Human-on-the-loop oversight

A supervision model where humans monitor and can intervene on an autonomous agent fleet as it runs, rather than approving each individual action, keeping bounded autonomy accountable.

grounded in: trend theme 'Who supervises autonomous agents' plus the system's own watchdog/gate/guardrails oversight of its unattended self-improve loop