Faster Observability for Agents, from Day One

When you connect an agent to Monte Carlo, the auto-created metric monitors for latency, token usage, and error rate now begin generating thresholds from the first data points. Monte Carlo starts with conservative thresholds and tightens them as more data arrives, handing off to full anomaly detection once the monitor matures.

Your agents are covered from day one — with active thresholds and alerting in place as soon as monitoring begins.