Agentic Red Teaming: Closing the Gap Between Engagements

Attackers are using AI to reason through identity relationships, privilege chains, and misconfigurations at machine speed—the fastest exfiltration ReliaQuest observed in 2025 took 6 minutes. Defenders need AI finding those same paths first.

Pentests, red teams, and BAS run on a cadence and produce outputs disconnected from detection and response workflows. The gap between engagements is the exposure.

In this session, ReliaQuest will break down how to safely use agentic AI to continuously red team your own environment, and how findings feed directly into detection, hunting, and response rather than a static report.

What you'll learn:

• How to use frontier and open source models to continuously map attack paths across your identities, assets, permissions, vulnerabilities, and controls

• What guardrails to put in place before testing your environment with AI: rules of engagement, scope, human oversight, and controls that prevent operational risk

• How to validate attack paths and feed findings into detection engineering, hunting, remediation, and response