Concept
Closed-loop completion
Closed-loop completion means a task is not marked done when the model emits output; the system verifies that output against the goal or constraints and feeds the result back until it converges. The name comes from control theory: an open-loop controller sends a command and…
The rest of “Closed-loop completion” is a premium feature: every concept in the library gets a precise, practitioner-focused write-up like this one, cross-linked straight from the paper summaries that use it.
Log in to unlock→