Global open-source solutions company Red Hat said on May 20 it is introducing new functions for Red Hat AI Factory with NVIDIA, a software platform it co-developed with Nvidia.

The update strengthens security features for continuously operating agents. It also provides an immediately deployable solution to help companies move beyond experiments and shift to production environments.

Red Hat said it supports an integrated and scalable foundation for production-grade agentic AI by combining the integrated AI platform capabilities of Red Hat AI and Nvidia AI Enterprise open models, advanced tools and frameworks.

OpenShell is an open-source project established by Nvidia. It provides a sandboxed runtime for autonomous AI agents. It is designed to help companies implement policy control, safe tool use and an auditable operating environment, and it manages how agents run, access permissions and inference processing paths. Companies can use OpenShell to control and monitor autonomous agent operations through a single integrated policy layer.

Red Hat said it is conducting joint engineering to integrate OpenShell into its full-stack AI platform to maintain oversight and policy compliance at the infrastructure level. Red Hat AI Factory with NVIDIA supports extending trust boundaries to the hardware level through confidential computing, beyond software-defined governance. Red Hat stressed it is designed to prevent runtime-stage agent compromise even if other agents are breached.

Chris Wright (크리스 라이트), Red Hat CTO and senior vice president for global engineering, said, "To turn enterprise AI experiments into an industrial engine, a sovereign and consistent foundation across the hybrid cloud is needed." He added, "Through strategic joint engineering with Nvidia, Red Hat provides the architectural control capabilities and open-source innovation companies need to confidently scale agentic AI."