An established interaction-design pattern in which a human acts as a supervisory checkpoint within an otherwise automated decision loop, retaining the ability to review or intervene; a superordinate design concept rather than a single countable occurrence.
Superordinate framework term; not operationalized as a per-interaction event. It names a class of control architectures (human review embedded in an automated loop); specific instances are studied through narrower constructs such as confirmation-gating or override behaviour rather than by counting the pattern itself.
Proposed measurement protocol (not yet empirically validated): Not directly measured; superordinate term.
https://andreasehstandlicenseofclarityloc.github.io/augmanitai-periodic/#topic-agentic-ai