Humyn Labs raises $47M to bridge the gap between AI training and real-world robotics
Humyn Labs, a company turning human experience into training for robots and artificial intelligence, today announced it has raised $47 million in early-stage funding to accelerate intelligent robotic deployment readiness in real-world environments across Brazil, South Korea and India.
Accel led the $26 million equity portion of the round, with participation from Prosus Ventures, BeeGlobal, Finserv Intelligence through Bajaj Finance, Bajaj Alts AI Opportunities Fund, Dallas Venture Capital, Lighthouse Canton and Jump Trading LLC. Lighthouse Canton led the remaining $21 million in venture debt, with participation from Trifecta Capital.
Founded in early 2026, Humyn’s proposal is that AI models for humanoid robots don’t learn dexterity from staged demonstrations; they must learn by observing how humans move in real environments. This provides actual behaviors showing how hands work on objects, reach, touch, handle and recover without explicit instructions.
Unlike generative AI models that learn from vast amounts of text, images and video, physical AI models must learn how to interact with the world around them. That means understanding not only what objects look like and where they are, but also how to manipulate them, how much force to apply and how to respond when something goes wrong.
“The results can be very, very, very dramatic and catastrophic,” co-founder and Chief Executive Manish Agarwal told SiliconANGLE in an interview.
Agarwal pointed out how mistakes that might be inconsequential for a chatbot can have serious consequences for a robot operating alongside people. That makes access to diverse, high-quality data about how humans physically interact with their environments critical for training robots to work safely and reliably in the real world.
To address this, the company built a video corpus captured from head-mounted cameras as participants performed everyday household tasks, comprising up to 200,000 hours of intentional human motion. The company also shipped a dataset of 120,000 faces across five regions for facial recognition, including adjustments in lighting, pose, expression, glasses, hairstyle, facial hair, accessories, age and locale.
That’s combined with more than 50,000 hours of representative speech data for underserved language markets, including Bengali, Tamil, Telugu, Odia, Assamese, Korean, Tagalog, Arabic, Vietnamese and Portuguese. Voice AI models that understand natural language already have vast amounts of data for widely spoken languages such as English, Mandarin or Spanish, but real-world customer support within South and Southeast Asian markets lacks routine coverage.
Humyn’s engine uses sound, sight, motion and touch as data inputs, which it curates, normalizes and labels so that AI models can understand the kinematics of motion, force and environmental factors. Together, this provides training patterns for physical AI models, world models, general foundation models, and audio and voice AI.
Bringing diversity to robotic data
Geographic diversity is also important because robots must understand the environments and cultural practices where they will operate. “Diversity gets you in the room. The quality ensures that you stay in the room,” Agarwal said.
To make this point, Agarwal emphasized the importance of geographic and cultural diversity in collecting training data. A household robot trained in the United States may struggle with tasks in an Indian home, where everything from cooking utensils and kitchen appliances to laundry practices and room layouts can differ. These regional differences mean that robots need experience with the specific environments and practices of the people they will eventually serve.
By collecting data across more than 20 countries, Humyn can train models on data that more closely resembles their intended deployments, potentially accelerating readiness for real-world work.
Agarwal added that the company is seeing interest from customers building household robots for domestic tasks and elder care, alongside industrial applications involving object manipulation and movement through factories, warehouses and commercial environments.
“The next phase in the deployment cycle is about how well these models perform in the wild,” Agarwal said.
In the end, data is only the beginning. Humyn wants to measure where robots fail in real-world conditions and use that to feed the AI flywheel and build better models. That means using virtual simulations, physical replicas and partnering with the enterprise and researchers to have field deployment labs to identify the differences between expected and actual performance.
The company said it intends to use the new funding to build deployment-readiness infrastructure in Brazil, South Korea and India, develop its hardware and enrichment engine, and expand its on-the-ground presence in San Francisco, London and Seoul. Agarwal said this will also involve testing robots in real-world environments to identify gaps between their expected performance and actual behavior, feeding those results back into model development.
Image: SiliconANGLE/Microsoft Designer
A message from John Furrier, co-founder of SiliconANGLE:
Support our mission to keep content open and free by engaging with theCUBE community. Join theCUBE’s Alumni Trust Network, where technology leaders connect, share intelligence and create opportunities.
- 15M+ viewers of theCUBE videos, powering conversations across AI, cloud, cybersecurity and more
- 11.4k+ theCUBE alumni — Connect with more than 11,400 tech and business leaders shaping the future through a unique trusted-based network
Are you an AWS customer? Support SiliconANGLE financially by buying your AWS services from our Marketplace portal page and links: https://siliconangle.com/aws-marketplace/
About SiliconANGLE Media
Founded by tech visionaries John Furrier and Dave Vellante, SiliconANGLE Media has built a dynamic ecosystem of industry-leading digital media brands that reach 15+ million elite tech professionals. Our new proprietary theCUBE AI Video Cloud is breaking ground in audience interaction, leveraging theCUBEai.com neural network to help technology companies make data-driven decisions and stay at the forefront of industry conversations.