10 Crucial Insights into NVIDIA and SAP's Trust-Driven AI Agent Collaboration
At SAP Sapphire 2025, NVIDIA and SAP unveiled a groundbreaking partnership to bring trust, security, and governance to specialized AI agents operating within enterprise systems. These agents—handling finance, procurement, supply chain, and manufacturing—require robust boundaries before they can autonomously execute workflows. Here are ten essential points to understand this strategic collaboration.
1. The Core Announcment: NVIDIA and SAP Expand Alliance
NVIDIA founder Jensen Huang joined SAP CEO Christian Klein's keynote via video at SAP Sapphire, revealing an expanded collaboration. The goal: enable enterprises to deploy specialized AI agents with built-in security and governance. SAP integrates NVIDIA's OpenShell—an open-source runtime—into the SAP Business AI Platform. This move addresses the critical need for trust as agents become autonomous.
2. What Are Specialized AI Agents?
Specialized AI agents go beyond simple assistants. They perform tasks within enterprise systems—like processing invoices, optimizing supply chains, or managing procurement—by interacting with databases, applications, and workflows. Unlike chatbots, these agents can execute actions autonomously, making security and policy enforcement paramount. They operate where business decisions are made daily.
3. The Trust Challenge: From Assistants to Autonomous Agents
As AI evolves from recommendation engines to autonomous agents, the trust equation shifts. An agent that can alter system records or cross application boundaries without human review needs strict safeguards. Enterprises require boundaries, policy enforcement, and a complete audit trail before agents can run in production. NVIDIA and SAP are co-developing solutions to meet this challenge head-on.
4. Introducing OpenShell: The Security Backbone
OpenShell is an open-source runtime designed for securely developing and deploying autonomous AI agents. It provides isolated execution environments, policy enforcement at filesystem and network layers, and infrastructure-level containment. If agent logic fails, OpenShell prevents damage by containing the breach. This runtime is now embedded as the security layer within SAP's AI platform.
5. How OpenShell Integrates with SAP Business AI Platform
Within the SAP Business AI Platform, OpenShell acts as the runtime security layer for all AI agents—including custom ones built in Joule Studio, SAP's agent-building environment. This means every agent, whether pre-built or tailor-made, benefits from the same governance and isolation controls. The integration ensures consistent security across diverse enterprise use cases.
6. The Five-Layer AI Cake: Where SAP Fits
Jensen Huang describes AI as a five-layer stack: energy, chips, infrastructure, models, and applications. Applications sit at the top, where AI creates economic value. SAP dominates the application layer with finance, procurement, supply chain, and manufacturing systems. This position makes SAP a key catalyst for agentic AI adoption, as agents must operate within existing process controls.
7. Enterprise Requirements for Agentic AI
Business agents need to understand roles, processes, permissions, and data boundaries. They require an execution environment that limits what an agent sees and can do. SAP's deep integration with enterprise operations provides the context needed—identity management, process rules, and data governance—to ensure agents act within safe parameters.
8. NVIDIA's Insider Perspective as an SAP Customer
NVIDIA itself runs finance, supply chain, and logistics on SAP systems. This real-world experience gives both companies shared insights into what enterprise-grade governance requires. NVIDIA engineers bring first-hand knowledge of challenges, helping shape OpenShell contributions that focus on runtime hardening, policy modeling, and identity integration.
9. Co-Development: OpenSource Contributions for Production Readiness
SAP engineers are working alongside NVIDIA to further develop OpenShell's codebase. Their contributions focus on making agentic AI production-ready: runtime hardening, policy modeling, enterprise identity integration, and auditing hooks. This collaborative effort ensures the open-source project meets the stringent requirements of large organizations deploying AI at scale.
10. The Road Ahead: Trustworthy Agentic AI at Scale
The partnership sets a new standard for deploying AI agents in critical enterprise environments. By combining SAP's application-layer dominance with NVIDIA's security runtime, the two companies provide a foundation for agents that are both powerful and trustworthy. As enterprises adopt autonomous agents, this collaboration offers a blueprint for balancing innovation with control.
In summary, NVIDIA and SAP are addressing the fundamental trust deficit that holds back widespread agentic AI adoption. With OpenShell as the security layer and SAP's platforms as the operational backbone, enterprises can confidently deploy specialized agents that drive efficiency without compromising governance. This is a pivotal step toward a future where AI agents operate seamlessly within business workflows, secured by design.