Spend Returned to Its Owners
Label-based reattribution moved roughly $21.6K per month of sandbox and shared spend off the central bucket and back to the teams actually spending it, with net-zero splits for shared estates.
CASE STUDY
FinOps
Reporting App
Three clouds. One report everyone reads.
A multicloud cost platform spanning AWS, Azure, and GCP: actuals, budgets, forecasts, vendor spend, and AI costs in one place, framed the way executives actually consume them. It replaced a monthly ritual of exports, spreadsheets, and deck assembly with a system that is always current.
$112.7M
of lifetime cloud spend under management
$4.4M
per month current run rate
33
business units graded A–F
3
clouds in one FOCUS-based feed
01
The problem
Multicloud reporting was a manual production: export from each cloud, reconcile in spreadsheets, screenshot into a deck, repeat every month. The numbers arrived late, disagreed with each other, and got debated instead of acted on. Shared and sandbox spend landed on whoever's account it happened to sit in, so the teams spending the money never quite saw their own bill.
02
The system
One FOCUS-based cost feed covers all three clouds, so every number shares a definition before it shares a page. The landing page is an executive summary organized around the four FinOps Framework domains, each with a red/amber/green status: leadership sees posture in one glance and drills from there. Budgets and rule-based forecast adjustments live in the app, vendor SaaS spend and revenue context sit alongside cloud costs, and the charts page renders the monthly report from live data in brand colors: the deck assembles itself. Behind every view sit five fiscal years of history across roughly 1,400 accounts and projects, 105 subscriptions, and 243 distinct billed services, and the whole platform went from first commit to production in under five months.
03
What makes it different
ACCOUNTABILITY
A Letter Grade Every Business Unit Understands
Each of 33 business units gets an A–F cost scorecard built from five weighted criteria, led by budget adherence, with disaster-recovery spend redistributed to its true owners and only completed months scored. Nobody argues with a spreadsheet tab; everybody understands a D+.
ALLOCATION
AI COSTS
AI Spend, per Engineer
AI coding-assistant usage is metered from telemetry down to cost per user per model: adoption, token burn, and dollars in one view, before the invoice arrives to explain itself.
FORECASTING
Projections Leadership Can Question
Regression-based full-year projection with visible assumptions and rule-based adjustments, so "are we going to make budget" has a number and a why, not a shrug.
04
The stack
AWSAzureGCPFOCUS
ReactNodePostgreSQLAG-Grid
RechartsDockerKubernetesOpenTelemetry
05