# OpenClaw Enterprise: A Free, Self-Hosted Control Plane for Persistent Agents, Ready Only for Internal Pilots Until 1.0

The OpenClaw Foundation released OpenClaw Enterprise on September 29: a free, open-source, vendor-neutral platform for running persistent agents in sensitive environments, with multi-tenancy and hard security boundaries. Its maintainers rate it for internal pilot workloads only, with 1.0 planned for later this year.

By Neomanex. Updated 2026-10-20.

**Our verdict: pilot OpenClaw Enterprise, and wait for 1.0 and the security reference architecture before you trust it with anything sensitive.** OpenClaw Enterprise (OCE) promises what teams running shared agents need: multi-tenancy and hard security boundaries, free, on your own infrastructure. But its own maintainers rate it for "internal pilot workloads", and the reference security architecture is still due "in the coming weeks". Our directory rates both OpenClaw Enterprise and the OpenClaw assistant as conditional. We have not run OCE ourselves.

## What happened

On September 29, 2026 the OpenClaw Foundation announced OpenClaw Enterprise.

- **What it is:** "an open source, vendor neutral platform for managing persistent agents in sensitive environments".
- **Features:** multi-tenancy, hard security boundaries, standardized agentic primitives, a swappable harness, model and sandbox, and governance and auditability across the agent lifecycle.
- **Deployment:** self-host from the repository, with docker-compose for local development or Kubernetes for internal deployment.
- **Cost:** "OCE is built to run on your own infrastructure and will always be free for any organization to use." The announcement says "open source" but names no licence.
- **Maturity:** suitable today for "internal pilot workloads", with 1.0 planned for later this year. A reference security architecture is due "in the coming weeks".
- **Origin:** OCE started at OpenAI, was donated to the OpenClaw Foundation, and was developed further with Red Hat and NVIDIA.

## Why it matters

Shared, persistent agents in sensitive environments are exactly the gap in the OpenClaw assistant. Our OpenClaw entry is conditional partly because OpenClaw's own security docs say it is not a hostile multi-tenant security boundary for mutually adversarial users. OCE is a separate product from the same foundation, and it advertises multi-tenancy and hard security boundaries.

That is the right target, but the evidence is not there yet. The security claims arrive before the reference architecture that is supposed to show how they hold together, and the maintainers themselves cap it at internal pilots. The announcement also does not say whether OCE runs the OpenClaw assistant or changes its shared-gateway limit, so the OpenClaw rating stays where it is.

## What changes for you

- **Evaluating multi-tenant agent hosting:** spin OCE up with docker-compose in a non-production environment and test the tenant isolation yourself.
- **Running anything sensitive:** hold production use until the reference security architecture and the 1.0 release ship.
- **Already on the OpenClaw assistant:** nothing changes. Do not read OCE's multi-tenancy as a fix for the assistant's shared-gateway limit; the announcement makes no such claim.
- **Licence review:** the post names no licence, so check the repository before your legal team signs off on "open source".

Canonical: https://neomanex.com/news/openclaw-enterprise-self-hosted-agent-control-plane-pilot