Enterprise AI has reached a fascinating crossroads. We've mastered generation. We've conquered reasoning. Now comes the most intriguing question yet: How should AI actually act in our businesses?
At ServiceNow Knowledge 2026, NVIDIA and ServiceNow unveiled their answer—a groundbreaking collaboration that transforms AI from a clever conversationalist into a capable coworker that can execute complex tasks autonomously while maintaining the security and governance enterprises demand.
Meet Project Arc: Your New AI Colleague
The centerpiece of this partnership is Project Arc, ServiceNow's revolutionary autonomous desktop agent designed specifically for knowledge workers, developers, IT teams, and administrators. Unlike standalone AI tools that operate in isolation, Project Arc integrates seamlessly with ServiceNow's AI Platform through their Action Fabric.
What makes Project Arc special? It can access your local file systems, terminals, and applications to complete multi-step tasks that traditional automation simply can't handle—all while maintaining the enterprise-grade controls your organization needs to deploy AI at scale.
"Project Arc represents the next step in our ongoing collaboration with NVIDIA, bringing autonomous execution to the desktop," explained Jon Sigler, Executive Vice President and General Manager of AI Platform at ServiceNow. "We're delivering the governance and security that enterprise AI requires."
Security First: The OpenShell Foundation
Here's where things get really interesting. Project Arc runs on NVIDIA OpenShell, an open-source secure runtime that creates sandboxed, policy-governed environments for autonomous agents. Think of it as a secure playground where AI agents can operate freely while being contained within strict boundaries.
With OpenShell, enterprises can define exactly:
- What an agent can see
- Which tools it can access
- How each action is contained and monitored
This addresses one of the biggest concerns about autonomous AI: ensuring it remains helpful without becoming harmful.
Building Specialized AI Skills
The partnership leverages NVIDIA's open ecosystem approach, allowing organizations to customize models and applications for their specific domains. The NVIDIA AI-Q Blueprint enables specialized research agents that can gather context, synthesize information, and support complex decision-making across business functions.
To ensure these systems perform reliably in real-world scenarios, the companies are advancing NOWAI-Bench—an open benchmarking suite specifically designed for enterprise AI agents. Unlike general benchmarks, these evaluations focus on the multi-step workflows where enterprise AI systems face their biggest challenges.
The Economics of Always-On AI
As AI agents become long-running and always-on, efficiency becomes crucial. This is where NVIDIA's AI factories shine. The NVIDIA Blackwell platform delivers impressive economics:
- More than 50x greater token output per watt than previous generations
- Nearly 35x lower cost per million tokens
For enterprises running agents across millions of workflows, these efficiency gains can determine how quickly AI moves from pilot programs to broad production deployment.
What This Means for Your Business
This partnership represents a fundamental shift in how we think about enterprise AI. We're moving beyond AI that responds to prompts toward AI that proactively executes complex workflows. The key ingredients for this transformation are finally coming together:
- Capable agents that can handle real business tasks
- Built-in guardrails that ensure safe operation
- Proven performance with measurable business impact
The Path Forward
AI is fundamentally changing how work gets done. The companies that will thrive are those that give their AI agents the infrastructure to act effectively, the context to make informed decisions, and the governance to keep every action accountable.
The NVIDIA-ServiceNow partnership isn't just about better AI—it's about creating a new category of digital workers that can augment human capabilities while maintaining the trust and control that enterprise environments require.
As we stand on the brink of this autonomous AI era, one thing is clear: the future of work isn't just about what AI can generate or reason about—it's about what AI can actually do.
Source: NVIDIA Blog by Kari Briski