Cloud, Kubernetes and AI spend.
Every cost has an owner.

UnitVane covers every bill: AWS, Azure, Google Cloud, Oracle Cloud, Kubernetes, AI and SaaS spend in FOCUS, allocated to your business units and teams, with budgets, anomalies, verified savings and a built-in BI studio. KubeOn goes inside clusters with a read-only agent, pod-level allocation, change proposals, optional write access and the GPU optimizer. UnitVane AI covers model spend with cost per 1M tokens and a managed LLM gateway. Start with one and add the others; they can be quoted on one agreement.

UnitVane is SaaS only. We run all three products, including upgrades, backups and monitoring. You grant read access to billing exports, install the read-only agent in clusters you want UnitVane to cover, and point apps at the managed gateway if you use UnitVane AI. There is no self-hosted edition.

The agent's ClusterRole grants get, list and watch on nodes, pods, namespaces, workloads, volumes, services, quotas and storage classes, plus read access to metrics.k8s.io. It never reads Secrets or ConfigMaps and has no write verbs. It runs as an hourly CronJob with no Service, no Ingress and no inbound port.

Only if you turn it on. By default KubeOn reads, and each saving is a change proposal your pipeline ships. You can also grant a separate controller write access, cluster by cluster and feature by feature, for rightsizing, node pools and Spot fallback, node autoscaling and live migration. Billing exports are never changed. UnitVane AI buys or renews reservations only on pools whose policy you set to do so, and every purchase waits for a named approver.

Billing from AWS (CUR 2.0), Microsoft Azure (Cost Management exports), Google Cloud (billing export to BigQuery), Oracle Cloud and any FOCUS 1.x file. Clusters on EKS, AKS, GKE, OKE, OpenShift, Rancher, VMware Tanzu, k3s and kubeadm. Data centers are priced with rate cards or amortized hardware cost.

Your billing exports stay in your accounts, and UnitVane reads them through an identity that cannot change them (on AWS, its only write is Athena query results). The cost data UnitVane keeps is stored in the region you choose. European data stays in a European region.

For cost management, very little. The agent collects infrastructure metadata (nodes, workloads, labels, requests and usage) and the hub reads billing line items; neither reads Secrets, ConfigMaps, logs, application data or request payloads. The UnitVane AI gateway is different: it reads every prompt and response in transit to route it, so any personal data in them passes through it. It logs metadata only by default, can mask email addresses, phone numbers and card numbers before a request leaves it, and can pin routes to EU providers and regions.

Cluster inventory appears within the first hour: the agent runs on the hour and the hub ingests at 20 past. Allocated costs follow as soon as your billing data is in, within the same hour when your export already has data. A brand-new AWS CUR export can take up to a day to deliver its first file. On first connect UnitVane reads 180 days of billing history and allocates it.

Allocation starts from the billed, net amortized cost of each node, so Savings Plans, reservations, credits and discounts are already in it. Each pod is charged on max(request, usage), idle and shared cost stay in their own columns, and anything UnitVane cannot attribute is listed as unallocated with its resource IDs.

Yes, with UnitVane AI. It measures effective cost per 1M tokens across Azure OpenAI, Amazon Bedrock, Google Vertex AI, OpenAI and Anthropic, sizes provisioned throughput, and offers a managed LLM gateway that routes requests and enforces team budgets. It shares owners, budgets and sign-in with KubeOn, and its token spend reaches UnitVane as FOCUS rows.