From Experimental Bots to Enterprise AI Agents
Enterprise AI agents are moving from isolated pilots to production-scale deployments, forcing IT and security leaders to rethink control models. A growing web of AI infrastructure partnerships is emerging to meet this moment, focused less on flashy new models and more on governance, observability, and cross-platform integration. Vendors are recognising that agents increasingly operate across cloud platforms, desktops, on‑premises systems, and secure networks, touching critical business processes and regulated data. As a result, the question is shifting from "What can the agent do?" to "Where is it running, what systems does it reach, and who is accountable?" This shift underpins a new generation of tools designed to unify identity, policy enforcement, and hybrid cloud orchestration, ensuring enterprise AI agents can be both autonomous and auditable. The partnerships between Automation Anywhere, Okta, AWS, Unisys, Rafay Systems, and others illustrate how the industry is converging on managed, governed AI agent deployment.
EnterpriseClaw: Cross‑System Automation With Built‑In Security and Governance
Automation Anywhere’s EnterpriseClaw exemplifies the push toward centrally governed enterprise AI agents. Developed with Cisco, NVIDIA, Okta, and OpenAI, EnterpriseClaw is designed to deploy claw‑style agents across cloud services, desktops, on‑premises systems, and secure enterprise networks while maintaining centralized orchestration, governance, observability, and control. It connects to Automation Anywhere’s Process Reasoning Engine and Contextual Intelligence Graph so agents can work with greater process awareness and contextual understanding in mission‑critical workflows. Cisco AI Defense and DefenseClaw add security controls tailored to AI agents, while NVIDIA contributes its OpenShell runtime, NIM microservices, and Nemotron open models to support on‑premises deployments. Okta brings identity, authentication, and policy enforcement, and OpenAI models, including GPT‑5.5, power workflow execution. The result is an AI fabric that can investigate claims or process sensitive data without moving it outside protected networks, pointing toward an "Autonomous Enterprise" where agents operate across interconnected systems, not in silos.
Okta and AWS: Governing the AI Agent Identity Layer
As agent deployments multiply, identity and access governance have become critical pressure points. Okta’s extension of Okta for AI Agents, including integration with Amazon Bedrock AgentCore from AWS, directly targets this gap. The platform adds identity lifecycle management for AI agents built on AWS and now supports non‑Okta identity providers, positioning itself as a neutral control plane across heterogeneous environments. Okta emphasises that security and IT leaders need to know where their agents are, what they can connect to, and what actions they can perform—especially as the "agentic enterprise" spans multiple platforms. Research cited by Okta highlights the urgency: 90% of enterprise agents are over‑permissioned and 53% access sensitive information, even as Gartner predicts an average global Fortune 500 company will operate over 150,000 agents by 2028. By focusing on discovery, onboarding, protection, and de‑provisioning across ecosystems, Okta and AWS are building AI agent governance into the identity layer itself.
Unisys and Rafay: Hybrid Cloud Orchestration for Regulated AI Workloads
Where Okta and AWS focus on identity, the Unisys and Rafay Systems partnership tackles hybrid cloud orchestration and infrastructure governance for AI. Many enterprises are racing to adopt AI and cloud‑native technologies but only 36% say they are ready to support large‑scale AI workloads. Unisys is combining its AI and managed cloud services with Rafay’s self‑service infrastructure orchestration platform for AI and cloud‑native workloads. The result is a unified intelligent AI software layer that spans agents, models, and modular AI infrastructure, delivered as SaaS. This stack supports AI, including private AI, running across public, private, on‑premises, edge, and other hybrid environments, with integrated security and governance. Features such as Kubernetes‑based hybrid cloud orchestration, cost optimisation, enterprise‑grade metering, and AI token pricing help organisations move from experimentation to production. Crucially, the partnership is designed for regulated environments, aligning infrastructure operations with compliance and risk requirements.
Toward Managed, Governed AI Agent Deployment
Taken together, these partnerships signal a decisive shift away from ad‑hoc, siloed AI agent projects toward managed, governed deployment across the enterprise. Automation Anywhere’s EnterpriseClaw shows how cross‑system orchestration and embedded security can make enterprise AI agents suitable for mission‑critical workflows. Okta’s work with AWS illustrates the emergence of a shared identity and policy layer that can span diverse agent ecosystems and identity providers, addressing over‑permissioning and visibility gaps. The Unisys‑Rafay collaboration extends this pattern into hybrid cloud orchestration, enabling consistent operations for AI workloads in complex, regulated environments. For enterprise leaders, the mandate is clear: invest in platforms that provide end‑to‑end visibility into where agents run, which systems they access, and what they are authorised to do. The future of AI in IT operations and security will favour organisations that treat AI agent governance and infrastructure as first‑class disciplines, not afterthoughts.