Physical AI startup Humyn Labs, which collects data from factories and people’s homes to train AI systems and humanoid robots, has raised $47 million in a Series A funding round.
The funding includes $26 million in equity financing led by Accel, with participation from Prosus Ventures, which has previously backed Swiggy, Urban Company, Meesho, PayU, and PharmEasy, among others.
BeeGlobal, Bajaj Finance’s Finserv Intelligence, Bajaj Alts AI Opportunities Fund, Dallas Venture Capital, Lighthouse Canton, and Jump Trading LLC also joined the startup’s cap table.
Meanwhile, Lighthouse Canton led the $21 million venture debt round, with participation from Trifecta Capital.
Founded in 2026 by Manish Agarwal, former CEO of Indian gaming company Nazara Technologies, and Ishank Gupta, a former executive at Belgian brewing company AB InBev, Humyn Labs is among a growing crop of companies that are collecting footage of human activity in real-world settings and feeding it to US-based AI and robotics companies, which need this data to train physical AI models.
McKinsey and Morgan Stanley put the physical AI market at $25-30 trillion by 2050.
How Humyn Labs collects robot training data
The Bengaluru-based startup pays people to wear cameras while they work in factories, warehouses, retail stores, farms and homes. Then, its software turns that first-person video into training data for frontier AI labs, humanoid robotics companies, voice AI firms, etc.
The company claims it has recorded more than 24 million video episodes, captured over 1 million hours of audio footage and has at least a dozen paying customers.
- What they record: Humyn Labs pays people to use head-mounted cameras, including its own egoS1 device and film themselves while they fold laundry, cook a meal or wash dishes, per Forbes and the company’s official website. While most human-generated data comes from video, sometimes collectors wear hardware on their hands and fingers to collect touch data, including movements of the fingers, wrists, elbows and shoulders, as well as the forces applied during these movements.
- How they source it: Humyn Labs taps into an existing network of partners – originally built through the founders’ previous gaming venture, KGeN – across more than 20 countries, with major hubs in India, Brazil, and South Korea. Workers are paid daily or weekly for their recordings and are supervised by local partners, TechCrunch reported.
- Privacy measures: According to coverage by Humanoid Home News and the BBC, Humyn Labs claims that privacy-sensitive details, such as faces and children, must be excluded or blurred during capture to protect household privacy. MediaNama could not independently verify these claims. Queries sent to Humyn Labs remained unanswered at the time of publishing.
- Where is the data stored? “We store data on servers located in India and may use service providers that process data outside India, in compliance with applicable law,” Humyn Labs’ privacy policy reads.
According to Crunchbase, investors pumped $47.4 billion into physical AI in the first six months of 2026, nearly four times what they invested in the second half of 2025. The $47 million investment in Humyn Labs suggests that investor appetite for physical AI startups is unlikely to wane anytime soon.
Over the past few months, India has emerged as a breeding ground for data to train robots and physical AI models, with startups like Pronto tapping into the country’s large gig workforce to record people doing everyday tasks.
However, the practice has also raised concerns around surveillance, consent and children’s data. In May this year, Pronto came under the scanner of the Ministry of Electronics and Information Technology (MeitY) after reports emerged that it was running a pilot that involved service professionals wearing cameras to collect video data of tasks they performed inside customer homes.
The controversy prompted the founders of rival home services startups Urban Company and Snabbit to publicly deny involvement in similar practices. Read our full coverage of Pronto’s in-home recording pilot here.
Then, there are concerns about data sovereignty. At SuperAI 2026, Yanliang Zhang, chief scientist at Western Robot, argued that hardware was not the main bottleneck in robotics, but the data needed to train them has “sovereignty” issues.
“Because of the regulations, the data we collect in Singapore might not be able to use in the (United) States. The data you collect from China, (it) is unlikely you can use (it) in your country. Because this kind of challenge force(s) us to re-collect data for the same problem. However, for the conventional automation business, you develop everything in your factory, tomorrow you can sell to the world. But today, cannot. Data is an asset. Data is national security. This is a challenge we are facing today,” Zhang said.
Earlier this month, MeitY summoned representatives of Pronto, Urban Company, Snabbit and robotics companies and asked whether they were selling or sending footage recorded inside Indian homes to physical AI developers outside India.
Also read:
- The physical world isn’t ready for physical AI
- Is footage from Indian homes going abroad to train robots? MeitY questions Pronto and other platforms
The post Humyn Labs raises $47 mn to collect training data for physical AI appeared first on MEDIANAMA.





