NVIDIA and SAP Partner to Launch Secure AI Agents for Enterprise
The landscape of corporate automation is shifting from simple assistants to autonomous agents capable of managing complex workflows. At the annual SAP Sapphire conference, NVIDIA and SAP announced a significant expansion of their partnership aimed at bringing trust, security, and governance to specialized AI agents. This collaboration integrates NVIDIA OpenShell, an open-source runtime designed for the secure deployment of autonomous AI, directly into the SAP Business AI Platform. As global enterprises look to integrate artificial intelligence into their core systems of record, this move provides the necessary "guardrails" for AI to operate within sensitive business environments like finance, procurement, and supply chain management.
Bridging the Trust Gap in Autonomous AI
For years, AI has functioned primarily as a co-pilot or assistant, requiring human intervention at every critical junction. However, as the industry moves toward agentic AI-systems that can touch systems of record and cross application boundaries-the "trust equation" has fundamentally changed. An autonomous agent that can execute a procurement order or adjust a supply chain schedule without constant human oversight needs strict policy enforcement and an audit trail to be viable for production.
NVIDIA and SAP are addressing this by co-designing OpenShell to provide isolated execution environments. This technology enforces security at the filesystem and network layers, ensuring that an AI agent cannot exceed its intended permissions. By embedding this into the SAP Business AI Platform, OpenShell acts as the runtime security layer for all SAP AI agents, including those custom-built via Joule Studio.
The Five-Layer Cake of Artificial Intelligence
NVIDIA CEO Jensen Huang has frequently described the AI ecosystem as a "five-layer cake" consisting of energy, chips, infrastructure, models, and applications. While NVIDIA dominates the lower layers with its hardware and infrastructure, SAP sits at the very top-the application layer. This is where AI creates tangible economic value by driving productivity for knowledge workers.
The integration of OpenShell allows these high-level applications to function safely. While Joule Studio asks, "Should this action happen at all?" based on business logic, OpenShell asks, "Can this agent action safely execute?" from a technical security perspective. Together, they close a security gap that application-layer software alone cannot fill, providing a comprehensive defense-in-depth strategy for the modern enterprise.
Accelerating the Path to Production
One of the primary hurdles for developers today is building the security "scaffolding" required to move an AI project from a pilot phase to a live production environment. To streamline this, NVIDIA is making NemoClaw-a reference blueprint for developing autonomous agents-available directly within SAP’s Joule Studio. This provides development teams with a structured route to deployment, allowing them to focus on the logic of their specialized agents rather than the underlying security infrastructure.
NVIDIA is not just a provider in this relationship; they are also a long-standing SAP customer, running their own global finance and logistics operations on SAP software. This shared context has allowed both companies to co-develop features focusing on runtime hardening, enterprise identity integration, and auditing hooks that are grounded in real-world business requirements. As AI agents become a staple of modern development, these tools ensure that they stay within the boundaries required by the world’s largest organizations.
