Syself Autopilot is fully managed Kubernetes that runs on your own Hetzner servers. You describe a cluster as a single Kubernetes object and apply it; Syself builds the cluster, keeps it healthy, and upgrades it. Your workloads run on standard, upstream Kubernetes, on machines in a Hetzner account that stays yours.
We have run Kubernetes on Hetzner in production since 2020. We build and maintain [cluster-api-provider-hetzner](https://github.com/syself/cluster-api-provider-hetzner), the most-starred Cluster API provider on GitHub, and contribute upstream to the Kubernetes projects it runs on. In 2023 we won the public tender for the [Kubernetes-as-a-Service part of the Sovereign Cloud Stack](/news/syself-awarded-scs-vp-05-tender-to-develop-kubernetes-as-a-service-for-sovereign-cloud-stack), a German government initiative, and built the cluster-stack approach there. Since then we have managed thousands of nodes for agencies, enterprises, and startups.
## How you work with Syself Autopilot
Everything is declarative. You define and create your clusters through a central Kubernetes API, the management cluster, where you apply and change the objects that describe them (learn more about [accessing the management cluster](/docs/hetzner/apalla/getting-started/access-the-management-cluster)). You apply a manifest, and Syself Autopilot reconciles your infrastructure to match it, then keeps it that way.
Syself Autopilot builds on [Cluster API](https://cluster-api.sigs.k8s.io/), the widely adopted, proven standard for managing Kubernetes clusters as Kubernetes objects, and adds a layer of automation on top. If you already know Cluster API, you will recognize some of the custom resources. If you do not, you do not need to: the whole flow follows the same Kubernetes conventions you already use, so it stays familiar and consistent.
Two ideas shape how Syself Autopilot runs your clusters:
- **Nodes are disposable.** A node is never patched in place. When it needs to change, for an upgrade, a fault, or a security fix, a fresh, tested node replaces it. Every node runs Syself Linux, an immutable operating system, built from source and verified at boot.
- **A cluster stack is one tested version.** The node operating system, the cluster topology, and every component inside the cluster ship together under a single version number, built and tested as one unit before any node boots. You track one number instead of a dozen.
To go deeper on the model, including how the management and workload clusters divide the work, how self-healing and upgrades run, and what makes the node OS verifiable, read the [Concepts](/docs/hetzner/apalla/concepts/overview) tab.
## Structure of this documentation
If you just want to get going, continue to the next page and follow the [Getting Started guide](/docs/hetzner/apalla/getting-started/prerequisites). By the end you will have a highly available, production-ready cluster running on Syself Autopilot.
> [!NOTE]
> Reaching the management cluster needs an Autopilot account, and there is no self-service signup yet.
>
> - **Already have an account?** Continue to [Prerequisites](/docs/hetzner/apalla/getting-started/prerequisites).
> - **Not yet?** [Reach out for a trial or free-plan account](/docs/hetzner/apalla/support/faq#how-do-i-get-access-and-start). Syself sets it up and emails you an invite to log in, along with the details you'll need later, like your namespace. You can install the tools in [Prerequisites](/docs/hetzner/apalla/getting-started/prerequisites) meanwhile; the account only matters once you [access the management cluster](/docs/hetzner/apalla/getting-started/access-the-management-cluster).
The tabs at the top of the navigation split the docs by job:
- Start with [Concepts](/docs/hetzner/apalla/concepts/overview) to understand how the platform works, and why, before you build.
- The task tabs (Cluster management, Servers and nodes, Networking, Storage, Observability, Workloads, and Security) each own one area of running clusters in production.
- [Platform](/docs/hetzner/apalla/platform/add-users) covers users and authorization: who can manage or view which clusters.
- If something goes wrong, the [Troubleshooting](/docs/hetzner/apalla/clusters/troubleshooting/cluster-creation-fails) guides and the [Support](/docs/hetzner/apalla/support/where-to-start) tab explain how to reach us and what to collect first.
## Using the docs
The version picker in the header selects the cluster-stack release the docs describe. The content changes with it, so pick the one that matches what you are doing:
- **Creating a cluster:** select the latest version, unless you specifically want a different one.
- **Upgrading an existing cluster:** select the current version of the cluster you are upgrading, so the upgrade guides show the right steps.
For anything else, such as troubleshooting or the task guides, use the version your clusters are on, or the latest version if you do not have a cluster yet.
> [!TIP]
> Press and to move between neighboring pages, and to search the docs from anywhere.