Imagine an AI agent needing a fresh 3D scan of a warehouse it has never seen. It can process the request, but someone still needs to go there and capture the actual environment. That’s the gap @vangrid_io is working to address, and its latest move on @arc is worth watching. Vangrid is building a system that lets AI agents pay human contributors to collect real world spatial data on demand. Here’s how the process works. An agent identifies a location it needs data from and funds a bounty in USDC. A contributor visits the site, captures the environment, and submits the data for review. Once accepted, the requester receives the resulting 3D reconstruction. Vangrid has also announced its first bounty payment on Arc, marking an early milestone for this approach. I find the connection between AI agents and human contributors particularly interesting. Agents can handle digital tasks, process information and make decisions, but getting accurate data about a specific physical location can still require someone on the ground. That data could support robotics, mapping, digital twins and Physical AI development, where understanding real environments matters. Of course, the model still needs to prove itself at scale. Data quality, verification, privacy and consistent demand will all play a role in its success. Still, the idea opens up an interesting possibility: AI agents could eventually commission real world services as easily as they request information from an API. With @vangrid_io people with smartphones could help supply the physical-world data these systems need. The agent economy may eventually extend beyond software, creating demand for tasks that only someone on the ground can complete.