A hybrid tree data structure is suitable for use in scenarios involving intermingled text and user interface elements. Trees of two different types can be combined via one or more proxy nodes. For example, one type can be efficient at processing user interface elements and another can be efficient at processing text. Operations suitable for user interface elements can be efficiently performed on portions of the hybrid tree having user interface elements, and operations suitable for sequential data can be efficiently performed on portions of the hybrid tree having sequential data. The structure is thus suited for representing documents or graphical user interfaces. A hypertext document can be represented via the hybrid tree to enable more efficient searching, navigation, rendering, or editing of the document. Hybrid tree services can be provided by an operating system service.