How an Agent Built a 3D Paris Gallery by Chaining Two Hugging Face Spaces

An AI agent can build a 3D web gallery by chaining together two separate Hugging Face Spaces. The process uses an image generation model to create pictures of monuments, which are then fed into a 3D reconstruction model to generate Gaussian splats. This workflow is enabled by a new `agents.md` file that provides a simple, documented API for AI agents to call and use the models within Spaces. This approach exemplifies a "building-block economy" for multimedia AI, where complex applications are assembled from composable, callable models without requiring deep integration work.