Building Defense for the Agentic Era: Kevin Mandia

On The a16z Show, a16z General Partner David George speaks with Armadin founder and CEO Kevin Mandia about the implications of machine-speed cyberattacks. Mandia’s core thesis is that AI changes the economics and operating pace of offensive security so fundamentally that defense will ultimately need to become autonomous. After 30 years in security and building Mandiant, he says AI prompted him to return to building a cybersecurity company.

The argument centers on attackers’ ability to probe thousands of paths simultaneously. That parallelism creates a challenge for defenses that depend on humans to detect threats and coordinate responses. Armadin’s approach begins with continuously attacking customers’ systems using AI to identify exploitable vulnerabilities before adversaries find them. Mandia reports that Armadin has found more than 90 zero-days in production environments this year, providing a concrete example of the discoveries behind its approach.

The immediate context is a potential mismatch between the speed of AI-enabled attacks and the speed of human-led security operations. Mandia argues that humans cannot remain in the detect-and-respond loop, and describes Armadin as building toward autonomous defenses capable of responding in real time. The distinction between its continuous offensive testing and its ambition for autonomous response matters: the notes describe a direction of development, not a completed transition to fully autonomous defense.

The discussion also considers how the security stack could change over the next few years. The notes do not specify the controls, implementation requirements or measures of effectiveness that would establish when autonomous response is ready for broad deployment. They also do not detail the reported zero-days or provide independent validation, leaving the scale of discovery clearer than its operational consequences.