CAST AI is a Kubernetes automation platform that replaces the default Cluster Autoscaler with an ML-driven provisioner that actively manages node selection, instance types, and spot capacity. Rather than only adding nodes when pods are pending, CAST AI continuously rebalances the cluster: live-migrating pods between nodes, switching instance families as pricing shifts, and placing workloads on spot with sub-minute on-demand fallback when interruption signals arrive.
Write access to the cloud account is required for autonomous actions. A monitoring-only mode provides cost visibility and recommendations without executing changes.
Supported clouds: AWS, GCP, and Azure. Free tier: monitoring and recommendations only. Growth and Enterprise tiers enable autonomous actions; pricing not publicly listed.
Key Features
ML-driven node provisioner: replaces Cluster Autoscaler with active bin-packing, instance-type selection, and spot migration — takes provisioning actions rather than waiting for pending pods to trigger scaling
Live migration: moves stateful and stateless pods between nodes without downtime during cluster rebalancing, enabling continuous optimization without service disruption
Spot management with interruption prediction: places workloads on spot capacity and monitors pool signals for impending reclamation, migrating to on-demand within approximately 60 seconds
Cost analytics: calculates per-namespace, per-workload, and per-node cost attribution with CPU and memory rightsizing recommendations based on actual utilization patterns
Monitoring-only mode: provides cost visibility, savings projections, and rightsizing recommendations without cloud write access — a staged adoption path before enabling autonomous actions
EKS, GKE, AKS, and self-managed Kubernetes support on AWS, GCP, and Azure; Terraform provider available for cluster configuration
Integrations
8 total
ci / cd
TerraformHelm
cloud
AWSGCPAzure
orchestration
Kubernetes
monitoring
DatadogPrometheus
Pricing
3 tiers
Free (Monitor)
Free
Free tier: Unlimited clusters, cost monitoring only — no optimization actions
Growth
—
$1,000/month + $5/CPU/month for full autonomous optimization
