Navique AI Core OperatorA governed AI platform, delivered as Kubernetes resources

Apply a handful of custom resources and the operator stands up a complete, self-hosted AI stack — LiteLLM gateway, LibreChat UI, Langfuse observability, the datastores, and the secrets plumbing. No manual Helm or kubectl steps.

Get Started

Quick Start

Core Concepts

One platform, many resources

Per-component CRDs for the gateway, UI, observability, datastores and secrets — independently composable, cross-referenced, and shareable across namespaces.

Browse the custom resources

Installs what it needs

Bundles the official upstream Helm charts (LiteLLM, Langfuse, CloudNativePG, External Secrets, and more) inside the binary and drives them with the Helm v4 Go SDK — lazily, only when a resource requires them.

How it works

Provenance-aware

Never re-installs an operator you already run; never deletes what it does not own. Datastores can be operator-managed, adopted, or external.

Provenance & lifecycle

Secrets are never inline

Credentials flow from External Secrets (Azure Key Vault, Vault, AWS, GCP…) or Sealed Secrets. Operator-generated keys are owned, GC'd, and rotatable.

Secrets management

Auto-wiring across components

Reference a Gateway from your ChatUI, or a Langfuse from your Gateway, and the operator provisions and injects the credentials so you configure no secrets by hand.

Cross-component wiring

Open core, air-gapped licensing

AGPL-3.0 open core that runs fully without a license. Enterprise features unlock with a signed, offline license — no license server, no phone-home.

Editions & licensing

From a stack of Helm charts to a single operator

The Navique AI Core Operator replaces a hand-managed Helm + ArgoCD deployment of the Forge / Navique AI platform. Instead of orchestrating sync waves and chart values by hand, you describe the platform you want as Kubernetes objects and let the operator converge the cluster to that state — installing capability operators, provisioning datastores, wiring credentials, and keeping everything healthy.

yaml

# Bring up a LiteLLM gateway backed by a managed Postgres database,
# exporting traces to a shared Langfuse — in one resource.
apiVersion: core.navique.com/v1alpha1
kind: Gateway
metadata:
  name: gateway
  namespace: forge-gateway
spec:
  secretsRef: { name: forge-secrets }
  database:
    mode: postgresCluster
    postgresClusterRef: { name: forge-pg, namespace: forge-data }
    databaseName: litellm
  instance:
    masterKey: { autoGenerate: true }
    saltKey: { autoGenerate: true }
  observabilityRef: { name: observability, namespace: forge-langfuse }
  models:
    - name: gpt-5.4
      modelName: azure_ai/gpt-5.4
      model: azure_ai/gpt-5.4
      refSecretKey: OPENAI_API_KEY

New here? Start with the Introduction, then follow the Quick Start to bring up a full platform on a cluster.

Navique AI Core OperatorA governed AI platform, delivered as Kubernetes resources

One platform, many resources

Installs what it needs

Provenance-aware

Secrets are never inline

Auto-wiring across components

Open core, air-gapped licensing

From a stack of Helm charts to a single operator ​

From a stack of Helm charts to a single operator