Beyond Cloud
Sign in Get early access
Get early access Sign in

Early accessTalos on Hetzner Cloud

Kubernetes in your own Hetzner account, provisioned by a workflow you can watch.

Beyond Cloud turns a Hetzner Cloud API token into a Talos Kubernetes cluster. Twelve explicit steps with retries and recovery, live status, and one workspace for your team's credentials, kubeconfigs and history.

Get early access Talk to us

For small platform teams and engineers who want production-shaped clusters without writing the provisioning tooling themselves.

Recreated from the Beyond Cloud app with example data. The step names are the real provisioning workflow.

From API token to kubeconfig in three moves

The servers are created in your Hetzner project, so you can see every resource in the Hetzner console and Hetzner bills you for them directly.

  1. 1

    Connect your Hetzner Cloud project

    Paste a Hetzner Cloud API token. It is encrypted with AES-256-GCM and stored as a secret that belongs to your workspace.

  2. 2

    Choose the cluster shape

    Pick a Hetzner location, how many control plane and worker nodes you want, and the server types. Hetzner list prices are shown before you create anything.

  3. 3

    Watch it come up

    A 12-step workflow creates the servers, load balancer and Talos configuration, then bootstraps Kubernetes. Each step reports live. When it finishes, download the kubeconfig.

One request starts the whole workflow

The dashboard and the API use the same endpoint. Send a name, a location and a node count. Beyond Cloud records the cluster, starts the provisioning workflow and returns right away.

# Create a 3 + 2 node Talos cluster in Nuremberg
curl -X POST https://api.usebey.com/api/kubernetes/clusters/talos \
  -H "Authorization: Bearer $BEY_TOKEN" \
  -H "X-Organization-ID: $BEY_WORKSPACE" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "staging-nbg1",
    "region": "nbg1",
    "control_plane_nodes": 3,
    "worker_nodes": 2,
    "server_type": { "control_plane": "cpx21", "worker": "cpx21" }
  }'

# 201 Created (abbreviated)
{
  "id": 2,
  "name": "staging-nbg1",
  "status": "PENDING",
  "control_plane_nodes": 3,
  "worker_nodes": 2,
  "workflowId": "…"
}

Requests are authenticated with your session token and scoped to a workspace. Personal API keys are not available yet.

The 12 provisioning steps

  1. 01 Validate configuration Checks the request and that the Hetzner token can reach your project.
  2. 02 Find Talos image Locates the Talos snapshot to boot servers from.
  3. 03 Create load balancer Creates the load balancer that fronts the Kubernetes API.
  4. 04 Generate Talos configuration Generates machine configs and cluster certificates.
  5. 05 Create control plane nodes Creates the control plane servers.
  6. 06 Create worker nodes Creates the worker servers.
  7. 07 Wait for node readiness Waits until every server reports running.
  8. 08 Configure load balancer Points the load balancer at the control plane.
  9. 09 Bootstrap Talos Bootstraps etcd and the control plane.
  10. 10 Generate kubeconfig Issues the admin kubeconfig for the cluster.
  11. 11 Verify cluster health Checks nodes and core components.
  12. 12 Finalize cluster Records endpoints and marks the cluster ready.

See where every cluster stands, and recover when a step fails

Provisioning on real infrastructure fails sometimes. Beyond Cloud records what happened, retries what it can, and shows you the rest.

Recreated from the Beyond Cloud app with example data.

Step history
Each run stores every step's status, timing and logs, so you can see exactly where it stopped.
Automatic retries
Failed steps are retried with exponential backoff, three times by default, before the run is marked failed.
Stuck-run recovery
A background check every five minutes picks up runs that have stalled for more than ten minutes.
Live status
Step progress streams to the dashboard over WebSocket. No need to refresh.
Cancel and re-run
Cancel a run that is in progress, or re-run a failed step after you fix the cause.
Import what you already run
Register an existing cluster as read-only, with its GitOps repository and Argo CD link alongside.

Your account, your servers, your keys

Beyond Cloud orchestrates. It doesn't resell compute. Clusters run on servers in your Hetzner Cloud project, and the workspace keeps the credentials and outputs your team needs.

  • Encrypted cloud credentials

    Provider tokens are encrypted at rest with AES-256-GCM. Lists show names and types, not values, and every change to a value is versioned.

  • Workspaces with members

    Clusters, secrets and workflow history belong to a workspace. Invite teammates with owner, admin or member roles.

  • Standard Talos and Kubernetes

    No proprietary agent sits between you and the cluster. You get a normal kubeconfig and use kubectl, Helm or Argo CD as usual.

  • Clean teardown

    Deleting a cluster runs its own workflow. It removes the provider resources first, then the records.

Early access, honestly priced

Beyond Cloud is in early access and its pricing isn't set yet. The product has no self-serve billing, and we'll agree any fee with you in writing before we charge you. Hetzner bills you for the servers and load balancers your clusters use.

Get early access Pricing and access details

Built by the people who run it

Beyond Cloud is built by a small independent team that runs its own Talos clusters on Hetzner. In early access you work directly with the engineers who build the product. Write to [email protected] and an engineer will answer.

More about us