NVIDIA said enterprise software companies are adopting its new Agent Toolkit, Nemotron models, NeMo-related blueprints, OpenShell runtime and CUDA-X agent skills to build autonomous AI agents for business, engineering, healthcare, software development and operations use cases.
NVIDIA announced new software and partner integrations aimed at helping enterprise software companies build autonomous AI agents.
According to NVIDIA’s June 1, 2026 announcement, the company introduced Agent Toolkit software, Nemotron models, NeMo-related blueprints, an OpenShell runtime and CUDA-X agent skills for enterprise AI agent development. NVIDIA said the tools are intended for use across enterprise software, engineering, healthcare, software development and business operations.
NVIDIA described Agent Toolkit as software for building AI agents that can be integrated into enterprise applications. In its investor release, NVIDIA said the announcement also includes Nemotron models, OpenShell runtime and CUDA-X agent skills.
The company’s press release, distributed through GlobeNewswire and published on NVIDIA’s investor site, frames the offering as infrastructure for “autonomous enterprise AI agents.” The announcement names a range of enterprise software and technology companies as partners, including Cadence, Dassault Systèmes, Siemens, Synopsys, CrowdStrike, Palantir, Microsoft, Canonical and Red Hat.
GamesBeat also reported that NVIDIA announced new software, open source models and partnerships for enterprise autonomous AI agents, citing Agent Toolkit, Nemotron, NeMo-related components, OpenShell and CUDA-X skills.
NVIDIA said the new software and integrations are designed for companies building agents into enterprise workflows. The source materials point to several broad categories: business operations, software development, engineering and healthcare.
The partner list suggests NVIDIA is positioning its agent software across several parts of the enterprise software market. Cadence, Dassault Systèmes, Siemens and Synopsys are associated with engineering and design software. CrowdStrike and Palantir operate in security and data platforms. Microsoft, Canonical and Red Hat provide enterprise software, operating systems or cloud-related infrastructure.
NVIDIA did not, in the provided excerpts, disclose customer adoption figures, pricing details or performance benchmarks for the new agent tools. The announcement also does not provide independent evaluations of how the agents perform in production environments.
AI agents have become a major focus for enterprise software vendors because they promise to connect large language models with software tools, data sources and business processes. NVIDIA’s announcement is notable because it ties that trend to the company’s broader software and accelerated computing ecosystem.
The inclusion of CUDA-X agent skills indicates that NVIDIA wants agent developers to make use of its existing libraries and acceleration stack. The OpenShell runtime, described by NVIDIA as a secure runtime, points to the operational requirements of running agents inside enterprise systems, where access control, execution safety and reliability are important concerns.
Still, the announcement should be read as a vendor launch and partnership statement rather than proof of broad deployment. The available sources establish that NVIDIA announced the toolkit, related models and integrations, and that several named companies are listed as partners. They do not establish how widely the technology is being used, how mature each integration is, or whether customers have measured productivity or cost improvements.
NVIDIA is expanding its enterprise AI software push with Agent Toolkit, Nemotron models, NeMo-related blueprints, OpenShell runtime and CUDA-X agent skills. The company says enterprise software leaders are using or integrating these components to build autonomous AI agents across multiple business domains.
For now, the announcement shows where NVIDIA and its partners want enterprise agent development to go: toward packaged tools, runtime environments and domain-specific integrations that can be embedded into existing software platforms.
NVIDIA announced new software and partner integrations aimed at helping enterprise software companies build autonomous AI agents.
According to NVIDIA’s June 1, 2026 announcement, the company introduced Agent Toolkit software, Nemotron models, NeMo related blueprints, an OpenShell runtime and CUDA X agent skills for enterprise AI agent development.
NVIDIA said the tools are intended for use across enterprise software, engineering, healthcare, software development and business operations.
Continue reading
