Art Matched to Your Space
Upload a room, receive a considered selection—each recommendation explained in context.
The feature recommends curated artworks based on a photo of the customer’s own room. Every recommendation is accompanied by a written explanation of why that particular piece belongs in that space.
At the core sits a multimodal architecture: a vision encoder extracts image embeddings from the room photograph, which are passed—alongside textual descriptions of the artwork—into a language model for joint reasoning.
Multiple models work in concert. The real craft lies in orchestrating the handoffs, shaping the inputs and keeping the outputs coherent and genuinely useful rather than merely plausible-sounding.