A check that makes sure an AI Agent really finished its task.
Your robot says it washed the dishes. You still check for spaghetti on the forks.
This check looks for proof of success. It helps with orders and expense claims.
Agent
It checks if the Agent's actions got the expected result.
Agent observability
Logs and traces give proof that the task is done.
Human-in-the-loop
A person checks the task when the system is unsure.