Natasha Gubernov

Artist Companion uses Meta’s Muse models as an on-request art coach for physical artwork. The browser rectifies a photo of the piece from four marked corners, samples its palette, and sends that measured context with two small images to a Node server that calls the Meta Model API. Muse Spark (vision and text) returns structured JSON for critique, guide suggestions, art direction, balance marks, a voice conversation and a one-shot comprehensive overview; Muse Image can repaint the work as a study in a suggested direction. Every response is validated server-side against the app’s real guide catalog and sampled colors, so the model can recommend but never invent geometry or palette entries. The app itself was built with Muse Code as the development assistant.