Google launches Gemini Design: an LLM tuned for UX teams

AI · 5 min read

Google today announced Gemini Design, a new member of the Gemini line tailored specifically for UX and product design workflows. The model is trained on annotated UI screenshots, design tokens, and interaction patterns to understand layout intent, microcopy and accessibility constraints at a higher fidelity than general-purpose LLMs.

Gemini Design ships with a REST API and SDKs that convert high-level prompts into structured design outputs: JSON layout proposals, token suggestions, and annotated image overlays. Google also released a Figma plugin that allows designers to call the model directly from the canvas and receive annotated layer suggestions and accessible color contrast checks.

Early partners reported faster iterations on landing pages and component documentation, with the model helping teams standardize tokens and generate responsive variants. Google emphasized privacy controls and an on-prem Vertex AI option for enterprise customers with sensitive product data.