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

A coding agent built a 3D website showcasing Paris monuments as 3D Gaussian splats by directly calling and chaining two Hugging Face Spaces without touching traditional image generators or 3D reconstruction tools. According to the source, Mitchell Hashimoto recently described a shift called the building block economy, where AI excels at assembling small, well-documented components rather than building monoliths from scratch. While this thesis has traditionally applied to code libraries, the same integration forces are now impacting multimedia AI. Hugging Face Spaces have adapted to this shift by providing a plain-text agents.md file on every Gradio Space. This file supplies agents with schema URLs, call and poll templates, file upload instructions, and authentication hints, enabling an agent to drive a Space end to end without hardcoded integrations or client libraries. The pipeline utilized chaining, passing the output of an image-generating Space into a 3D reconstruction Space. Beyond asset generation, the agent handled glue work such as flipping outputs upright, auto-framing monuments, compressing files to .ksplat for faster loading, building a Three.js viewer, and deploying the application as a static Space. Human inputs were limited to taste-level feedback. Once established, the pipeline allowed new galleries for Japan and Egypt to be generated using a single sentence of prompt modification per country, highlighting how the marginal cost of creating multimedia applications falls toward the cost of describing them.