EnterpriseClaw, developed with Cisco, NVIDIA, Okta, and OpenAI, lets organizations deploy AI agents across cloud, desktop, and on-premises systems under centralized control
SOPA Images / Getty Images
Automation Anywhere has announced EnterpriseClaw, a capability built in collaboration with Cisco $CSCO, NVIDIA, Okta, and OpenAI that allows organizations to deploy claw-style AI agents across cloud platforms, desktops, on-premises systems, and secured enterprise networks, the company said.
The category of claw-style AI agents works by performing tasks within live applications, local terminals, browsers, and system environments. The company noted that agents in this category have historically been built around single-user or self-contained setups, whereas real enterprise operations involve distributed teams, varied infrastructure, and compliance-sensitive systems. The platform aims to extend that agent capability into production enterprise environments, giving organizations unified oversight of what agents can access, what they do, and how they are governed.
The collaboration divides responsibilities across specialized domains. Security comes from Cisco through its AI Defense and DefenseClaw technologies. NVIDIA brings two components: OpenShell, its open-source agent runtime, and NIM microservices running on Nemotron models for organizations that need on-premises deployment. Identity and authentication across agents is handled by Okta. Through its participation, OpenAI makes its models available for agent construction and operation within the platform, with GPT-5.5 among those supported.
Underpinning the platform are two proprietary Automation Anywhere technologies — the Process Reasoning Engine and the Contextual Intelligence Graph — which the company said give agents a stronger foundation for handling complex business processes than a standalone language model can offer.
"For AI to have a transformational impact on business, it needs to be able to do work where the work actually happens," Automation Anywhere CEO Mihir Shukla said in a statement. "Many claw-style AI agents are incredibly powerful, but most were designed for isolated environments or individual users — while enterprise operations span teams, cloud platforms, desktops, on-premises systems, and highly regulated infrastructure."
The partners emphasized security and identity as central concerns. "The widespread use of AI agents introduces an entirely new attack surface, making proactive, integrated security non-negotiable for enterprise adoption," said DJ Sampath, senior vice president and general manager of AI Software and Platform at Cisco, in a statement. Okta CPO Ely Kahn said in a statement that agents require "first-class identities" and described the company's integration as extending to agents the same authentication and least-privilege authorization standards that govern human users.
Jason Lee, OpenAI's head of enterprise technology, said in a statement that the platform gives organizations a path to deploy frontier AI models within agentic workflows suited to regulated and mission-critical work.
Organizations can bring their own agents into the platform regardless of how those agents were built — whether developed in-house or through external frameworks — and run them in parallel with automations already in place. Automation Anywhere has made the platform accessible through a preview program, with a full commercial release planned before the end of the year.
Join 500,000+ readers who start their day with Quartz.
By subscribing, you agree to our Terms of Service and Privacy Policy.
