ServiceNow Launches Project Arc Desktop AI Agent with NVIDIA Security

At Knowledge 2026 in Las Vegas, ServiceNow unveiled Project Arc, an enterprise autonomous desktop agent that operates within NVIDIA's OpenShell runtime environment. The announcement represents a significant expansion of the two companies' partnership, extending agentic AI governance from desktops to data centers. This isn't just another chatbot—it's software designed to complete multi-step work across enterprise tools without requiring pre-built workflows.

Every action the agent takes runs inside NVIDIA OpenShell, a sandboxed runtime environment that contains autonomous activity while maintaining audit capabilities. ServiceNow AI Control Tower monitors behavior and logs files read, commands executed, and APIs called. The result is an autonomous desktop agent that enterprise security leaders can fully audit and approve with confidence (which is exactly what CISOs have been asking for, honestly).

According to the official ServiceNow press release, Project Arc is powered by ServiceNow Action Fabric and grounded in the ServiceNow Configuration Management Database. The agent has deep intelligence about the enterprise, pulling from workflows, systems, and operational history so that actions are informed by how work actually gets done. Employees can access it through a desktop application, enterprise collaboration tools, or email.

The physical reality of this matters. Unlike standalone AI agents that operate in isolation, Project Arc connects natively to the ServiceNow AI Platform. It can access local file systems, terminals, and applications installed on a machine to complete complex, multistep tasks that traditional automation can't handle. Think of it as an agent that doesn't just suggest actions—it executes them while staying within policy boundaries.

ServiceNow AI Control Tower is now included in the NVIDIA Enterprise AI Factory validated design, extending enterprise governance to the infrastructure layer where large-scale AI model workloads run. For organizations managing AI at scale, this integration provides a unified governance layer for data centers. AI factory deployments governed by AI Control Tower gain continuous value, risk, and security management across the full model lifecycle.

The companies are also advancing NOWAI-Bench, an open benchmarking suite comprising two frameworks: EnterpriseOps-Gym, which evaluates multi-step agentic performance spanning IT service management, customer service, and HR workflows, and EVA-Bench, a voice agent evaluation framework for enterprise settings. Both benchmarks are available as open-source releases, and NVIDIA is integrating them into NeMo Gym for automated model evaluation.

NVIDIA evaluated its Nemotron 3 Super model on EnterpriseOps-Gym, with results showing performance among open-weight models. EVA-Bench serves as a benchmark for developing NVIDIA's Nemotron VoiceChat model. These aren't theoretical exercises—they're designed to measure real-world performance where enterprise AI systems often encounter actual challenges.

Joe Davis, executive vice president of AI Engineering & Delivery at ServiceNow, stated the partnership aims to make AI real for the enterprise. "Whether it's autonomous AI agents that can be trusted on the desktop, governance that extends to the data center, or open benchmarks that hold the entire industry accountable, this is enterprise AI that's built to last." Kari Briski, vice president of Generative AI for Enterprise at NVIDIA, emphasized that delivering autonomous agents securely at scale requires governance spanning models, software, and AI infrastructure.

Project Arc is available as an early preview. The AI Control Tower integration with NVIDIA Enterprise AI Factory validated design is generally available. NOWAI-Bench is available as an open-source release. These availability distinctions matter for enterprises evaluating deployment timelines.

Investing.com reported that ServiceNow remains a prominent player in the Software industry despite shares down 53% over the past year. The company maintains strong revenue growth of 22%, and analysts including Benchmark, TD Cowen, and Piper Sandler have maintained Buy or Overweight ratings citing AI growth outlook and sustainable revenue prospects.

Whether enterprises actually deploy autonomous agents at scale remains the real question. The technology exists, the governance frameworks are in place, and the benchmarks are public. But the friction between IT security teams approving autonomous execution and business units demanding faster AI adoption hasn't disappeared—it's just been given a new interface to manage.

Artūras Malašauskas is an AI Systems Integrator with 20+ years of production-grade web engineering experience. He has designed, shipped, and scaled enterprise Python/PHP systems for logistics, SaaS, and public-sector clients. For the past year, he has focused exclusively on AI integrations: deploying open-source LLMs, building generative media pipelines (image, audio, video), and engineering multi-agent workflows for real production environments. His standard: reproducibility, security, cost-efficient inference—no vaporware. He documents and evaluates emerging AI tooling, separating verified capabilities from marketing noise. Technical editor at: muza-ai.eu, ai-verslas.lt, ai-naujinos.lt Connect on LinkedIn