Speed, Judgment and Behavior: AI's Defense Mandate

Agentic AI , Artificial Intelligence & Machine Learning , Events Speed, Judgment and Behavior: AI's Defense Mandate Capitol Meridian Partners' Niloofar Razi on Innovation Sandbox, AI-Driven Offense Tom Field (SecurityEditor) • March 26, 2026     Share Post Share Credit Eligible Get Permission Niloofar Razi, operating partner, Capitol Meridian Partners Cybersecurity can no longer stop at the system boundary. Organizations must understand how humans and artificial intelligence agents behave, and intervene before attackers exploit that behavior, said Niloofar Razi, operating partner at Capitol Meridian Partners. See Also: Why Healthcare Faces Rising Risks From Shadow AI The attack surface has moved inward into agent workflows, development pipelines and the human decisions that attackers are learning to manipulate. That means organizations must look inside the build process and govern how agents act, not just monitor what they produce. AI gives defenders the speed to match that threat, moving through decision cycles in milliseconds. But speed alone is not enough. "The system can do all of that in milliseconds. Humans do it in minutes if they're really good. But the judgment piece uniquely still belongs to the human," Razi said. In this video interview with Information Security Media Group at RSAC Conference 2026, Razi also discussed: What this year's Innovation Sandbox finalists reveal about where the cybersecurity industry is headed; How AI is enabling startups to scale quickly; How education must evolve to produce AI-first thinkers equipped with analytical rigor, not just technical skills. Razi is an investor and entrepreneur with more than 30 years of experience in the technology industry, including two decades focused on cybersecurity and national security. She is the president of The Stratham Group, an alternative asset management firm. Razi is also a distinguished visiting professor at Vanderbilt University's Institute of National Security.