The AI agents that audit the network non-stop and generate reports. They run on a schedule and react to config-change events — "continuous" = frequent sweeps + event triggers, not magic real-time.
Currently L3 · conditional: agents plan & execute audits autonomously, but hold risky actions (remediation, high-severity) for human approval. Low-risk fixes run at L4.