For the past few years, AI has largely been defined by copilots that assist with tasks such as writing, coding, and research. That model is now evolving. Enterprises are beginning to explore autonomous AI systems that can execute workflows, coordinate across tools, and operate continuously. The challenge is no longer capability. It is control.
Bringing autonomous AI into enterprise environments requires clear governance, strong security boundaries, and confidence in how systems access and use data. Without these, scaling beyond experimentation is difficult.
At NVIDIA GTC 2026, this challenge took center stage. NVIDIA introduced NemoClaw, an open-source stack for building agentic AI, along with NVIDIA OpenShell, a secure runtime environment designed to let autonomous systems operate within defined policies and guardrails. Together, these technologies make it possible to run AI systems more safely, with greater visibility into how they behave and interact with enterprise data.
These capabilities are a critical step toward moving agentic AI from experimental use cases to enterprise-ready deployment.
This is where Lenovo plays a key role.
Turning capability into enterprise reality
While NVIDIA provides the foundation for secure execution, Lenovo enables organizations to deploy, scale, and manage these systems in real-world environments. That includes the infrastructure, services, and governance frameworks required to move from proof of concept to operational impact.
Claw-based systems running on Lenovo infrastructure powered by NVIDIA GB10 Grace Blackwell Superchip support long-running, always-on AI workflows. These systems are designed to handle complex, continuous operations locally, with high performance and large memory capacity. Running AI workloads locally reduces reliance on external APIs, improves responsiveness, and helps organizations maintain stronger control over sensitive data.
Lenovo’s ThinkStation PGX powered by NVIDIA GB10 superchip and 128 GB of unified memory delivers the compute capabilities to support multiple long-running AI agents. These platforms support continuous agent workloads, while also enabling high-performance AI inferencing, allowing organizations to run models and agents locally with speed, efficiency, and greater control over data.
Built on Lenovo’s Hybrid AI Advantage with NVIDIA, this foundation allows organizations to deploy AI across environments. Sensitive workloads can remain on-premises while organizations scale seamlessly across cloud and larger enterprise infrastructures.
Governance and scalability by design
A key differentiator is Lenovo’s AI services, along with Lenovo xIQ, a suite of AI-native platforms designed to simplify deployment and enforce governance.
With these capabilities, organizations can:
- Manage and monitor AI-driven environments
- Align access with enterprise policies
- Control how tools and data are used
- Roll out AI in a structured and scalable way
This ensures that autonomous AI can scale across the enterprise without introducing unnecessary risk.
Delivering measurable outcomes
Across AI deployments both internally and externally, Lenovo has seen significant results[1] including:
- Up to 90% adoption of AI assistants
- Productivity improvements of up to 30% in knowledge work
- Efficiency gains of up to 40% across support and operational teams
These outcomes are proof that when AI is deployed with the right infrastructure, governance, and user enablement, organizations can move quickly from pilots to real operational impact.
What’s next
Enterprises are already applying these capabilities across a range of use cases. In customer support, AI systems can assist representatives by gathering context, suggesting responses, and handling routine inquiries. In sales, they can track interactions, generate follow-ups, and keep CRM systems continuously updated. As research and knowledge tools, they can pull from internal data sources to generate reports, summarize insights, and support faster decision-making.
As adoption grows, the conversation is shifting.
The focus is moving from what AI can do to how effectively it can be deployed at scale and how consistently it can deliver meaningful business impact.
[1] Internal Lenovo data