Vision LLM analyzes notification snapshots and generates contextual titles and descriptions. Preserves known names from face recognition and describes vehicles with make/model/color when visible. When ...