Maxinsights raises $30M to turn everyday chores into robot training data

Maxinsights, a Santa Clara startup, raised $30M in seed funding led by Peak XV, with Tola Capital, South Park Commons, INCE Capital, and Alumni Ventures also participating. The company pays more than 5,000 contractors across six countries to wear head, wrist, and chest cameras while doing everyday tasks like cooking, cleaning, and loading dishwashers, then turns that footage into training data for robots.

Founder and CEO Yuguang Yuan told Forbes the goal is to build an archive of real, first-person physical work the way the internet gave language models an archive of text to learn from. The company says it has captured more than 2 million hours of footage, with 1.5 million hours ready for use by models, and has over 30 customers including frontier AI labs. It says it supplied all 200,000 hours of first-person data behind Genesis AI's GENE-26.5 model and calls itself the largest data provider behind Dyna Robotics' Dyna-2 model.

Alongside the funding, Maxinsights launched MaxLoop, software that checks incoming footage quality, and MaxGlove, a sensor glove built with Genesis AI that records grip force and contact data in sync with video. The company says its quality checks cut review costs by roughly 80% and shrank its human review team by 90%, while processing about 450,000 hours of footage monthly.

Why it matters

Physical AI investment is surging, and training data from real human movement, not staged demos, is becoming a key bottleneck and business.

Source: The Next Web