Search Object Types
Almost every action that touches the screen has a Search object parameter — the description of what the robot should click, read or wait for. Search objects fall into three categories:
- Recognized UI elements – detected directly by the AI model from the screen image.
- Positional search – locate an element relative to another reference element.
- Special objects – define working areas, masks or relative coordinates.
Choosing the right one is the difference between an automation that survives a UI change and one that breaks on the next release of the target application.
How to pick
- If the element has stable, unique text or is a recognizable control, use a recognized element (Text, Button, Input, …).
- If its own text changes every run but a nearby label does not, anchor it with a positional search (On the right of the label, Below the column header).
- Fall back to Area coordinates only when nothing else works — coordinates break as soon as the window moves or resizes.
Recognized UI Elements
The detection model processes a screenshot and returns the elements that are
Positional Search
A positional search finds a target by its position relative to another element. Use
Special Objects
These objects do not represent interactive controls. They describe regions of the