A New York startup has launched a free household-chore service to collect first-person video data for humanoid robot AI [1, 2].
The initiative represents a strategic shift in AI development, moving from digital data scraping to the collection of physical, real-world motion data. By recording how humans navigate domestic environments, the company aims to solve the complex problem of robotic dexterity in unpredictable home settings.
Launched in May 2026 [2], the service is called "Shift" [1, 2]. The company provides household labor to residents at no cost [1]. In exchange for the free service, staff members wear hats equipped with cameras that record their every movement and interaction within the home [1, 2].
According to the startup, the service is designed for efficiency, with a maximum response time of two hours for a request [1]. This high-volume data collection is intended to give the company a competitive edge in the global race to develop functional physical AI [1, 2].
Harry Kilberg leads the project [1, 2]. He said the ultimate goal is to create humanoid robots that can help with various household chores that humans do not want to do [1].
The data gathered via the staff's perspective allows the AI to learn the nuances of cleaning, organizing, and navigating a home, tasks that are difficult to simulate in a laboratory. By using human proxies to map these movements, the startup intends to accelerate the training of robots that can eventually replace the human staff of Shift [1, 2].
“The ultimate goal is to create humanoid robots that can help with various household chores that humans do not want to do.”
This approach signals a move toward 'embodied AI,' where the bottleneck for robotic advancement is no longer processing power, but high-quality physical interaction data. By subsidizing human labor to generate this data, the startup is treating the New York residential market as a live laboratory to bridge the gap between digital intelligence and physical execution.
