Bengaluru-based Humyn Labs has raised $47 million, split between a $26 million equity round led by Accel and $21 million in venture debt led by Lighthouse Canton, to supply training data to robotics and AI companies in the United States. Co-founder Manish Agarwal detailed the round and the company’s business to TechCrunch.
Trifecta Capital also participated in the venture debt, while Prosus Ventures, BeeGlobal, Bajaj Finance’s Finserv Intelligence and Bajaj Alts AI Opportunities Fund, Dallas Venture Capital, Lighthouse Canton and Jump Trading joined the equity side. The company was founded in February 2026 by Agarwal, formerly chief executive of Nazara Technologies, and Ishank Gupta, a former AB InBev executive. The two had previously founded Kratos Studios, now known as KGeN.
What Humyn Labs does
Humyn Labs uses its own two-camera hardware to record people performing everyday tasks, then processes and labels the footage, tracking hand movements and object interactions to turn the raw recordings into structured AI training data. It sells that data to frontier AI labs, robotics companies and developers building so-called world models, which simulate how objects and environments behave.
The company says it has 12 paying customers, with about 90% of its revenue coming from the United States even though most of its data collection happens elsewhere. It gathers footage in more than 20 countries across Latin America and Asia, with Brazil, India and South Korea as its main hubs, working through local partners and a contributor network that includes recordings made in people’s homes.
Why the round includes venture debt
Humyn Labs pays its data-collection workers and local partners daily or weekly, but typically waits 30 to 60 days to get paid by its AI lab customers. Agarwal said the venture debt portion of the round exists specifically to bridge that gap between fast outgoing payments and slower incoming ones, rather than to fund general growth.
The company has 51 full-time employees, most of them based in Bengaluru. It generated $10 million in revenue between April and October this year, and says it has $20 million in signed contracts with AI lab customers, plus another $20 million in deals that are not yet finalised.
What the founders said
Agarwal told TechCrunch that “scale, diversity, and quality are important aspects, and they need to work in tandem” for the training data to be useful to AI developers. He also described the people doing the recording as incentivized workers who still need monitoring and supervision to ensure the input they capture is usable.
Gupta gave TechCrunch the customer count and revenue mix figures, underscoring that the business depends on steady demand from a small number of large AI lab customers even as its data collection operation spans dozens of countries.
Frequently asked questions
What does Humyn Labs actually sell?
Labelled video and sensor data of people performing everyday tasks, captured with its own camera hardware, which AI labs and robotics companies use to train models on real-world human movement and object interaction.
Where does the data come from?
The company collects footage in more than 20 countries, with Brazil, India and South Korea as its main hubs, using local partners and a contributor network that includes people’s own homes.
Whether Humyn Labs can keep adding large AI lab customers at the pace its venture debt assumes is likely to shape its next funding round.



Leave a Reply