Skip to main content
Each step below ends with something you can see, so you know it worked before you move on. What you need: an Ankra account (sign up at platform.ankra.app), and either a Kubernetes cluster you can run helm against, or nothing at all - the free playground gives you a cluster.

1. Get a cluster

The playground is a real, writable Kubernetes cluster that Ankra runs for you. It is ready in about two minutes and needs no cloud account.
1

Start it

On your first visit, the dashboard offers See a demo app first - click Start the playground. You land on the new cluster’s overview and can watch it provision.
2

Check it is ready

When it is ready, the overview shows Your app is live with the public URL of a small starter app Ankra deployed for you. Open it.
The free playground expires after the organisation has been away for 3 hours (a week on a paid plan), with a warning first. Everything you do here works the same on your own cluster.
Want Ankra to build a new cluster on Hetzner, AWS, DigitalOcean or another provider, or create EKS, GKE or AKS for you? See Add a cluster - then come back to step 2.

2. Deploy your first Stack

A Stack is what a cluster should run: Helm charts (called add-ons) and your own Kubernetes YAML (manifests), deployed together in the right order. You will deploy Grafana, the dashboarding tool, so there is something real to look at.
1

Open the Stack Builder

Open your cluster, go to Stacks, and click + Create. The builder canvas opens.
2

Add the add-on

Click Add Add-on, search for Grafana, and select it. The defaults are fine - click the add-on on the canvas if you want to see or change its Helm values.
3

Deploy it

Name the Stack (for example dashboards) and click Create Stack. This saves and deploys it in one step, straight through the agent - no Git repository needed.
4

Check it is running

The Stack’s status moves to deployed within a minute or two. Go to Kubernetes → Workloads → Pods and you will see the grafana pod Running.
You can build the same Stack by asking the AI instead: press ⌘ + J (Ctrl + J on Windows and Linux) and type “Add a Stack with Grafana to this cluster”. The AI picks the chart, builds the Stack and deploys it for you.

3. Ask the AI about it

Ankra’s AI sees what you are looking at - the pod’s logs, events, manifest and recent deploys - so you can ask plain questions without pasting anything.
1

Open the assistant on the pod

Open the grafana pod from step 2 and press ⌘ + J. Leave the mode on Ask, which reads and explains but does not change anything already running.
2

Ask a question

Try “Is this pod healthy, and how do I open Grafana and log in?” or “What would happen if this pod restarted?”. The answer is based on your cluster’s actual state.
When you want the AI to make changes, switch to Agent (it asks you to confirm each change) or Plan (you approve a whole plan once). See AI Assistant for what each mode may do.

4. Optional: keep the cluster’s configuration in Git

Stacks work without Git. Connect a GitHub repository when you want every change to be reviewed, versioned and reversible - Ankra then commits each Stack change to the repository, and commits you push are deployed back to the cluster.
1

Connect a repository

Open the cluster, go to Settings → GitOps, choose a GitHub credential (or add one), and pick the repository. Nothing extra is installed on the cluster.
2

Check the commit

Your dashboards Stack appears in the repository under clusters/, as Helm values and manifests you can review like any other code.
See GitOps for how two-way sync and conflicts work.

You are done

You now have:
  • a cluster connected to Ankra, showing Healthy
  • a Stack deployed to it, with its pod Running
  • an answer from the AI based on your cluster’s real state

What’s next

Ship your own app

Get a repository running on a cluster and redeployed on every merge.

Build a monitoring stack

Prometheus, Grafana and Loki as one Stack.

Use the CLI and your AI assistant

ankra chat, ankra cluster apply, and the MCP server for Claude and Cursor.

Reuse a Stack everywhere

Turn a working Stack into a template for every cluster.

Variables

One Stack, different values per cluster or environment.

Get alerted

Page your team in Slack when something in the cluster goes wrong.
Stuck on a step? Contact support or ask in the Ankra community Slack.