This SimSpace Success Session demonstrates how to train and test AI cybersecurity agents before they touch production, led by Jason Rivera, SimSpace’s solution engineering lead, formerly US Army, NSA and CrowdStrike. It opens with the threat case for agentic defense: breakout time falling from 18 minutes in 2018 to 27 seconds last year, and adversaries using AI for lures, fake personas and post-exploitation scripting. The rest runs live inside a cyber range, where four agents work against each other. An autonomous pen-test agent exploits an internet-facing web server, a threat-hunting agent queries Splunk from plain-language intent, an IT agent detects the downed server and restores it, and Dropzone AI triages the resulting CrowdStrike alerts. Watch the session for the ground-truth comparison showing which alert the triage agent missed, and why.
