There’s really no such thing as free house cleaning. You already knew that. But the German startup MicroAGI is banking on you forgetting that basic rule of thumb, at least long enough to let a stranger in wearing a camera.
The company launched its Shift app in New York City last month, advertising free professional house cleanings for residents willing to let workers film everything they do. The pitch is wrapped in that familiar startup optimism: here’s a free service, and in exchange, you’re helping train the next generation of household robots. What’s not to love?
Plenty, actually.
According to Ars Technica reporting, the Shift app connects New Yorkers with what it calls “trusted professional house cleaners” who wear cameras capturing first-person cleaning footage. That data gets fed into MicroAGI’s AI systems, meant to teach robots how to navigate and clean human homes. The company describes itself as a team on a mission to accelerate embodied AI, which is tech jargon for robots that can move around and interact with the physical world.
The promotional video set to “Empire State of Mind” makes it all look相当 slick. Harry Kilberg, the company’s US general manager, appears alongside cleaned homes and smiling workers, promising that the platform already pays “tens of thousands of people” across fifteen countries to record daily work. The main Shift website claims more than ten thousand “operators” have been collectively paid over five million dollars in the first quarter of 2026.
Here’s the thing though. While you’re watching that cheerful promotional content, you’re also being asked to hand over your home address, phone number, and access instructions. The fine print reveals you’ll be charged if you cancel with less than twenty-four hours notice or aren’t available at the appointment time. The terms of service also include liability waivers for property damage, theft, or personal injury. That’s a lot to sign up for what amounts to a couple hours of free cleaning.
The privacy side of this gets murky fast. The Shift app’s FAQ claims names, faces, and personal information are automatically anonymized, with sensitive details blurred before any data gets used. The privacy policy mentions “advanced machine learning models” running on smart glasses to perform irreversible transformations like face blurring before uploading to cloud servers. But there’s no explicit mention of whether you can ever request your videos be removed from training datasets. It’s also unclear whether the anonymization is robust enough to prevent someone’s home from being identified later.
This puts MicroAGI in familiar territory with other data collection startups. Companies like Encord and Micro1 have been recruiting ordinary people to record everyday tasks for robot training, hiring thousands of workers across dozens of countries according to reporting from MIT Technology Review. The business model is simple: pay people small amounts to generate massive amounts of training data that would otherwise cost a fortune to produce.
What makes this particular pitch interesting is how it layers the free cleaning service on top of the broader recruiting operation. The promotional video for the free cleanings briefly mentions that the app’s primary function is paying people twenty dollars per hour plus bonuses to wear a “recording headstrap” and capture videos of household or professional tasks. The free cleaning is almost like a loss leader to get more New Yorkers comfortable with the concept of strangers filming inside their homes.
MicroAGI has been aggressively targeting NYC, with blog posts aimed at university students, teachers, restaurant and delivery workers, and specific neighborhoods. They’ve also posted on Craigslist looking for residents in Boston, and the founder has hinted at expansion to London, Munich, and Zurich.
At the end of the day, what you’re really deciding is whether a few hours of free cleaning justifies letting a company build a digital replica of your living room that may exist in some training dataset indefinitely. The startup calls it “no catch.” That’s one way to look at it.
