Google’s A2UI Team introduced A2UI Express as the new default inference format for generative UI. A2UI Agent SDKs compile model output into validated A2UI JSON messages, and developers can substitute other DSLs with their own compilers. The release also demonstrates TypeSafe’s Jev decision model selecting among predefined layouts. Google reports that Express can reduce latency by up to threefold and use up to threefold fewer tokens than Direct JSON, while Jev-based layout decisions can reduce latency by up to 80-fold.