NVIDIA and SAP Bring Trust to Specialized Agents

NVIDIA Blog Products

Summary

NVIDIA and SAP have expanded their collaboration to integrate NVIDIA OpenShell into SAP Business AI Platform, enabling secure and governed deployment of specialized AI agents in enterprise systems.

<div id="bsf_rt_marker"></div><p><span style="font-weight: 400;">From finance and procurement to supply chain and manufacturing, specialized AI agents are moving into the enterprise systems where business decisions are made, data is accessed and workflows run at scale. </span></p> <p><span style="font-weight: 400;">Announced today at SAP Sapphire — where NVIDIA founder and CEO Jensen Huang joined SAP CEO Christian Klein’s keynote by video — SAP and NVIDIA’s expanded collaboration helps enterprises run specialized agents with security and governance controls. </span></p> <p><span style="font-weight: 400;">SAP embeds </span><a target="_blank" href="https://build.nvidia.com/openshell"><span style="font-weight: 400;">NVIDIA OpenShell</span></a><span style="font-weight: 400;"> — an open source runtime for securely developing and deploying autonomous AI agents — into SAP Business AI Platform. In addition, SAP engineers are codesigning OpenShell alongside NVIDIA, contributing back to the open source project.</span></p> <p><span style="font-weight: 400;">OpenShell provides isolated execution environments, policy enforcement at the filesystem and network layers, and infrastructure-level containment that guards against damage when agent logic fails. </span></p> <p><span style="font-weight: 400;">Within SAP Business AI Platform, OpenShell is the runtime security layer for all SAP AI agents, including custom agents built in Joule Studio — SAP’s environment for building and managing end-to-end enterprise agents.</span></p> <p><span style="font-weight: 400;">For enterprises, the shift from AI assistants to autonomous agents changes the trust equation. An agent that can touch systems of record, cross application boundaries and operate without review at every step needs boundaries, policy enforcement and an audit trail before it can become part of production work. That’s what SAP and NVIDIA are collaborating to address.</span></p> <h2><b>Why the Application Layer Matters</b></h2> <p><span style="font-weight: 400;">Huang has described </span><a href="https://blogs.nvidia.com/blog/ai-5-layer-cake/"><span style="font-weight: 400;">AI as a five-layer cake</span></a><span style="font-weight: 400;">: energy, chips, infrastructure, models and applications. </span></p> <p><span style="font-weight: 400;">Applications sit at the top, where AI creates economic value and drives productivity for knowledge workers. As a global leader in enterprise applications and business AI, SAP is an important catalyst in the application layer, running finance, procurement, supply chain and manufacturing workflows where agents must operate within policy, identity and process controls.</span></p> <p><span style="font-weight: 400;">That makes SAP’s position at the core of enterprise operations a key driver for enterprise adoption of agentic AI. </span></p> <p><span style="font-weight: 400;">Business agents need to understand roles, processes, permissions and data boundaries. They also need an execution environment that limits what an agent sees, what it can do and where inference runs.</span></p> <h2><b>Codeveloping the Agentic AI Foundation</b></h2> <p><span style="font-weight: 400;">NVIDIA brings its own perspective as a longstanding SAP customer — running finance, supply chain and logistics on SAP, giving both companies shared context for what enterprise-grade governance requires in practice.</span></p> <p><span style="font-weight: 400;">SAP engineers are working alongside NVIDIA’s to further develop OpenShell’s open source codebase, focusing contributions on what enterprises need to run agentic AI in production: runtime hardening, policy modeling, enterprise identity integration, and auditing and governance hooks.</span></p> <p><span style="font-weight: 400;">SAP and NVIDIA technologies work in unison to address important enterprise requirements for deploying trustworthy agents. </span></p> <ul> <li style="font-weight: 400;" aria-level="1"><span style="font-weight: 400;">NVIDIA OpenShell asks: Can this agent action safely execute? </span></li> <li style="font-weight: 400;" aria-level="1"><span style="font-weight: 400;">Joule Studio runtime — the enterprise control layer within SAP Business AI Platform — asks: Should this action happen at all? </span></li> </ul> <p><span style="font-weight: 400;">Together, they close a gap that application-layer security alone cannot.</span></p> <h2><b>A Faster Start for Agent Builders</b></h2> <p><span style="font-weight: 400;">SAP customers building custom agents will have a faster path to production. </span><a target="_blank" href="https://www.nvidia.com/en-us/ai/nemoclaw/"><span style="font-weight: 400;">NVIDIA NemoClaw</span></a><span style="font-weight: 400;"> — a reference blueprint for developing and deploying autonomous agents — will be available directly in Joule Studio. This means development teams get a structured route from initial build to trusted production deployment, without having to engineer security scaffolding from scratch.</span></p> <p><span style="font-weight: 400;">AI agents will create value only when enterprises can trust them with their data. </span></p> <p><span style="font-weight: 400;">For many organizations, that data is in SAP — which houses the systems of record across finance, procurement and supply chain that run their business. </span></p> <p><span style="font-weight: 400;">Together, SAP and NVIDIA are making agents ready to act, while staying within the boundaries enterprises require.</span></p> <p><i><span style="font-weight: 400;">Learn more about </span></i><a target="_blank" href="https://build.nvidia.com/openshell"><i><span style="font-weight: 400;">NVIDIA OpenShell</span></i></a><i><span style="font-weight: 400;"> and </span></i><a target="_blank" href="https://www.nvidia.com/en-us/ai/nemoclaw/"><i><span style="font-weight: 400;">NemoClaw</span></i></a><i><span style="font-weight: 400;">.</span></i></p> <p><i><span style="font-weight: 400;">See </span></i><a target="_blank" href="https://www.nvidia.com/en-eu/about-nvidia/terms-of-service/"><i><span style="font-weight: 400;">notice</span></i></a><i><span style="font-weight: 400;"> regarding software product information. </span></i></p>
Original Article
View Cached Full Text

Cached at: 05/13/26, 12:20 AM

# NVIDIA and SAP Bring Trust to Specialized Agents Source: [https://blogs.nvidia.com/blog/sap-specialized-agents/](https://blogs.nvidia.com/blog/sap-specialized-agents/) From finance and procurement to supply chain and manufacturing, specialized AI agents are moving into the enterprise systems where business decisions are made, data is accessed and workflows run at scale\. Announced today at SAP Sapphire — where NVIDIA founder and CEO Jensen Huang joined SAP CEO Christian Klein’s keynote by video — SAP and NVIDIA’s expanded collaboration helps enterprises run specialized agents with security and governance controls\. SAP embeds[NVIDIA OpenShell](https://build.nvidia.com/openshell)— an open source runtime for securely developing and deploying autonomous AI agents — into SAP Business AI Platform\. In addition, SAP engineers are codesigning OpenShell alongside NVIDIA, contributing back to the open source project\. OpenShell provides isolated execution environments, policy enforcement at the filesystem and network layers, and infrastructure\-level containment that guards against damage when agent logic fails\. Within SAP Business AI Platform, OpenShell is the runtime security layer for all SAP AI agents, including custom agents built in Joule Studio — SAP’s environment for building and managing end\-to\-end enterprise agents\. For enterprises, the shift from AI assistants to autonomous agents changes the trust equation\. An agent that can touch systems of record, cross application boundaries and operate without review at every step needs boundaries, policy enforcement and an audit trail before it can become part of production work\. That’s what SAP and NVIDIA are collaborating to address\. ## **Why the Application Layer Matters** Huang has described[AI as a five\-layer cake](https://blogs.nvidia.com/blog/ai-5-layer-cake/): energy, chips, infrastructure, models and applications\. Applications sit at the top, where AI creates economic value and drives productivity for knowledge workers\. As a global leader in enterprise applications and business AI, SAP is an important catalyst in the application layer, running finance, procurement, supply chain and manufacturing workflows where agents must operate within policy, identity and process controls\. That makes SAP’s position at the core of enterprise operations a key driver for enterprise adoption of agentic AI\. Business agents need to understand roles, processes, permissions and data boundaries\. They also need an execution environment that limits what an agent sees, what it can do and where inference runs\. ## **Codeveloping the Agentic AI Foundation** NVIDIA brings its own perspective as a longstanding SAP customer — running finance, supply chain and logistics on SAP, giving both companies shared context for what enterprise\-grade governance requires in practice\. SAP engineers are working alongside NVIDIA’s to further develop OpenShell’s open source codebase, focusing contributions on what enterprises need to run agentic AI in production: runtime hardening, policy modeling, enterprise identity integration, and auditing and governance hooks\. SAP and NVIDIA technologies work in unison to address important enterprise requirements for deploying trustworthy agents\. - NVIDIA OpenShell asks: Can this agent action safely execute? - Joule Studio runtime — the enterprise control layer within SAP Business AI Platform — asks: Should this action happen at all? Together, they close a gap that application\-layer security alone cannot\. ## **A Faster Start for Agent Builders** SAP customers building custom agents will have a faster path to production\.[NVIDIA NemoClaw](https://www.nvidia.com/en-us/ai/nemoclaw/)— a reference blueprint for developing and deploying autonomous agents — will be available directly in Joule Studio\. This means development teams get a structured route from initial build to trusted production deployment, without having to engineer security scaffolding from scratch\. AI agents will create value only when enterprises can trust them with their data\. For many organizations, that data is in SAP — which houses the systems of record across finance, procurement and supply chain that run their business\. Together, SAP and NVIDIA are making agents ready to act, while staying within the boundaries enterprises require\. *Learn more about*[*NVIDIA OpenShell*](https://build.nvidia.com/openshell)*and*[*NemoClaw*](https://www.nvidia.com/en-us/ai/nemoclaw/)*\.* *See*[*notice*](https://www.nvidia.com/en-eu/about-nvidia/terms-of-service/)*regarding software product information\.*

Similar Articles

How Autonomous AI Agents Become Secure by Design With NVIDIA OpenShell

NVIDIA Blog

NVIDIA launches OpenShell, a secure-by-design runtime for autonomous AI agents that isolates agent operations in sandboxes and enforces security policies at the system level rather than relying on behavioral prompts. The toolkit, part of NVIDIA Agent Toolkit, enables enterprises to run coding agents and agentic workflows with unified policy management and compliance oversight.

HPE AI Factory With NVIDIA Expands for the Era of Agents

NVIDIA Blog

NVIDIA and HPE are expanding their AI factory collaboration with the NVIDIA Vera CPU for agentic AI, the NVIDIA Agent Toolkit for HPE Private Cloud AI, and NVIDIA Confidential Computing across the portfolio, enabling enterprises to move agentic AI into production.

NVIDIA Brings Trusted, 24/7 AI Agents to Telecom Operations

NVIDIA Blog

NVIDIA announces new AI agents and tools for telecom operations, including synthetic data generation and secure agent runtimes, showcased at DTW Ignite 2026. The platform aims to enable autonomous networks by combining domain-specific models, privacy-safe synthetic data, and policy-based guardrails.