Soft launch: we're still taming the tentacles—thanks for wiggling the knobs with us.
Features
ClusterClaw runs your OpenClaw gateway and control surface on Google Cloud. Pick the hosting lane that matches cost vs. isolation — both are managed by us.
Stateless Go services scale on demand; Postgres uses a compact Cloud SQL instance. Best when you want pay-for-what-you-use economics without a dedicated cluster control plane.
1 vCPU · 512Mi per service · min instances 0 (raise for steady traffic)
Dedicated GKE Autopilot namespace with platform-enforced limits, Envoy Gateway-style routing, and the same Helm chart we operate in production. Higher baseline cost, negotiated limits, compliance onboarding.
Dedicated namespace · Helm-managed data plane · sales-assisted onboarding
Everything you need to run OpenClaw in production — without building the control plane yourself.
Logical isolation on Cloud Run; dedicated Kubernetes namespace when you purchase the enterprise lane.
Metered inference proxy with BYOK or platform keys — no hidden markup on model calls.
Create, modify, and tear down OpenClaw runtimes from the admin console or Connect API.
Orchestrator provisions GKE namespaces (oc-customer-name) or Cloud Run services from the same control plane.
Enterprise add-on for durable multi-agent workflows — always-on daemons, gates, and auditable state.
Connect over HTTP RPCs from proto/openclaw/v1 — documented in the customer portal with OpenAPI.
LabClaw, CodeClaw, ShieldClaw, PlanClaw, and the rest represent the roles we automate — from experiments to production isolation. Same characters, serious infrastructure.
Ready to deploy? Create a deployment · Sign in
