Recommender-sourced savings · Google Cloud
One run turns your real Google Cloud bill into an interactive savings report — trends, forecasts, anomaly alerts, and the exact resources to fix. Headline savings come from Google's own Recommender API, so they already reflect your sustained and committed use discounts. No subscriptions required.
Trusted by cloud engineers, FinOps practitioners & consultancies
The Google Cloud FinOps Toolkit
An always-on service that runs the whole loop — or one-time downloads you own forever.
Always-on and multi-cloud. One launch scans your Google Cloud — and Azure, and AWS — read-only in your own environment, then an AI analyst reasons over your real findings and tells you what to fix first. It runs all four rungs.
Cancel anytime · keep what you downloaded
Scans one or more Google Cloud projects across 10 modules: Google's own cost Recommenders, orphaned disks/IPs/snapshots, committed use discounts, Cloud Storage class & lifecycle, networking and Cloud NAT, billing budgets, Vertex AI FinOps-for-AI, and cost anomaly detection with webhook or Pub/Sub alerts. Interactive HTML report with the FinOps Insights band, plus a JSON export.
Secure checkout · payments by Paddle.com
Three toolkits in one: a GCP Cloud Cost Governance Framework, a Migration Readiness Kit (assessment, scoring, 6 Rs, stakeholder report), and a GCP Labelling & Naming Standard with an organisation-policy starter.
Secure checkout · payments by Paddle.com
A one-page monthly cost-review checklist for Google Cloud — billing reports, budgets, Recommender, CUD/SUD coverage and the waste sweep. Aligned to the Google Cloud Architecture Framework. The fastest way to see the CloudFinOpsKit approach.
A board-ready report built for Google Cloud, from your real billing data.
A plain-English walkthrough: what gets installed for you, how to sign in, and what to click. The launcher installs Python, the Google Cloud CLI and every package for you, into a private folder that leaves nothing behind.
Industry data puts wasted cloud spend near 30%. A disciplined practice recovers 20–35%.
Illustrative estimate at a 24% benchmark recovery rate — not a quote. The tool reports figures from your own Google Cloud billing data and Recommender output.
Everything above in one download: the GCP Tool + the Governance, Migration & Labelling Pack, the FinOps Dashboard Pack and the 150-prompt AI Prompt Library.
$60 $50
Run more than one cloud? The All-Cloud FinOps Bundle is every toolkit, doc and workbook for Azure, AWS and Google Cloud in a single download — $150 of toolkits for $115.
See the Azure toolkit → See the AWS toolkit → Get all three clouds — $115Headline savings come from Google's own Recommender API — never in-tool guesswork. Idle findings are metric-confirmed, and the tool never infers idle from missing data.
Recommender computes against your actual usage and pricing, so sustained and committed use discounts are already baked in. No double-counting a discount you already have.
A dedicated Vertex AI module: per-model token analysis, output-token bloat, cost per successful output, and self-hosted endpoints holding replicas that serve nothing.
Runs on your existing read-only GCP access via ADC or a service account. No agents, no infrastructure, no recurring fees. Costs the same to assess 5 resources or 50,000.
A 150-prompt GCP FinOps AI Prompt Library ships with the tool — CUD and SUD modelling, BigQuery slot economics, GKE bin-packing, anomaly RCA, and consultant-grade multi-angle audits for ChatGPT, Gemini or Claude.
The one-page monthly Google Cloud cost review. No spam.
Thanks! Your free checklist is downloading now.
Headline savings come from Google's own Recommender API, which computes projected savings against your actual usage and pricing — sustained use and committed use discounts are already reflected. Anything the tool derives itself is labelled as derived on the finding, so you always know which is which.
Recommended, not required. Without it the tool still runs; the spend trend and forecast are skipped rather than estimated. We'd rather show you less than show you a guess.
Read-only. roles/viewer plus roles/recommender.viewer covers the scan; add roles/billing.viewer for budgets and BigQuery read on the billing-export dataset for spend trend. The scan never changes anything in your projects.
Yes — pass several project IDs with --projects, or let it use your Application Default Credentials project. Findings carry the project so allocation works out of the box.
Yes. Per-model token analysis for Vertex publisher models — output bloat, token spikes, zombie deployments, cost per successful output — plus self-hosted custom endpoints holding replicas that serve nothing. One honest limit: self-hosted endpoints emit no token metrics to any cloud's monitoring, because the counts live inside your model server. For those the tool reports cost and utilisation and says so, rather than implying coverage it doesn't have.
Python 3.9+ with pip install -r requirements.txt, or the included Dockerfile if you'd rather not manage an environment. A scheduled-run template ships with it for monthly runs and anomaly alerting.
Yes. Customise and reuse them across your own projects and clients. You may not resell the toolkits as-is.
Yes — a 14-day money-back guarantee via the support link in your Paddle order-confirmation email.
