microagi's Shift offers free apartment cleaning if AI can watch
Founded by two former Formula One engineers and an AI researcher, microagi is testing whether chores can become robotics training data.
By Ryan Merket ยท Published
Why it matters
Shift shows how far robotics companies may go to get real-world training data: subsidizing everyday labor today in the hope that machines can learn it tomorrow.
Bercan Kilic (@bercankilic), Yoan Iliev and Anton Poletaev's German AI lab microagi is using Shift Robotics to offer $0 apartment cleanings in New York City in exchange for first-person video of the work, Business Insider reported.
The bargain is simple and unusually literal: Shift cleaners wear head-mounted cameras while doing dishes, mopping floors and folding laundry. Shift then turns that footage into training data for AI labs and robotics companies, according to Business Insider. The consumer sees a free cleaning. microagi sees a way to manufacture a data corpus for machines that do not yet have the equivalent of the internet-scale text and image troves that powered large language models.
Kilic and Iliev are former Formula One aerodynamic engineers, while Poletaev previously worked as a researcher at The Alan Turing Institute, according to Business Insider. Their lab, founded in a Munich hacker house last year and headquartered in Germany, says it is working on "end-to-end physical AGI," a phrase that points to the real target: not cleaning apartments as a services business, but collecting enough examples of physical work to train systems that can eventually act in homes, warehouses and other messy environments.
The viral offer is also a data acquisition machine
Shift's launch video on X has drawn more than 8 million views, according to Business Insider, and the initial 250 cleaning slots sold out almost immediately.
Harry Kilberg, Shift's US general manager and the face of the video, told the outlet, "We had thousands and thousands of people trying to book."
Kilberg said he expected the response. "We knew this idea is world-changing, so we thought it would go viral," he told Business Insider.
That is the optimistic version of the pitch. The harder business question is whether a free household service can generate data valuable enough to pay for labor, operations and quality control. Kilberg told Business Insider that "the unit economics are a lot better than you think," saying microagi's internal processing makes the data higher quality and able to command a premium from AI labs and robotics companies. No buyer names, prices, gross margins, funding figures or valuation were disclosed in the reporting.
The worker side is also still opaque. Business Insider reported that Kilberg declined to say how much people are paid to record themselves doing chores. He said Shift operates in 15 countries and has 14,000 operators collecting real-world data, but those figures are company claims through Kilberg rather than independently verified numbers.
Privacy is part of the product claim
Recording inside apartments raises immediate privacy questions. Kilberg told Business Insider that faces and screens are automatically blurred and that the system captures no audio. microagi also uses the footage for its own internal research, according to the report.
Those safeguards matter because Shift's best data is also its most sensitive: first-person footage of private spaces, objects and routines. The more varied the homes, tasks and geographies, the more useful the data could be for robotics companies trying to handle the irregularity of real life. But the same variety also makes consent, anonymization and building access more important than they would be for a conventional cleaning marketplace.
Kilberg described Shift to Business Insider as a marketplace "aimed at accelerating the transition from an economy where people work out of necessity, to one where everyday goods and services can become increasingly abundant and accessible." In the near term, though, the model still depends on people doing the chores and wearing the cameras.
A small launch aimed at a big robotics bottleneck
Shift's bet lands as OpenAI (@OpenAI), Nvidia, Meta and Tesla are all pushing deeper into robotics or physical AI, according to Business Insider. Data suppliers that fed the chatbot boom, including Scale AI, Turing and micro1, have also moved into real-world data collection, the outlet reported.
The reason is structural. Chatbots could learn from vast stores of internet text. Robots need examples of bodies moving through physical spaces, handling objects, correcting mistakes and adapting to clutter. Business Insider cites UC Berkeley roboticist Ken Goldberg's "100,000-year data gap" as shorthand for how far robotics trails language models in available training examples.
Shift is starting with cleaning in New York, but Kilberg told Business Insider the plan is to expand across the US and add free or subsidized services such as cooking and plumbing. He also said early users had already been recording themselves doing tasks at home, posting flyers to clean neighbors' apartments because microagi would "foot the bill," and recording work in bodegas and soup kitchens.
That makes Shift less a cleaning startup than a test of whether founders can turn subsidized human work into the missing raw material for robotics. If microagi can prove the data is scarce, clean and valuable, the free apartment cleaning is customer acquisition for a much larger market. If not, it is an expensive way to buy video of people doing the jobs robots still cannot reliably do.
Social posts: (none)