Wear a Tiny Camera, Film Your Chores, Earn $6 to $18 a Task: Memories.ai Pays Ordinary People to Teach Robots

Picture a new side gig: you clip on a camera lighter than an egg and film yourself chopping vegetables, stocking shelves or fixing a bike, and you get paid if the footage qualifies. It's not a joke. San Francisco startup Memories.ai launched Memories Upload on Nov. 10, a platform that pays ordinary people around the world to record everyday physical work from their own point of view so the footage can train robots.

Background: robots don't lack computing power, they lack lessons in how to do the job

Chatbots got smart because the internet holds oceans of text to learn from. A robot learning to pour coffee or fold laundry has no equivalent library: it needs to see how a human hand grips a cup and avoids a spill. So whoever collects large amounts of first-person video of people working holds a scarce raw material for robot training. Fortune reports that humanoid robot company Figure's Index platform also pays people to record everyday tasks, and that Physical Intelligence has tested combining human video with remote-control data.

What happened

According to Fortune's exclusive on Nov. 10 (8:00 AM Eastern): Memories Upload is a global gig platform where contributors take recording tasks and wear the company's own LUCI Pin camera. The pin weighs under 45 grams, attaches magnetically to the head, chest or wrist, records 4K video with audio, and two of them can be paired for stereo views that help estimate depth. Fortune reports payouts of $6 to $18 per task, with tasks such as food prep, stocking shelves and repairing equipment. The company says its network recorded more than 1 million hours of first-person video in the platform's first month in beta and aims for more than 10 million hours by March 2027. It was founded by former Meta researchers and has raised $16 million.

Memories.ai says its customers include three of the top five frontier AI labs plus robotics developers that buy training data, but it declined to name them, citing non-disclosure agreements. That is the company's own claim and can't be independently verified. It also says it uses Nvidia's Cosmos models to caption videos and label actions automatically, so developers can search footage for a specific hand movement the way they search text.

What about privacy?

Filming at work inevitably catches coworkers, passersby and information on screens. Per Fortune, the company says camera wearers and coworkers who regularly appear must sign consent forms; before footage leaves the country where it was captured or reaches customers, automated tools blur other people's faces, badges, screens and license plates, and staff manually review a sample of each batch. The company's terms of service, which carry no date on the page, also say contributors pay a refundable deposit to borrow a camera and that Memories.ai owns the recordings.

What it means for everyday people

In the short run, it's one more gig for people who want extra income, but the $6-to-$18 pay, the company's ownership of the footage and the chance that your hands and voice could be identified all deserve a careful read of the terms first. In the long run, if "humans showing robots how to work" video grows at scale, household and warehouse robots could learn faster. At the same time, who gets filmed and how that data is used will become a more practical privacy question.

Original source published: Fortune, Nov. 10, 2026, 21:00 Beijing time / 8:00 AM EST