Anton Chuvakin discusses how AI agents in security operations affect the metrics that matter to leadership. AI agents show a 4:1 cost ratio, and they reshape mean time metrics for detect, triage, contain, investigate, and close.
Chapters
That's a really interesting roadmap you're laying out there. I want to maybe take a step back and uplevel this a little bit. So for your overall SOC, what are the metrics you're reporting up to your leadership? And then how do you think that the agent and all the agentic work you're doing is going to shift those metrics?
Sure. So we have 2 different types, just like everyone else does. We have the one that the fancy drawings for senior leadership that they want to see. The return on investment obviously is like, hey, how much Produce or throughput are we getting from this, you know, AI agent today? And today we were seeing a 4-to-1 cost ratio. So that's not bad, but we could do better. And then of course we also see the meantime to detect in terms of not just developing a detection later on, but the meantime to triage, contain, investigate, close the entire ticket in response. We're measuring against that which is AI labeled and which ones aren't.