Claru AI

Claru is a data company that builds the training material powering "physical AI" — the technology behind robots, self-driving vehicles, and systems that understand the real world. Instead of giving AI models generic internet data, Claru captures real, licensed video of everyday life: people cooking in kitchens, working on factory floors, driving city streets, and moving through warehouses. Each video clip is carefully enriched with extra information like depth maps, body-pose estimates, and labels that explain what's actually happening in the scene. This helps AI systems learn how the physical world really works. With over 10,000 trained collectors spread across 100+ cities on six continents, Claru delivers the large, high-quality datasets that advanced AI labs need to train robots and world models — built to fit each customer's own format and pipeline.

Key Features

• Real-world, licensed video captured by 10,000+ trained collectors across 100+ cities and six continents — not synthetic or scraped data.
• Every clip is enriched with depth maps, pose estimation, segmentation masks, and structured metadata.
• Expert human labelers add what machines miss, including intent, context, and edge cases.
• Covers diverse environments: egocentric video, game worlds, driving, cinematic, manufacturing, cooking, warehouse, and human activity.

Use Cases

• Training robots and embodied AI to understand and act in real-world environments.
• Building "world models" and video-generation systems that need realistic physical-scene data.
• Improving autonomous driving and vision-language AI with large-scale, human-verified footage.