We treat artificial intelligence as a catalyst for design discovery more than a post-production accessory. During the earliest material research phase, custom diffusion models generate high-resolution studies of ceramic glazes—testing how color, porosity, and surface texture respond to Stuttgart’s humidity and light angles across the year. These AI-driven images function as analytical sketches. The visuals also power our narrative workflow. We curate the most evocative frames, annotate them with environmental data, and weave them into storyboard sheets that describe a resident’s journey from street to skycourt. Each storyboard acts as a brief for the parametric model: balcony depths, façade perforations, and communal terraces are dimensioned directly from the atmospheres depicted in the AI imagery. This loop—generate, narrate, parametrise—turns abstract research into spatial sequences that can be costed, constructed, and lived in.