microagi is a Munich-based physical AI research and deployment company founded in September 2025 by five co-founders with backgrounds spanning Formula 1 engineering, academic AI research, and tech entrepreneurship. The company's core product is Atlas, a hardware- and model-agnostic platform that captures real-world demonstration data from workers on factory floors using cameras and sensor-equipped gloves, trains robot control policies in simulation on that data, and deploys the resulting models onto existing industrial robots — all without disrupting live production lines. Its proprietary data corpus, one of the largest collections of real physical-work footage in the world, is also gathered through its Shift platform, which pays over 20,000 operators across 15 countries roughly $20/hour to record physical tasks.

microagi targets industrial customers in automotive, logistics, and food processing, positioning itself as the "last mile" solution that bridges the gap between impressive robot demos and reliable, day-to-day factory deployments. The platform is model- and hardware-agnostic, meaning it works with whichever robot suits the customer's task regardless of manufacturer. Atlas now trains and serves models on NVIDIA Blackwell infrastructure on Google Cloud, backed by engineering teams from both partners.

In July 2026, the company closed a $55 million seed round — the largest seed round in German startup history — led by Hummingbird Ventures, with Northzone, LocalGlobe, Village Global, and redalpine participating. The company is headquartered in Munich, with a global robotics R&D hub next to ETH Zurich in Zurich, and additional offices in London, Istanbul, and New York. In September 2026, microagi hired Georgia Tech professor and former NVIDIA researcher Animesh Garg as Chief Research Officer, bringing seven members of his research group through the acquisition of a stealth startup, further accelerating its post-training capabilities.