Automated Resource Tuning for Kubernetes
Fewer nodes.
Fewer OOM kills.
No manual tuning.
Everything sized perfectly, laid out efficiently and within your policies.
What Kubex does for Kubernetes
-
Container rightsizing
CPU and memory requests and limits set from real usage, for every container.
- Requests and limits tuned together
- Undersized workloads raised before anything is trimmed
- New containers start at the right size
-
Node optimization
The right node types and counts for the workloads you actually run.
- Works with Karpenter, picking the instance types it provisions
- Managed node groups and Cluster Autoscaler too
- Freed capacity turns into fewer nodes
-
Bin packing
Long-running pods placed so nodes fill evenly and empty ones can go.
- Consolidates without breaking disruption budgets
- Respects affinity, taints and topology rules
- Runs on a schedule you set
-
HPA tuning
Target utilization and replica bounds tuned with rightsizing, so the two work together and workloads avoid saturation and idle replicas.
-
JVM tuning
Java heap sized together with container memory, so JVM workloads stop over-requesting memory without risking OOM kills.
Real results in leading enterprises
-
Global pharmaceutical leader
$585K a year
907 cores and 8.8 TB reclaimed in two weeks
-
Global media and entertainment leader
$1.2M back
6,000 cores removed across the estate
-
Global enterprise software leader
$140K a year
From the first Kubernetes cluster alone
-
Guardrails on every change
If a check fails, nothing moves.
- HPA awareness
- ResourceQuota
- LimitRange
- Node headroom
- PodDisruptionBudgets
- Readiness before and after
- Protected namespaces
-
Built for platform teams
Installs with one Helm chart.
- Zero-downtime, in-place resize on Kubernetes 1.33+
- New pods right-sized at admission
- Policies as CRDs, GitOps and RBAC ready
- Pause any workload, or allow upsize-only or downsize-only
- Deployments, StatefulSets, DaemonSets, Jobs, CronJobs and Rollouts
Get more from the Kubernetes stack you already run
No rip and replace. Kubex gives your autoscalers accurate requests, so they scale efficiently.
-
Karpenter
Picks smaller, better-fitting nodes and consolidates further when pod requests are accurate.
-
HPA
Every replica it adds inherits a right-sized template, so scale-out doesn’t multiply waste.
-
VPA
Kubex is an alternative to VPA, with better sizing from long-term workload models and smarter resizing that works alongside HPA.
-
Cluster Autoscaler
Fewer unneeded scale-ups, tighter packing and faster scale-down.
-
KEDA
Event-driven replicas, including scale-to-zero, launch at the right size from the first pod.
-
Argo CD and Flux CD
Pods are adjusted at admission, not in Git, so your sync never sees drift. No flood of PRs to merge before recommendations take effect.
Agentic harness
The market leading agentic & deterministic engine accessed from where you work.
The Kubex Agent works through the deterministic engine and inside your policies, so every answer is grounded and every action is governed.
-
Agent
One Kubex Agent. Ask it questions, approve its changes, or let it act on its own.
-
Skills
What the agent knows how to do: rightsizing, node selection, bin packing, HPA and JVM tuning. Pull the same skills into Claude, Cursor or your own tools.
-
Tasks
Scheduled prompts that put skills to work on your clusters. Turn on the ones that ship, or write your own.
-
Proposals
Every change the agent, skills and tasks find comes forward as a proposal. Review and approve them for automation.
Example tasks
Turn these on, or build your own.
-
Size new containers
Gives new deployments a sensible starting size. -
Scale ahead of daily peaks
Adjusts cyclical workloads before the peak arrives. -
Pre-warm nodes
Adds nodes before known load arrives. -
Flag OOM risk
Finds workloads close to their memory limits. -
Report savings by team
Sends a weekly summary by namespace. -
Your own task
Write it as a prompt, pick the skills, set the schedule.
Runs on any Kubernetes








