The most valuable data for Physical AI might be sitting right outside your door. A staircase. A narrow corridor. A shop entrance. The uneven pavement a delivery robot could easily struggle with. These details rarely make it into the maps AI systems rely on, yet they matter when intelligence has to operate in the real world. This is the angle I find interesting about @vangrid_io Its network has reported more than 1.25 million captures, turning everyday environments into spatial data that can help machines understand the places they move through. And I keep coming back to one question: What if the next generation of AI infrastructure isn’t built only in data centres, but also by ordinary people documenting the world around them? That would make human presence part of the infrastructure powering Physical AI. We’re used to thinking of our phones as tools for consuming information. @vangrid_io is exploring a world where they can help create something machines genuinely need. That shift deserves more attention.