HOT RADAR
industryHEAT 13.8°

AI CURATED HOT STORY · 11/18/2026

Humyn Labs Raises $47M to Supply Real-World Factory Data for US Robots

2 reports archived2 independent sourcesupdated 11/19/2026, 01:28:01
Synthesis & Latest Updates
2 Sources Cross-Validated

Indian data-collection startup Humyn Labs has raised $47 million in Series A funding, comprising $26 million in equity led by Accel and $21 million in venture debt led by Lighthouse Canton. The capital will be used to scale real-world physical manipulation data collection for US AI and robotics companies, alleviating the data bottleneck in embodied AI training.

Humyn Labs' core approach involves workers wearing custom-built collection glasses to record first-person operation videos in real factory floors, replacing clean-lab footage and avoiding distribution gaps with production lines. The data is cleaned and decomposed into atomic action primitives such as reach, grasp, lift, and place, accompanied by a self-built benchmark, forming a closed loop of collection, cleaning, and evaluation. The goal is to enable policies to learn the same skills across scenarios. The company has 51 full-time employees, mainly in Bangalore, and relies on local partner networks in multiple countries.

The funding highlights the urgent need for high-quality real-world data in embodied AI. As US robotics companies accelerate productization, Humyn Labs' scalable data collection capability could become critical infrastructure, driving robots from labs to real production lines.

LATEST/Humyn Labs has closed a $47 million Series A round led by Accel for equity and Lighthouse Canton for venture debt. The funds will scale real-world factory data collection for US robotics companies, using custom glasses and atomic action primitives to build a closed-loop pipeline.

Hourly heat curve

31 fully observed hours
Now
13.8
Peak
23.602:00
24h change
+30%

The line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.

Coverage timeline

Total 2 reports · Latest first
  1. Rohan Paul (@rohanpaul_ai)T2·58 pts
    • Self-built capture glasses record egocentric data on real factory floors, avoiding the lab-to-production distribution gap
    • Data is filtered before labeling and decomposed into atomic primitives like reach, grasp, lift and place so policies reuse skills across scenes
    • An in-house research team builds benchmarks on top of the data, closing the loop from capture to cleaning to evaluation