Skip to main content
Running Kubernetes well usually means a pile of Helm values, a GitOps tool to babysit, a runbook per failure, and one person who knows how it all fits together. Ankra is the AI platform engineer that takes that pile off your team: you describe what a cluster should run, Ankra deploys it and keeps it in sync, and its AI reads your logs, events and deploy history to tell you why something broke and how to fix it. It is built for platform engineers and developers who want production Kubernetes without hand-maintaining the machinery around it.

Connect a cluster and deploy a first Stack

​
What you can do with Ankra

Get a cluster

Connect the cluster you already run (EKS, GKE, AKS, k3s, on-prem, even minikube) with one helm command, or have Ankra build a new one on your own cloud account.

Deploy what it runs

Compose Helm charts and manifests into a Stack, visually or by asking the AI, and deploy the same Stack to every cluster that needs it.

Find out why it broke

Ask “why is this pod failing?” on any page. The AI already has the logs, events, manifests and recent deploys, and proposes a fix you approve.

Automate it

Everything in the UI is also available from the ankra CLI, from your AI assistant over MCP, and from your CI pipelines.

​
How it fits together

  • An organisation is your team’s account: members, roles, credentials and billing.
  • A cluster is any Kubernetes cluster running the lightweight Ankra agent. The agent connects out to Ankra, so it works behind firewalls and NAT.
  • A Stack is what a cluster should run: add-ons (Helm charts) and manifests (your own YAML), with variables for what differs between clusters.
  • Git is optional. Stacks deploy straight through the agent. Connect a repository when you want every change reviewed and versioned, and edits flow both ways - see GitOps.

​
Start in 15 minutes

The Quickstart takes you from nothing to a connected cluster, a deployed Stack and your first answer from the AI. You do not need a cluster to follow it: the free playground gives you a real one in about two minutes.

Open the Quickstart

One path, a visible result at every step.

​
Run your CI on your own clusters

Ankra CI runs a repository’s tests, image builds and security scans as Kubernetes Jobs on a cluster you already connected, and reports back as a check on every pull request. Get started with Ankra CI moves your first repository in about 20 minutes.

Get started with Ankra CI

From GitHub Actions or GitLab CI to a green Ankra pipeline check.

​
Work from your own tools

The web UI, the CLI, the MCP server and the in-product AI all use the same tools and the same permissions, so anything you can do in one you can do in the others.
See the Ankra CLI guide.
Automating Ankra or building an agent on it? Use the MCP server or the CLI rather than driving the web UI with a browser. Both use scoped tokens, and every call lands in your organisation’s audit log.

​
Where to go next

Add a cluster

Import one you have, or pick a provider for a new one.

How Stacks work

The model behind everything you deploy.

AI Assistant

What the AI can see, and what it may change.

Security

What Ankra can access, and how to restrict it.

Cloud Cost

What each cluster, namespace and Stack costs.

API Reference

The REST API behind every surface.
Need a hand? Contact support or ask in the Ankra community Slack.