The OpenClo Foundation, which manages the open-source AI agent tool OpenClo, said on Monday it has released its enterprise AI agent management platform OpenClo Enterprise (OCE) as open source.

The foundation said OCE manages agents that operate continuously in sensitive environments.

The foundation said that nearly a year after the launch of OpenClo, adoption of always-on agents remains slow. It said companies have demanded common standards for security, safety and governance to adopt agents in earnest.

OCE provides an enterprise-grade control plane for agents. It offers multi-tenancy, security boundaries and standardized agent components, as well as management and auditing functions across the agent lifecycle. Core elements such as harnesses, models and sandboxes can use external solutions or in-house implementations as needed.

The foundation currently recommends OCE for in-house pilot work. It said it released the platform early to develop it with the community before rolling out version 1.0 later this year.

OCE was originally developed by OpenAI and donated to the OpenClo Foundation. It is now an independent project, with Red Hat and Nvidia also participating in development.

Red Hat and OpenAI are running OCE in internal pilots. OpenAI is already operating OpenClo agents with full privileges over the code base and plug-ins.

The OpenClo Foundation cited security as OCE's top priority. It said it will release a reference architecture within weeks showing how safeguards work in practice.