CloudFinOpsKit

Free · instant download · no card

Your AI bill doubled. Which model, which deployment, which feature?

AI is billed in tokens, and your cost report shows one line per deployment. A deployment costing $4,000 a month looks the same whether it serves real traffic or replays a bloated prompt to nobody. This free checklist gives you the 12 checks that find token waste on Azure OpenAI, Amazon Bedrock and Vertex AI, in the order that reclaims money fastest.

Covers Azure OpenAI & AI Foundry Amazon Bedrock & SageMaker Vertex AI

What's on the checklist

Pure waste first (nothing to trade off), then capacity, then token efficiency. The model right-sizing check comes last because it's the only one that needs a quality test.

Pure waste

Delete, don't debate

  • Zombie deployments: no requests in 30 days, still billing
  • Idle SageMaker, Azure ML & Vertex endpoints
  • Notebooks left running overnight
Often the biggest line Capacity you aren't using
  • Under-used Azure OpenAI PTUs
  • Idle Bedrock Provisioned Throughput
  • Paying for reserved and pay-as-you-go at once
Token efficiency

One parameter to one refactor

  • Output bloat: 800+ tokens per reply
  • Context bloat & a cold prompt cache
  • Premium models on simple tasks

Plus: failed-call overhead, cost per successful output, and token anomaly watch so it doesn't creep back.
Get the free checklist →

Who it's for

If AI features are shipping faster than anyone can explain the bill, this is for you.

CTOs & engineering leads
  • Answer "why did the AI bill double?" with numbers
  • Know which deployments are worth what they cost
Platform & ML engineers
  • Exact metrics to pull on each cloud
  • Fixes in payback order, not a theory lecture
FinOps & finance
  • A monthly AI cost review with a scorecard
  • Unit economics finance can follow

Prefer it done for you in about 10 minutes? The read-only assessment runs all 12 checks automatically and prices every finding from your real bill. Or meet the FinOps Agent (from $15/mo).