Imagine getting paid a dollar an hour to wear a camera on your head while you fix someone’s AC.
That’s the deal Human Archive is offering India’s gig workers. And depending on who you ask, it’s either a smart on-ramp to the AI economy — or something that sounds way too dystopian to ignore.
I spent some time reading about this startup, and I genuinely can’t decide if I’m impressed or unsettled. Maybe both.
Wait, what exactly are they doing?
Human Archive partners with smaller home-service startups. Here’s how it works: you need a plumber or an electrician at home. You get a choice — pay less if you let them record the whole thing on video. Pay full price if you’d rather not be on camera. The worker shows up wearing a cap with a camera strapped to it, filming everything from their point of view.
The real customer isn’t you, though. It’s the AI model that’ll train on all that first-person footage later.
Human Archive says it’s creating a “flexible earning opportunity.” Workers get about a dollar an hour. For context, other platforms in India pay ₹250–400 per hour — roughly $3 to $5. So the rate here is noticeably lower. The company’s argument? This kind of work literally didn’t exist before. They’re not competing with gig platforms; they’re creating a new category.
I’m not sure that argument holds up when you look at the math, but that’s their story and they’re sticking to it.
From iPhones to full-body motion capture
The hardware side is genuinely fascinating. The founders — brothers Parth and Rudra Patel — started with iPhones. Then they built their own caps. Now they have seven different hardware products: tactile gloves that capture touch, full-body motion capture suits, wrist cameras, depth-sensing headsets. All of it synced together.
Over a thousand headsets are already deployed. Wing VC’s Zach DeWitt calls this a moat: “No one else in the world has been able to synchronize and collect headset RGB-D, force feedback, full-body motion capture, and wrist camera data at scale.”
That’s a pretty wild claim — and if it’s true, it means Human Archive is sitting on a dataset that AI labs would kill for.
The messy part
This is where it gets interesting. Urban Company — one of India’s biggest home services platforms — rejected a partnership. CEO Abhiraj Singh Bhal went public on X saying they wouldn’t participate in this kind of data collection. Human Archive fired back, suggesting Urban Company would “soon be forced to reconsider or risk losing relevance.”
Pronto, another platform, also said no. Co-founder Rushil Agarwal claims their founder called him “stupid” for suggesting the idea. Pronto denies that happened.
These rejections tell you something. The big players don’t want to touch this model — at least not publicly.
The real question nobody’s answering
A worker gets $1/hour. Their video, their movements, their tactile feedback — all of it trains AI that could be worth billions. The gap between what the worker earns and what the data is worth is so wide it makes you uncomfortable just thinking about it.
Human Archive frames this as participation in the AI economy. Critics call it something else.
I don’t have the answer. But I do know this: the robots of tomorrow are being trained on the labor of today’s gig workers. And the people wearing those camera caps right now? They’re earning less than what most coffee shops pay.
Human Archive is forging ahead anyway. Building datasets. Running experiments with “every major lab and university.” The physical AI train is leaving the station — and they’re making sure India’s workers are on it.
The question is whether they’re passengers or fuel.
What do you think — is this the future of AI training data or a modern exploitation story dressed up in startup language? I’d genuinely love to hear your take in the comments.