Cisco AI Defense now supports a hybrid, infrastructure-agnostic deployment model for AI security, running as a pure software layer across AWS, Azure, GCP, and on-premises environments using NVIDIA accelerated computing. The platform provides three core capabilities: supply chain security (scanning models, datasets, and MCP servers before deployment), runtime guardrails (enforcing security, privacy, and safety policies on every request), and agentic security (protecting multi-agent communication and tool invocations via NVIDIA OpenShell). Deployment is Kubernetes-native with validated support for EKS, AKS, GKE, and Red Hat OpenShift, allowing enterprises to maintain consistent security policies regardless of infrastructure.
Table of contents
Security That Follows Your InfrastructureConsistent AI Security Across Every DeploymentAny Cloud, Any NVIDIA AI InfrastructureWhere to Start38 Impressions