CloudPilot AI
Interactive tools

THE CLOUDPILOT AI TOOLKIT

See how optimization works.

Interactive models for Java runtimes, memory, and Kubernetes.
Explore the mechanics behind each recommendation.

Explore the tools

6 tools
From runtime signals to resource decisions
01Runtime sizing

Java Runtime Evaluation

Turn Heap usage, GC pressure, and stability into coordinated Xms, Xmx, and container-memory recommendations.
Illustrative container-memory layers: Heap, Non-Heap, and safety buffer. Xms and Xmx coordinate Heap boundaries.Container memoryJVMSafety bufferNon-HeapHeapXmsXmx
Java · JVM · KubernetesOpen tool
02Capability comparison

Kubernetes Optimization Comparison

Explore CloudPilot AI, Kubecost, Crane, OpenCost, and Amazon EKS Auto Mode across five capability areas.
Five comparison dimensions: workload, Java runtime, infrastructure, delivery, and pricing. No scores or vendor capability claims are shown.WorkloadJavaInfraDeliveryPricing
Kubernetes · FinOpsOpen tool
03Memory tiers

Memory Extension

Keep hot pages in RAM. Move eligible cold pages through cache and disk with an observable recreate workflow.
Illustrative memory flow: hot pages stay in RAM; eligible cold pages move through cached memory to disk.Hot pagesEligible cold pagesRAMCacheDiskRetainReclaim when eligible
Kubernetes · MemoryOpen tool
04On-demand loading

Hermes Lazy Loading

Mount images using SOCI metadata. Fetch only the layer spans a workload actually reads.
Illustrative image-layer spans: a SOCI index identifies spans, and only requested spans are fetched.Image layer spansSOCI indexFetched on readNot yet requested
Container images · SOCIOpen tool
05Startup resources

Java Resource Optimization

Coordinate Heap, Memory Request, and Limit. Add CPU only during startup with ResourceStartupBoost.
Illustrative ResourceStartupBoost curve: CPU is added during startup, then returns to its baseline. No measured values are shown.CPU allocationStartup boostBaselineStartupSteady state
Java · Memory · CPUOpen tool
06Batch jobs

Job NodePool Optimization

Let batch jobs wait for freed capacity and prefer Active nodes to reduce burst overprovisioning, fragmented long-job tails, and slow node cleanup.
Traditional NodePoolJob NodePoolWait → reuse
Kubernetes · Batch jobsOpen tool

Conceptual previews, not live measurements. Open a tool to explore the full model.