We sat down with Harsh Verma of Palo Alto Networks to discuss the future of AI and security. Harsh bridges the gap between research models and production systems, detailing why "intent"—not identity—is critical for securing autonomous agents. As AI automates coding, he argues the engineering role is shifting from implementation to judgment and systems design. We explore why the most dangerous AI failures are "authorized" ones and how engineers build durability by owning real-world outcomes.