Humyn Labs, a startup based in Bengaluru, India, has announced the closure of a $47 million Series A funding round aimed at scaling its operations to provide training data for artificial intelligence and robotics companies in the United States. The investment includes $26 million in equity led by Accel and Cactus Partners, with participation from Prosus Ventures, BeeGlobal, Bajaj Finance’s Finserv Intelligence and Bajaj Alts AI Opportunities Fund, Dallas Venture Capital, Lighthouse Canton, and Jump Trading. Additionally, the funding package features $21 million in venture debt led by Lighthouse Canton Growth Debt Fund II and Trifecta Capital.
The company addresses a critical challenge in the development of “physical AI”: the need for high-quality, real-world data to train robots that can perform everyday human tasks. While competitors such as Mecka AI, XDOF, and MicroAGI are also active in this space, Humyn Labs utilizes proprietary two-camera hardware to record individuals performing natural activities. The resulting footage is processed and labeled via software to track hand movements and object interactions, allowing AI models to learn from diverse human behaviors.
Co-founder Manish Agarwal, a former CEO of Nazara Technologies, and co-founder Ishank Gupta, a former executive at AB InBev, established Humyn Labs in February. The duo leverages their existing network of contributors in emerging markets to collect data in home environments across more than 20 countries in Latin America and Asia, with key operations in Brazil, India, and South Korea. According to Agarwal, the company focuses on recording natural workflows rather than staged activities, utilizing local partners to oversee collection sites and ensure data quality.
Humyn Labs currently serves 12 paying customers, including frontier AI labs and robotics developers, with approximately 90% of its revenue derived from the U.S. market. The company reported generating $10 million in revenue between April and October, alongside $20 million in signed contracts and another $20 million in pending deals. The venture debt portion of the new funding is designed to bridge the gap between daily or weekly payments to workers and the typical 30-to-60-day payment terms from customers.
As India emerges as a hub for physical AI data collection, the practice has drawn scrutiny regarding privacy and consent, particularly when recordings occur in private homes. Agarwal stated that workers are incentivized and monitored to ensure proper input, and that the company works with local partners to manage the collection process. The startup currently employs 51 full-time staff, primarily in Bengaluru, in addition to its global network of partners.
Physical AI needs more than just video, though. We need tactile and haptic data too. They are missing a big piece.
Venture debt to pay workers weekly is smart. Keeps the collection pipeline steady without burning cash reserves.
I wonder if this data is useful if it’s mostly from emerging markets and not Western households.
Making ten million dollars in just six months is actually impressive for such a young startup. They must have strong traction.
Is two-camera hardware really enough to capture complex spatial awareness? I feel like depth sensors are much more necessary.
Recording people in their homes without them realizing it sounds like a huge privacy nightmare waiting to happen.
It is exciting to see Bengaluru positioning itself as the global hub for physical AI data collection. Great progress.
Accel leading a robotics data fund is a massive signal. This space is heating up way faster than I expected.
Wait, they’re using home data for robots? I hope they are actually paying contributors fairly for all that footage.