ZenML is an open-source framework (Apache 2.0) for orchestrating AI pipelines and agents from experiment to production. Teams define their workflows in Python and run them on their own or rented infrastructure. With Kitaru, agent sessions can be replayed on production data before a change ships. Besides self-hosting, ZenML Pro offers a managed control plane. It is a European alternative to proprietary MLOps platforms from US cloud providers.
ZenML GmbH is the publishing entity, and the website mentions an office in Munich. With ZenML Pro, the server, authentication and the metadata about pipeline runs run on ZenML's infrastructure, while compute, datasets and models stay in your own stack. Secrets are held in ZenML's environment by default but can be moved to a customer-managed secrets store. The documentation does not name a hosting region for ZenML Pro. When you self-host, you decide where all data resides. No parent company outside Europe is stated.