Insights

Field notes on agent enforcement.

How Zero Trust principles for AI agents translate into controls that actually run on the machine where the agent does — at the OS layer, not in the prompt.

PlatformJun 7, 2026

Agent security is moving into silicon. You still need a portable layer above it.

NVIDIA’s RTX Spark and Microsoft’s Windows agent primitives push enforcement toward hardware — validation that the boundary belongs below the agent. Here’s why a cross-platform, agent-agnostic layer is still the missing piece.

Read →
Zero TrustJun 7, 2026

Least agency needs an enforcement point. That point is the OS.

Anthropic’s Zero Trust for AI Agents pairs with OWASP’s Least Agency principle — governing what an agent can do, not just what it can access. Why that has to be enforced at the OS, and how Naevik does it.

Read →