Autonomy Slider
An autonomy slider lets users set how independently an agent may act—from suggest-only to draft to execute—often per task or surface.
It implements “human-in-the-loop” as a continuous control, not a single approval modal.
Share
What it means
UI for autonomy levels such as Suggest, Co-pilot (draft for review), and Autopilot (act with guardrails), with clear defaults per risk.
Why designers should care
Different tasks need different oversight. A slider encodes policy in the interface and reduces both over-trust and alert fatigue.
Example
Email agent defaults to draft-only for external sends; user slides to auto-send for internal status updates after three approved drafts.
Common mistakes
- • Global on/off for autonomy with no per-action granularity.
- • Autopilot with no activity feed or undo.
- • Hiding the current autonomy level until something fails.