NVIDIA has announced the OpenShell Reference Architecture for building, containing, observing, and enforcing safe AI agents on enterprise infrastructure and at the edge. Announced at GTC Berlin, the design builds on NVIDIA OpenShell secure runtime software and pairs NVIDIA Vera Rubin infrastructure for model inference, NVIDIA Vera CPUs for agent harnesses and tool use, and NVIDIA BlueField-4 DPUs running NVIDIA Sentry for continuous out-of-band monitoring and in-silicon policy enforcement. Jensen Huang said agents can think on Rubin, act on Vera, and be governed on BlueField, with safety and control built into the computing architecture. If an agent moves outside its software boundary, Sentry is said to quarantine and stop it in milliseconds; Sentry can also boost agent capacity by up to 20%. The architecture runs on enterprise operating systems, including Canonical, Red Hat, and Microsoft Windows, and extends to the edge via RTX Spark, DGX Spark and DGX Station.
NVIDIA has announced the OpenShell Reference Architecture for building, containing, observing, and enforcing safe AI agents on enterprise infrastructure and at the edge. Announced at GTC Berlin, the design builds on NVIDIA OpenShell secure runtime software and pairs NVIDIA Vera Rubin infrastructure for model inference, NVIDIA Vera CPUs for agent harnesses and tool use, and NVIDIA BlueField-4 DPUs running NVIDIA Sentry for continuous out-of-band monitoring and in-silicon policy enforcement. Jensen Huang said agents can think on Rubin, act on Vera, and be governed on BlueField, with safety and control built into the computing architecture. If an agent moves outside its software boundary, Sentry is said to quarantine and stop it in milliseconds; Sentry can also boost agent capacity by up to 20%. The architecture runs on enterprise operating systems, including Canonical, Red Hat, and Microsoft Windows, and extends to the edge via RTX Spark, DGX Spark and DGX Station.