Google launches Muse: a prototype generation model turning images and text into interactive prototypes

AI · 6 min read

Muse combines visual encoders and a layout‑aware decoder to output structured prototypes, including navigation flows and annotated interactions. The model detects semantic elements like forms, tables, and navigation bars, and can enrich them with accessibility labels and keyboard flows.

Google released a toolkit for exporting Muse outputs into Figma, Adobe XD, and HTML/CSS skeletons, and included a small dataset demonstrating Muse's capabilities on cross‑platform flows. The company also shared research notes about handling ambiguity in low‑fidelity sketches.

Muse is available to Google Cloud customers with a free trial tier for designers and startups. Google warns that while Muse speeds ideation, human review remains essential to ensure usability and legal compliance for generated content.