Humyn Labs Raised $47 Million to Film People Doing Chores So Robots Can Learn Them
Bengaluru startup Humyn Labs raised $47 million in Series A funding on November 18 to record people doing everyday tasks, giving US robotics companies the real-world data their robots need.

The Humyn Labs funding round announced on November 18 is a bet on a simple idea: robots cannot learn to do everyday tasks from the internet, because the internet has almost no footage of people doing them. Bengaluru-based Humyn Labs raised $47 million in Series A financing to fix that gap by filming people doing real work, in real places, with its own hardware. The round mixes $26 million in equity led by Accel with $21 million in venture debt led by Lighthouse Canton, and it comes just nine months after the company was founded in February 2026.
What Humyn Labs Actually Does
Humyn Labs records how people do everyday work so robots can learn from it. The company built its own two-camera hardware to capture people performing ordinary tasks, and it collects that data in more than 20 countries across Latin America and Asia, with Brazil, India, and South Korea as its main hubs. The customers are the people building the robots: frontier AI labs, robotics companies, and developers of world models. The company says it has 12 paying customers, and about 90 percent of its revenue comes from the United States, which is where most of the serious humanoid robotics work is happening right now.
The problem it solves is real and underappreciated. Language models learned from trillions of words scraped from the web. Robots have no equivalent. There is no giant public dataset of someone folding laundry, loading a dishwasher, or picking up a dropped screwdriver, filmed from the angles a robot needs. Simulation helps, but simulated hands do not fumble, slip, or improvise the way real ones do. Companies building robots that work in homes and warehouses need footage of the physical world, and somebody has to go out and record it.
The Numbers, With a Grain of Salt
The company reports $10 million in revenue between April and October, another $20 million in signed contracts, and a further $20 million in deals that have not been finalized. It also says it crossed $50 million in annualized revenue within six months of operations. That last figure needs a footnote: it is a run rate, a recent month projected across a year, not $50 million already banked, and the numbers are company-reported rather than audited. The $10 million actually collected is the solid number, and it is still impressive for a company that is nine months old.
The funding will go to three things: deployment-readiness infrastructure in Brazil, South Korea, and India, where robots get tested in real environments before they ship; the company's data-capture hardware and processing engine; and on-the-ground presence in San Francisco, London, and Seoul, close to the customers paying for the data. Accel partner Pratik Agarwal backed the round, with Prosus Ventures, BeeGlobal, Bajaj Finance funds, Dallas Venture Capital, Jump Trading, and former Epic CFO Anita Pramoda joining the equity, and Trifecta Capital joining the debt.
Why Investors Are Paying for Footage
The timing is the story. Physical AI, robots that act in the world rather than chat about it, is where venture money is rushing this year. Mirendil raised $1 billion to build AI agents for frontier research, and LangChain raised $300 million for the infrastructure those agents run on. Every one of those bets assumes robots will soon do useful physical work. Humyn Labs is selling the picks and shovels for that bet: the training data without which the robots stay clumsy. A July 2026 Counterpoint Research report estimated India's data collection and annotation services for physical AI could be a $10 billion opportunity by 2030.
There is a real question underneath the excitement, and it is about quality. Filming people doing chores is easy. Filming them in a way that actually teaches a robot is hard: the angles, the timing, the force, the recovery from mistakes. Humyn Labs' claim is that its proprietary hardware and enrichment pipeline capture the details that matter. That claim is unproven in public so far, and the 12 unnamed customers are not enough to judge it. What is proven is demand. Twelve paying customers in nine months, for a product category that barely existed two years ago, tells you the robotics labs are hungry.
The Bottom Line
Humyn Labs is a nine-month-old company selling footage of people doing laundry to the richest AI labs in the world, and it just raised $47 million to do more of it. That sentence sounds like a joke, but it is a reasonable description of where physical AI stands: the models are ready, the hardware is nearly ready, and the data is the bottleneck. If robots are going to fold your laundry in five years, somebody has to film a thousand people folding laundry first. Humyn Labs wants to be the company that films them, and Accel just bet $26 million that the footage is worth more than the cameras.


