collecting real-world data at scale is not just a technical problem. it is also a coordination problem. @vangrid_io approaches this by connecting two groups with different needs: organisations that require specific places captured and contributors who can collect that data using ordinary phones. a buyer can post a bounty, define the location and requirements, and fund the request with usdc. a contributor can then complete the capture and submit it for review. payment follows when the requester accepts the work. this demand-driven model matters because it gives contributors a clear task instead of asking them to record random places and hoping someone eventually needs the footage. it also gives buyers a way to request the exact environments they need, from warehouse interiors to market streets and loading areas. for physical ai, the challenge is not simply gathering more footage. it is building a repeatable system where real-world data can be requested, collected, verified and purchased. that is the economic loop vangrid is working to establish.