AI company offers free cleaners in New York to train robots

The free cleaning was only the lure. Shift, the New York City brand tied to Germany-based MicroAGI, offered limited-time home service in exchange for first-person footage it said would train the next generation of household robots.

The company said customers could book an appointment and have an apartment cleaned at no charge while cleaners recorded laundry folding, vacuuming, mopping, dishwashing, bathroom scrubbing, fridge organizing and other chores. Shift said its system captured first-person video, motion and sensor data, and metadata including GPS location, phone model, camera parameters and the operator’s height.

That data collection sits at the center of the business model. Shift’s privacy policy says recordings may capture the interior and contents of a home, along with other people who are present. The company says sensitive details such as faces, names, screens and ID cards are blurred or anonymized before processing, but the setup still turns private domestic labor into a training set for automation.

Harry Kilberg, Shift’s U.S. general manager, said the launch drew demand for “thousands and thousands of bookings,” and described the company’s goal as to “democratize the AI economy.” Shift also said it already operates in more than 15 countries and works with more than 10,000 businesses and households, a scale that suggests the New York promotion is being used as both a customer acquisition tactic and a data-gathering exercise.

The broader strategy reaches well beyond cleaning. Shift has signaled plans to expand into more cities and add services such as handymen, repairs, errands, cooking and plumbing data collection. The company says it is already advertising that it can help shape the future of robotics, making the free-cleaning offer look less like a giveaway than a pilot program for a much larger labor market.

The privacy and liability questions are just as large. Reporting on the service says Shift’s legal documents disclaim responsibility for theft, injury or property damage, while cleaners are treated as independent contractors rather than employees. That leaves a familiar imbalance in place: the household carries the risks, while the company collects the footage and the potential value.

Shift’s bet is straightforward. Real homes, real messes and real human routines are the scarce ingredient in embodied AI. By paying for recordings of ordinary domestic work, the company is trying to build the dataset it believes robots will eventually need to do that work themselves.