The Missing Data Stack for Physical AI

Explore the challenges and opportunities of physical AI through this 53-minute podcast episode featuring Nikolaus West, CEO of Rerun, who breaks down why AI that interacts with the real world requires fundamentally different approaches than traditional software. Discover how visualization, adaptability, and specialized tooling are essential for making robotics and spatial computing more practical and effective. Learn about the complete physical AI lifecycle, from data collection challenges in dynamic environments to the constraints AI faces in robotics applications. Examine why IoT hype often falls short of reality and how better data integration can provide crucial insights for robotics development. Understand the importance of data pipelines, quality considerations, and design trade-offs in robotics systems. Gain insights into open-sourcing AI tools and platform integration strategies that can accelerate physical AI development. The discussion covers the evolution from traditional robotics to modern physical AI applications, highlighting the unique infrastructure requirements and data stack considerations needed for AI systems that must operate in unpredictable real-world environments.