In Manhattan, Robotic Arms Wash Windows 45 Floors Up

At 1133 Avenue of the Americas, a forty-five-story office tower in the heart of Manhattan, two mechanical arms glide down the glass facade, squeegees in hand. Together they are called Ozmo, they belong to a company named Skyline Robotics, and they perform a task that until recently was done by men dangling from a platform, sometimes five hundred feet above the sidewalk. On the roof, an operator watches the job unfold on a screen. No one hangs from a rope anymore.

Cleaning the windows of a skyscraper is among the most dangerous jobs there is. According to the records of the U.S. workplace-safety agency, over roughly fifteen years nearly ninety reported accidents among window cleaners left sixty-two people dead, almost all from falls. In the single stretch from 2022 to 2023, seven window washers died nationwide. Handing the work to a machine is not an engineer's vanity, then: it is, first of all, a matter of lives. But delegating a task this visible, this physical, opens a quieter set of questions about what we gain and what we give up in return.

A job nobody wants anymore

Behind the arrival of the robots lies an emptiness. In the United States, roughly three-quarters of window cleaners are over forty, and fewer than one in ten is between twenty and thirty. The trade is aging, recruits poorly, and struggles to replace those who hang up the harness for good. The machine does not merely displace a workforce: it fills a seat that fewer and fewer people want to take.

The time saved seals the case for building managers. Skyline says Ozmo cleans up to three times faster than a human crew. A facade finished in days rather than weeks means fewer platforms tied up, fewer streets blocked below, cleaner glass more often.

For anyone who did the work, the benefit fits in a single image: the rope stays empty. The repetitive motion, the exposure to wind, the constant vertigo, all of it passes to the machine. Comfort, here, is not a luxury; it is the absence of danger. It is also time returned, the time no longer spent rigging, climbing, waiting for a clear weather window.

Holding to the glass, reading the wind

Lowering a robot down a wall of glass requires that it know where it is. Ozmo combines a laser rangefinder with computer vision to map the facade, its curves, its edges, its frames. Force sensors gauge how much pressure each pane needs, and software corrects the arms' position when a gust shoves them. The challenge is not the scrubbing: it is staying precise dozens of floors up in the wind.

Others have placed different bets on the same problem. The company Verobotics launched a robot in Dallas, Ibex, that climbs the facade by suction, with no crane or rail on the roof: under ten kilograms, two legs, five cameras and some fifteen sensors to find its own way. KITE Robotics, for its part, builds an untethered machine carrying its own battery and water reservoir, free of any cord.

Three approaches, one idea: take the human off the wall and hand the glass to a mechanism that does not tire, does not shake, does not look down.

Someone is still watching from the roof

The autonomy, though, is partial. Today Ozmo is piloted and monitored by an operator stationed on the roof, screen in front of him. Full independence is promised for later, not for now. The robot has not erased the worker: it has moved him from the rope to the console. The "human-free" promise is, for the moment, a promise.

Then there is the matter of dependence. Cleaning your windows becomes a service you rent from a provider, often exclusively within a territory, as with the partnership struck in London to bring Ozmo to the British capital. The building does not own the robot; it subscribes to a vendor, to its prices, its schedule, its availability. The harness is gone, the contract has taken its place.

And there is cost and reliability. A young company that raises some ten million dollars to expand has not yet proven it will last, nor that its machine handles every facade, every wind, every recess a human hand once reached with a flick. The question of lost jobs, for its part, does not fully vanish, even if the shortage of recruits softens it.

What the machine sees on the way down

One detail slips by and deserves a pause. To guide themselves, these robots scan the facade continuously: cameras, laser and sensors sweep every square meter of glass. Yet behind that glass are offices, meeting rooms, sometimes apartments. A device descending slowly along a building while filming everything it passes is not only a cleaner; it is also a moving eye, whose data flows back to a server.

Nothing suggests these images are used beyond the cleaning itself. But the tool exists, and it watches. Handing window washing to a machine also means accepting that an automated gaze passes regularly in front of windows behind which people thought themselves unseen.

The founding intuition still holds: a job that kills should not keep killing once a machine can stand in. Facade robots spare falls, speed up maintenance, fill a real shortage. The promise will truly hold the day the human leaves the danger zone without merely trading the rope for a subscription and a camera lens trained on his windows.