Helm.ai

Helm.ai develops AI-first software and simulation platforms for autonomous driving at Levels 2 through 4. Founded in 2016 and headquartered in Redwood City, California, the company builds scalable foundation models for automakers and partners with major global OEMs. Its technical work spans unsupervised learning, deep learning, applied mathematics, computer vision and simulation.

At the centre of its approach is Deep Teaching, a proprietary unsupervised learning method that combines real-world data, deep learning and applied mathematics to train adaptable, high-performance foundation models at scale. The method is designed to reduce dependence on HD maps, lidar, fleet data collection and human annotation.

The product line splits between on-vehicle deployment and offline foundation models for training and validation. Helm.ai Vision provides perception software for on-vehicle use, while Helm.ai Driver covers full-stack autonomy. GenSim-2, VidGen-2 and WorldGen-1 serve as offline foundation models used in training and validation workflows.

Helm.ai has raised over $102 million and was named to the 2025 Deloitte Technology Fast 500. The company operates in the automotive, autonomous driving and robotics sectors.