
18 Nov 2026
Humyn Labs raises $47 million to sell robot makers footage of people doing chores
Bengaluru-based Humyn Labs raised a $47 million Series A, $26 million of it equity led by Accel, to record people doing everyday tasks and turn the footage into training data for robots.
Robots can't learn from the internet the way chatbots did, so someone has to film the dishes getting done. The revenue figures are the company's own, and the customers aren't named.
On Wednesday, Nov. 18, TechCrunch reported that Bengaluru-based Humyn Labs raised $47 million in a Series A. A Series A is a round of venture funding, money from investment firms in exchange for a stake, used to grow a company that is already selling. Of that money, $26 million is equity led by Accel. Equity, here, means the investors bought a share of the company. Prosus Ventures, BeeGlobal, Bajaj Finance's Finserv Intelligence and Bajaj Alts AI Opportunities Fund, Dallas Venture Capital, Lighthouse Canton, and Jump Trading joined that equity. Another $21 million is venture debt, a loan rather than more shares, led by Lighthouse Canton with Trifecta Capital. The size of the round, the split, and that investor list are the company's, as Humyn told TechCrunch.
What the company does, as TechCrunch describes it: its own hardware, with two cameras, records people doing everyday tasks. Software then tracks hand movements and how hands meet objects, and labels the clips. Labeling means tagging what happens in each clip so a model can learn from it. Humyn collects that footage in more than 20 countries across Latin America and Asia. Brazil, India, and South Korea are the main hubs. It films in people's homes through the contributor network of KGeN, the founders' earlier company. Those lines are the company's, as told to TechCrunch.
The founders are Manish Agarwal, former chief executive of the Indian gaming company Nazara Technologies, and Ishank Gupta, a former executive at AB InBev, the brewing company. They founded Humyn Labs in February 2026. Those lines are TechCrunch's.
The business figures are the company's, and the customers are unnamed. Humyn told TechCrunch it has 12 paying customers, including frontier AI labs, robotics companies, and developers of world models. A world model is a model aimed at how the physical world changes, rather than only the next word in a sentence. About 90 percent of revenue comes from the United States. Ishank Gupta gave TechCrunch that share. Manish Agarwal said the company made $10 million in revenue from April to October, has another $20 million in signed contracts, and has a further $20 million in deals that are not finalized yet.
The debt has a cash-timing reason, in Agarwal's account. Humyn pays workers and data partners daily or weekly, then waits 30 to 60 days for the AI-lab customers to pay. The loan covers that wait. He also said debt dilutes the owners less than selling more equity. Dilute, here, means each new share sale leaves the existing owners with a smaller slice. That explanation is his, as TechCrunch reported it.
TechCrunch names three companies already collecting this kind of real-world footage for robot training: Mecka AI, XDOF, and MicroAGI.
Agarwal's line to TechCrunch: "Scale, diversity, and quality are important aspects, and they need to work in tandem."
The picture is the two co-founders outdoors, in black shirts with an orange humyn(labs) logo, trees and a parked car behind them. Ishank Gupta is on the left, with glasses and a beard. Manish Agarwal is on the right, with glasses and a mustache, his arm on Gupta's shoulder. It is the company photograph TechCrunch published with the story. The credit is Humyn Labs, via TechCrunch.
In plain terms, a Bengaluru startup that films household tasks and sells the labeled footage to robot makers raised $47 million. Accel led the equity. A loan covers the gap while customer payments lag. The revenue, the 12 customers, and the 90 percent U.S. share are Humyn's figures, as told to TechCrunch. The customers are unnamed.
RELATED
Sources
- TechCrunch
techcrunch.com



















