humyn-labs-co-founders_164602

Humyn Labs Raises US$47 Million to Supply Real-World Training Data for Robotics and AI

Indian startup Humyn Labs has raised US$47 million in Series A funding as demand grows for real-world data to train artificial intelligence systems capable of performing physical tasks. The Bengaluru-based company collects and processes footage of people performing everyday activities, helping robotics companies develop AI models that can better understand and interact with the physical world.

The funding consists of US$26 million in equity led by Accel, alongside participation from Prosus Ventures, BeeGlobal, Bajaj Finance’s Finserv Intelligence, Bajaj Alts AI Opportunities Fund, Dallas Venture Capital, Lighthouse Canton and Jump Trading. An additional US$21 million in venture debt was led by Lighthouse Canton, with participation from Trifecta Capital.

Helping AI Learn from Human Activities

As robotics companies develop machines capable of performing tasks in homes, workplaces and industrial environments, access to high-quality training data has become increasingly important. Unlike AI systems trained primarily on text or digital information, physical AI requires data that captures how people move, handle objects and complete everyday activities.

Humyn Labs addresses this challenge by recording human activities using its own hardware equipped with two cameras. The company then processes the footage through its software, which identifies movements, interactions with objects and other details that can be used to train AI models.

Its data collection operations span more than 20 countries across Latin America and Asia, with Brazil, India and South Korea serving as its primary hubs. The company works with local partners to organise collection activities and build a diverse dataset reflecting how people perform tasks in different environments.

Founded by Experienced Technology Entrepreneurs

Humyn Labs was founded in February 2026 by Manish Agarwal, former CEO of gaming company Nazara Technologies, and Ishank Gupta, who previously held an executive position at AB InBev.

The founders previously established Kratos Studios, now known as KGeN, which operates a network of contributors across emerging markets. That existing network has helped Humyn Labs establish data collection operations and reach workers in different countries.

The startup says its approach focuses on recording people performing their regular work and household activities rather than staging artificial demonstrations. This allows the company to capture more natural movements and interactions that could help AI models better understand real-world situations.

Growing Revenue and Customer Demand

Humyn Labs says it currently serves 12 paying customers, including AI research laboratories, robotics companies and developers of world models. Around 90% of its revenue comes from the United States, reflecting strong demand from companies developing advanced AI systems.

According to Agarwal, the startup generated US$10 million in revenue between April and October 2026 and has secured another US$20 million in signed contracts. A further US$20 million in potential deals remains under negotiation.

The company currently employs 51 full-time staff, most of whom are based in Bengaluru, while also working with local partners across its international data collection network.

Funding Supports Expansion and Operations

The US$21 million debt component of the funding round will help Humyn Labs manage the costs of expanding its data collection operations. The company pays workers and collection partners frequently, while its enterprise customers generally take 30 to 60 days to settle invoices.

The additional capital provides working cash to support this difference in payment timing, allowing the company to expand its operations without relying entirely on additional equity funding.

With the new investment, Humyn Labs is positioning itself to become a major supplier of real-world training data as robotics and physical AI companies compete to build machines capable of operating beyond controlled environments.

As demand for more diverse and realistic training datasets increases, the company is betting that its international network of workers, collection hardware and data processing software will give it an advantage in an increasingly competitive market.

Tags: No tags

Add a Comment

Your email address will not be published. Required fields are marked *