The OpenClaw Foundation has introduced OpenClaw Enterprise, an open-source platform for managing persistent AI agents in sensitive enterprise environments. The launch, announced on September 30, is designed to address security, safety, and governance concerns that have slowed the adoption of agentic systems. The foundation is developing the project in the open ahead of a planned 1.0 release later this year.
Addressing Security and Governance Gaps
Many IT organizations currently restrict or ban agentic platforms because common security, safety, and governance standards are missing. OpenClaw Enterprise adds an enterprise grade control plane with multi-tenancy, audit trails, and permission controls to help close that gap. The approach is intended to let organizations deploy capable agents without reducing their functionality in sensitive settings.
An Open and Vendor Neutral Foundation
OpenClaw Enterprise is being developed as an open source, vendor neutral platform before its 1.0 release. The system supports multi-tenancy, hard security boundaries, and standardized agentic primitives while providing governance and auditability across the agent lifecycle. Organizations can also swap core components, including the model, sandbox, and harness, with third party or internal implementations.
Development and Community Collaboration
The initiative began at OpenAI and was later donated to the OpenClaw Foundation, where it became an independent project. Since that transition, the platform has been further developed in collaboration with Red Hat and NVIDIA. The foundation said it prioritized open sourcing the work early to build in the open and co-develop the technology with a wider community ahead of the 1.0 release.
Pilot Use and Internal Adoption
OpenClaw Enterprise is currently intended for internal pilot workloads rather than full production use. Red Hat and OpenAI are already piloting the platform internally, and OpenAI has deployed OpenClaw agents with full access to codebases and plugins. Internal teams report that an agent called Androidclaw has helped triage issues, locate pull requests, and respond to technical problems quickly.
Deployment and Self Hosting Options
Organizations can self host OpenClaw Enterprise today by cloning the repository and following the getting started guide. Local development is supported through Docker Compose, while internal deployment can be carried out on Kubernetes. These options give teams control over their own infrastructure while testing agentic workloads in pilot environments before wider rollout.
Security Architecture and Hard Boundaries
Security is the primary focus of OpenClaw Enterprise. The platform combines hard boundaries between trusted and untrusted workloads with sandboxing, LLM-based reviews, and granular permission controls. The foundation plans to publish a reference architecture in the coming weeks to show how these protections operate together in practice for enterprise deployments.
Enterprise Agent Landscape
The launch arrives as enterprise buyers place a higher priority on AI agent security. Red Hat has separately positioned its AI portfolio around agentic AI and running AI workloads on OpenShift, although its public materials do not currently show OpenClaw integrated with that stack. The foundation's project remains distinct and focused on open infrastructure rather than a specific vendor ecosystem.
OpenClaw Enterprise represents an early but meaningful step toward open and secure AI agent infrastructure for the workplace. The OpenClaw Foundation is inviting developers, operators, security teams, and organizations to help shape the platform before its 1.0 release. As enterprise interest in agentic AI grows, the project aims to provide the security, governance, and deployment controls that allow organizations to run persistent agents on their own terms.