Nvidia Expands AI Agent Ecosystem With New Toolkit for Adobe, SAP, Salesforce

At its GTC conference, the Santa Clara-based company unveiled the Nvidia Agent Toolkit, an open-source software platform designed to help enterprises deploy AI agents that not only respond to commands but also think, plan, and act autonomously.

It’s a significant pivot from hardware dominance into the increasingly crowded arena of enterprise software, and Nvidia is bringing some very big friends along for the ride.

Adobe, Salesforce, SAP, ServiceNow, Siemens, Atlassian, and more than a dozen other industry giants have signed on to integrate the toolkit into their platforms.

What is the Agent Toolkit, exactly?

At its core, the Agent Toolkit is a collection of open-source building blocks for AI agents. It includes open models (Nvidia’s own Nemotron family), open agent blueprints (NVIDIA AI-Q), open skills like cuOpt for optimization tasks, and now a new addition: NVIDIA OpenShell.

OpenShell is arguably the most consequential piece of the puzzle. It’s a runtime environment that gives AI agents the access they need to browse systems, call APIs, and execute tasks, while enforcing policy-based guardrails around security, network access, and data privacy. 

Nvidia CEO Jensen Huang framed the moment in typically grand terms. “Claude Code and OpenClaw have sparked the agent inflection point — extending AI beyond generation and reasoning into action,” he said in a statement. “Employees will be supercharged by teams of frontier, specialized and custom-built agents they deploy and manage.”

The cost question gets an answer

One of the persistent criticisms of enterprise AI is the cost of running powerful models at scale. Nvidia is positioning its AI-Q Blueprint as a direct answer to that concern.

The blueprint uses what Nvidia calls a hybrid architecture. Frontier models handle high-level orchestration and reasoning, while the lighter, open-source Nemotron models do the heavy lifting on research and retrieval tasks. The result, according to Nvidia, is a more than 50% reduction in cost per query without sacrificing accuracy.

The company claims its AI-Q-powered agent currently tops both the DeepResearch Bench and DeepResearch Bench II accuracy leaderboards.

LangChain, whose open-source agent frameworks have been downloaded over 1 billion times, is partnering with Nvidia to bring AI-Q, OpenShell, and the Nemotron model family into its deep agent library.

The partner list reads like a corporate who’s who

The breadth of the enterprise adoption is noteworthy. Each company appears to be carving out a distinct use case:

Salesforce is building a reference architecture where employees can use Slack as the primary conversational layer for its Agentforce agents, pulling from both on-premises and cloud data environments. SAP is using NeMo to let customers design their own agents through Joule Studio on the SAP Business Technology Platform.

Adobe is adopting the toolkit to run long-running creativity, productivity, and marketing agents in what it describes as a “more secure and cost-efficient environment.”

In the semiconductor industry, Siemens is launching a Fuse EDA AI Agent powered by Nemotron to autonomously orchestrate workflows across its electronic design automation portfolio, from initial chip design through manufacturing. Cadence and Synopsys are making similar moves for chip design and verification workflows.

On the security front, CrowdStrike unveiled a Secure-by-Design AI Blueprint that embeds its Falcon platform directly into Nvidia’s agent architectures, while Cisco’s AI Defense will provide security controls for OpenShell deployments.

Healthcare and life sciences aren’t being left out either. IQVIA said it has already deployed more than 150 agents across internal teams and client environments, including 19 of the top 20 pharmaceutical companies, and is now integrating Nemotron into its unified IQVIA.ai platform.

Nvidia says OpenShell and the Agent Toolkit are available on build.nvidia.com, running across a range of inference providers and cloud partners. For those who prefer local development, OpenShell can be downloaded from GitHub and run on NVIDIA GeForce RTX PCs, RTX workstations, and Nvidia’s own DGX Spark and DGX Station supercomputers.

Enterprise deployments are supported across major cloud providers, including AWS, Google Cloud, Microsoft Azure, and Oracle Cloud Infrastructure.

Also read: The latest agentic AI trend shows why vendors are racing to move enterprise AI beyond chat and into systems that can plan, reason, and act.

The post Nvidia Expands AI Agent Ecosystem With New Toolkit for Adobe, SAP, Salesforce appeared first on eWEEK.