is attempting to scale up this proposal. In September, AgiBot announced a partnership with the Chimelan Group to deploy over 300 robots at a resort complex in China. The project encompasses applications ranging from visitor guidance and entertainment to education, retail, and hospitality services. Rather than simply showcasing a humanoid performing an isolated task, the initiative brings together various types of robots in a public-facing environment; however, there is a significant difference between deploying hundreds of machines and proving they can operate reliably throughout an entire workday.

A hotel cannot halt service every time a robot encounters an unexpected situation; a retail store must contend with children, misplaced items, and customers making highly varied requests; and a humanoid navigating among people needs to know when to stop, ask for help, or simply decline to execute a command. These challenges differ vastly from those encountered in an arena setting or a dance performance. Then there is the economic factor: the "H" model is listed at approximately $89,000 in a commercial offer cited by industry media.

To justify an investment of that magnitude, it is not enough for a robot to simply execute a task in front of cameras; it must perform it consistently, safely, and at a cost that makes sense for the end user. Nevertheless, what we are witnessing in 2026 represents a significant shift in how these machines are being developed.

In previous instances, we saw a humanoid facing off against a man inside a cage, robots fighting one another, and AI systems beginning to plan movements while an operator selected each action; now, we find these same technologies being developed for environments where people work, shop, and spend their lives. Perhaps the greatest challenge for humanoids is not learning to fight, dance, or carry objects, but learning to do all of that in a world that was not designed for them. After all, if these machines truly begin to populate hotels, stores, and eventually our own homes, an inevitable question arises: What will it be like to live alongside robots capable of acting on their own? And what can we expect from this evolution over the next 5, 10, or 25 years?