Maxinsights raised $30M in seed funding led by PeakXV, with participation from Tola Capital, South Park Commons, INCE Capital, and Alumni Ventures. The platform pays contractors to record unscripted everyday activities — cooking, cleaning, car washing — to generate training data for physical AI models. The company has collected 2M hours of real-world activity, with 1.5M hours model-ready.
Unlike staged data collection, Maxinsights captures variation and edge cases that robots encounter in real deployment environments. The company built proprietary quality control infrastructure including MaxLoop, an AI system that scores incoming footage and reduced review costs by 80%, and MaxGlove, a sensor glove co-developed with Genesis AI to capture tactile data like grip force and object weight that video alone cannot provide.
The founding team includes robotics and autonomous driving veterans from Waymo, Google, Microsoft, and Meta. CEO Yuguang Yuan previously co-founded Mega Gear Robotics (acquired), while CTO Kun Wang was an early Waymo engineer and later co-founded autonomous driving startup QCraft. The company supplied all 200K hours of egocentric data used to train Genesis AI's GENE-26.5 model and is the largest data provider for Dyna Robotics' Dyna-2 World-Action Model.
Maxinsights targets 1M hours of force data collection in 2027 using MaxGlove and its existing contractor network across six countries. The platform processes 450K hours of footage monthly, a throughput enabled by treating quality control as an engineering problem rather than a manual staffing function — a structural advantage competitors struggle to replicate after the fact.



