Skip to content
DIGIUSHER
Book my 15-min call
Domain editions

Platform reference

One engine. Five domains. Each with the depth its terrain deserves.

A domain edition is the connector set, waste-scenario library and signature value metric DigiUsher applies to one class of technology cost.

The platform capabilities underneath are deliberately generic: they work identically on any cost row. What differs per domain is where the data comes from, where the waste hides, and which ratio its owners actually argue about.

Book my 15-min discovery call → Read the TVR framework
Cloud
Cost per workload
AWS · Azure · GCP · OCI
AI
Cost per merged PR
models · agents · GPU
Data
Cost per pipeline run
Databricks · Snowflake · Atlas
K8s
Cost per service
EKS · AKS · GKE · on-prem
The five editions

Every domain, in the vocabulary its owners already use.

Cloud cost management · AI and LLM cost governance · Kubernetes cost optimization · data platform economics · on-premise and SaaS cost allocation

Cloud Edition

AWS · Azure · GCP · OCI · Alibaba Cloud

Native billing ingestion from every hyperscaler with amortized, effective and billed cost reconciled, so the number finance sees and the number engineering sees are the same number. Rightsizing, commitment management, storage economics and network waste run as scenario libraries; scheduled start/stop alone typically captures around 70% of non-prod compute cost. €1M realized in 45 days for a European enterprise on exactly this discipline.

Signature value metric Cost per workload Turns migration, consolidation and re-platforming debates from opinion into arithmetic.
Cloud edition in full →

AI Edition

Models · Agents · GPU

Three cost surfaces, one discipline: managed platforms (Bedrock, Vertex, Azure AI), direct providers (Anthropic, OpenAI, Gemini, Copilot) and engineering agents (Claude Code, Cursor, Codex, Windsurf, Devin and more), plus the GPU fleet underneath, down to MIG partition. Agent spend is joined to delivered work via commit attestation; prompts and responses are stripped at ingestion by architecture. Agentic workloads consume 5–30× the tokens of a chatbot per task, and 98% of FinOps teams now manage AI spend.

Signature value metric Cost per merged PR Whether engineering-AI spend compounds into shipped software, with traced coverage as its honest denominator.

Data Cloud Edition

Databricks · Snowflake · MongoDB Atlas · BigQuery

System-table-level ingestion, not invoice summaries: DBUs by cluster and job, credits by warehouse and query, Atlas tiers by project. The AI workloads growing inside each platform, such as Mosaic AI, Cortex and Vector Search, are separated as their own categories rather than buried in a platform total.

Signature value metric Cost per pipeline run A pipeline whose run cost doubled after a schema change is caught the week it happens, not at renewal.
Data Cloud edition in full →

Kubernetes Edition

EKS · AKS · GKE · OpenShift · on-prem clusters

Node, pod, cluster, daemonset, replicaset, deployment and namespace level truth via native cloud mechanisms and label ingestion, cloud and on-premise in the same view. Rightsizing runs across five dimensions: GPU (MIG matching, time-slicing, bin-packing), CPU, memory, storage and network. CPU-only rightsizing optimizes the cheapest resource on an AI-era node; DigiUsher sees all five. Idle capacity gets its own line, because headroom nobody owns is headroom nobody reduces.

Signature value metric Cost per service Efficiency outliers visible across the service catalog; cost per request for platform targets.
Kubernetes edition in full →

On-Premise, SaaS & ANY

VMware · mainframe · subscriptions · future domains

CAPEX amortization, ISV licenses and facilities normalized into FOCUS beside cloud spend, giving honest cloud-versus-data center unit economics with mainframe MIPS included. SaaS is allocated by actual seat activity rather than seat count, which is how shelfware surfaces. And ANY: a cost domain that does not exist yet lands in the same schema and inherits every platform capability on day one. AI arrived exactly this way. A new domain costs a connector, not a re-platform.

Signature value metrics Cost per VM · per active seat · per MIPS Which subscriptions are shelfware; what the data center honestly costs against cloud.
On-prem, SaaS & ANY edition in full →

Supported today native billing ingestion

  • Amazon Web Services AWS
  • Microsoft Azure Microsoft Azure
  • Google Cloud Google Cloud
  • Oracle Oracle Cloud (OCI)
  • Alibaba Cloud Alibaba Cloud

Coming soon on the connector roadmap

  • Huawei Cloud (coming soon)
  • Tencent Cloud (coming soon)
  • OVHcloud (coming soon)
  • DigitalOcean (coming soon)
  • Hetzner (coming soon)

Every provider lands in the same FOCUS schema, so a multi-cloud total is directly comparable rather than approximately aligned. A new provider is a connector, not a platform release, which is why the roadmap above is measured in weeks rather than versions.

Why one engine matters

Metrics only compose when the domains share a schema.

Point tools each compute their own fraction of the answer. A platform holding the whole estate in one FOCUS-conformant store can compute the whole number: AI-inclusive cost per customer, in a single formula, with one auditable lineage.

Domain metrics composing into one number Five domain metrics, cost per workload, cost per merged pull request, cost per pipeline run, cost per service and cost per active seat, converge through a shared FOCUS schema into one composite figure: AI-inclusive cost per customer. Cloud cost per workload AI cost per merged PR Data Cloud cost per pipeline run Kubernetes cost per service On-prem & SaaS cost per VM · seat ONE FOCUS SCHEMA same columns, same lineage COMPOSITE $0.94 AI-inclusive cost per customer ↓ 12% QoQ Point tools each compute one row of the left column. Only a platform holding the whole estate can compute the right.
The composition property: because every domain lands in the same schema, its metric can be combined with the others in a single formula with one auditable lineage.
  1. 01

    One ingestion path

    Every domain lands in FOCUS at the moment of ingestion, with no proprietary intermediate schema, no translation lag, and roughly 30% lower data-processing cost.

    See
  2. 02

    One allocation pipeline

    Auditable sequenced chargeback runs identically over a Snowflake credit, a GPU hour and a SaaS seat. Attribution quality is graded the same way in every domain.

    Attribute
  3. 03

    One governed change path

    A recommendation becomes a pull request against the repository that owns the resource, whatever the domain. Human approval applies it. Nothing changes silently.

    Optimize
  4. 04

    One value language

    Signature metrics per domain, composing into cost per customer. Savings tracked identified → applied → verified-realized, on the same ledger across the estate.

    Realize
Questions about coverage

Direct answers on domains and connectors.

What is a DigiUsher domain edition?

A domain edition is the connector set, waste-scenario library and signature value metric that DigiUsher applies to one class of technology cost. All five run on the same FOCUS-native engine and the same See, Attribute, Optimize, Realize stages, so adding a domain does not mean adding a platform.

Which cost domains does DigiUsher cover?

Five: Cloud (AWS, Azure, GCP, OCI, Alibaba Cloud), AI (managed platforms, direct model providers, engineering agents and GPU), Data Cloud (Databricks, Snowflake, MongoDB Atlas, BigQuery), Kubernetes (EKS, AKS, GKE, OpenShift and on-premise clusters), and On-Premise, SaaS and ANY future domain.

What is a signature value metric?

The one ratio that tells a domain's owners whether spend is compounding into output: cost per workload for cloud, cost per merged PR for AI, cost per pipeline run for data platforms, cost per service for Kubernetes, and cost per VM, active seat or MIPS for on-premise and SaaS. Because every domain shares one schema, these compose into an AI-inclusive cost per customer.

What happens when a new cost domain appears?

It lands in the same FOCUS schema and inherits every platform capability on day one: allocation, unit economics, anomaly detection, forecasting and governed workflow automation. AI itself arrived this way. A new domain costs a connector, not a re-platform.

44 more answers: TVR, FinOps, AI cost, Kubernetes, allocation and vendor selection →

Start with the domain that worries you most.

First insight within 48 hours of connecting a source. Extend across the estate when you are ready.

Book my 15-min discovery call → Take the product tour