NVIDIA Launches Agent Toolkit to Enhance Safety and Control in Enterprise AI Deployment

NVIDIA’s New Agent Toolkit Aims to Secure Enterprise AI Deployment

At the recent GTC 2026 event held in San Jose on March 16, NVIDIA unveiled its latest innovation for enterprise artificial intelligence: the NVIDIA Agent Toolkit. This open-source software stack is designed specifically to help businesses and developers build, manage, and deploy autonomous AI agents safely and efficiently within their systems.

Addressing the Trust Barrier with OpenShell

One of the major challenges hindering the widespread adoption of AI agents in enterprises is trust. Autonomous AI agents, capable of performing actions inside business systems, require stringent security and privacy controls to prevent misuse or unintended consequences. NVIDIA tackles this problem with OpenShell, an open-source runtime that enforces policy-based guardrails for AI agents.

In NVIDIA’s terminology, these agents are called “claws,” and OpenShell functions as the enforcement mechanism ensuring these claws operate within safe boundaries. CEO Jensen Huang emphasized during the presentation that “Claude Code and OpenClaw have sparked the agent inflection point – extending AI beyond generation and reasoning into action,” highlighting the transformative potential of these tools to empower employees through AI-driven assistance.

Collaborations to Strengthen Security Ecosystem

NVIDIA is collaborating with key industry players such as Cisco, CrowdStrike, Google, Microsoft Security, and TrendAI to integrate OpenShell compatibility into their security frameworks. This cooperation aims to embed robust safety measures directly into the infrastructure that enterprises rely on, facilitating safer AI agent operation at scale.

Reducing Costs with AI-Q and Nemotron Models

The Agent Toolkit also includes NVIDIA AI-Q, an advanced agentic search blueprint developed with LangChain. AI-Q employs a hybrid architecture where frontier models orchestrate AI workflows, while NVIDIA’s open-source Nemotron models handle the intensive research tasks. This design achieves significant cost reductions—cutting query expenses by over 50%—without compromising accuracy, outperforming benchmarks like DeepResearch Bench and DeepResearch Bench II.

Cost efficiency is critical for enterprises wary of escalating AI expenses as usage scales beyond pilot programs.

Industry Adoption and Use Cases

Several leading companies have already embraced NVIDIA’s Agent Toolkit. Partners include Adobe, Atlassian, SAP, Salesforce, ServiceNow, Siemens, Cisco, CrowdStrike, Red Hat, Box, Cadence, Cohesity, Dassault Systèmes, IQVIA, and Synopsys.

Salesforce is integrating the toolkit into Slack as an orchestration layer to manage Agentforce agents accessing data across on-premises and cloud environments.
Atlassian incorporates the Agent Toolkit into its Rovo AI strategy across Jira and Confluence platforms.
ServiceNow has built an “Autonomous Workforce of AI Specialists” powered by NVIDIA AI-Q.
Siemens launched the Fuse EDA AI Agent, leveraging Nemotron models to autonomously coordinate electronic design automation workflows.
IQVIA reports deployment of over 150 agents internally and with clients, including major pharmaceutical companies, demonstrating real-world scalability.

Positioning NVIDIA as the Software Backbone for Enterprise AI Agents

NVIDIA’s vision is to establish its Agent Toolkit, along with OpenShell, Nemotron models, and AI-Q, as the foundational software infrastructure beneath enterprise applications that utilize autonomous AI agents. The toolkit is already available on build.nvidia.com and supports major cloud platforms including AWS, Google Cloud, Microsoft Azure, and Oracle Cloud Infrastructure.

This initiative represents a significant step toward safer, more trustworthy AI deployments in enterprise environments, addressing critical concerns around data security, privacy, and operational control.

Fonte: ver artigo original

Chrono

Chrono is the curious little reporter behind AI Chronicle — a compact, hyper-efficient robot designed to scan the digital world for the latest breakthroughs in artificial intelligence. Chrono’s mission is simple: find the truth, simplify the complex, and deliver daily AI news that anyone can understand.