AI training startup Shift has announced an unusual offer: free home cleaning services for customers willing to let the company record the cleaning process. The footage will be used to train future robots to perform household chores. The company, which already pays tens of thousands of people across 15 countries to record everyday tasks through its app, sees this as a scalable way to gather high-quality training data.
The cleaning service is marketed as a win-win: customers get a spotless apartment, and Shift gets valuable data to improve its AI models. The company's co-CEO and co-founder, Bercan Kilic, explained that the value of the training data generated from each cleaning session outweighs the cost of providing the service. In promotional materials, Shift emphasizes that the data collected will help create robots capable of autonomously cleaning homes in the future.
Cleaners wear a special 'magic hat' equipped with a camera that records their work from a first-person perspective. This footage captures every action—vacuuming, mopping, dusting, washing windows, scrubbing dishes, and tidying up—along with the environment and objects encountered. Shift says the data is processed to blur or anonymize sensitive information such as faces, names, and personal details visible on screens or ID cards before it is used for AI training.
The company stresses that cleaners are vetted through its partners, although they are not Shift employees. The service initially launched in New York City, with plans to expand 'very soon' to San Francisco, London, Zurich, and Munich. Free cleanings are available for a limited time, and Shift's website explicitly states that dirtier homes are especially useful for training, as more challenging environments provide richer data. However, cleaners may decline tasks they are uncomfortable performing.
Shift's approach fits into a broader trend in AI development: the growing demand for real-world video recordings of human activities to train robotic systems. Companies like Google, OpenAI, and others have explored similar methods using wearable cameras or remote-controlled robots to collect data. Shift's model is particularly notable because it turns a consumer service into a data collection pipeline, potentially reducing the cost of obtaining diverse, high-quality training data.
The video promoting the service shows a cleaner in a crisp white uniform performing various tasks while wearing the camera hat. It ends with the tagline: 'Every home cleaned today lays the groundwork for a home that cleans itself tomorrow.' Beyond cleaning, Shift's video hints at future expansion into other household and trade services such as plumbing, cooking, and building. This suggests the company envisions a broad range of robotic applications that require training on human expert performances.
Critics may raise concerns about privacy, even with anonymization. Footage recorded inside a home can still reveal a great deal about a person's lifestyle, belongings, and habits. Shift says it will blur faces and personal information, but the overall environment—wall colors, furniture arrangements, even the types of objects present—could remain identifiable. Additionally, the recording happens in real time, meaning there is a potential risk of exposure if data is compromised. Shift has not detailed its security protocols beyond claiming that data is safely stored and processed.
Another point of contention is the labor aspect. While customers receive free cleaning, the workers themselves are not Shift employees and are hired through partner agencies. This raises questions about job security, wages, and the long-term impact of automation on cleaning jobs. Shift's business model essentially monetizes the labor of cleaners by using their physical work to train the machines that may eventually replace them. The company frames this as part of a transition, but the ethical implications are significant.
The service is currently limited to New York City, but Shift has a history of global data collection. Its app, which pays users to record tasks, operates in 15 countries spanning North America, Europe, and Asia. The expansion into home cleaning marks a shift from passive data collection (users voluntarily recording at home) to an active, organized service. It also demonstrates a move toward gathering more structured and consistent visual data, which is crucial for training robots to perform physical tasks reliably.
Shift's co-CEO and co-founder Bercan Kilic has a background in robotics and computer vision. Before founding Shift, he worked on AI projects at major tech companies and research labs. The startup has raised significant venture capital, though the exact amounts have not been publicly disclosed. Its core product is a platform that connects people willing to record their everyday activities with companies and researchers needing training data. Shift processes the raw footage, adds annotations, and sells the datasets.
The home cleaning initiative is just one example of how AI companies are creatively sourcing data. Others have used similar tactics: one startup offered free food delivery in exchange for recording the delivery process; another paid people to narrate their daily commute while wearing eye-tracking glasses. As competition for high-quality training data intensifies, such barter arrangements are likely to become more common.
Despite the allure of free services, customers should weigh the trade-off carefully. A free cleaning might be worth the privacy cost for some, especially those who already use smart home devices that record audio or video. But the long-term consequences—both for personal data privacy and for the labor market—are non-trivial. Shift's offer underscores a fundamental tension in AI development: the need for massive amounts of real-world data often collides with individual rights and societal norms.
For now, Shift's cleaning service is a proof-of-concept for its data collection strategy. If successful, it could scale to other cities and other types of domestic work, creating a vast library of human task executions. The company's broader vision is clear: a future where robots clean, cook, and maintain homes autonomously, powered by data gathered from today's human workers. Whether that future is desirable depends on how well the transition is managed and whose interests are prioritized.
Source: The Verge News