Find out where your AWS AI budget actually goes.
One Athena query against your own CUR. No AWS access required from us. We score your AI governance maturity across four dimensions and deliver a clear, prioritized action plan within 24 hours.
No cost, no commitment
To run the query and share results
Governance Score and action plan delivered
Never leaves your AWS account
Three Steps. No AWS Access Required.
We provide a single Athena SQL query tailored to your CUR table. You run it in your own AWS environment. Your data never leaves your account.
Export the query output and share with the Cloud Scal3 team. The file contains billing metadata only — no workload content, no application data.
We score your governance maturity across four dimensions and deliver a clear, prioritized action plan within 24 hours. Specific to your environment.
The Query
One SQL statement. Every AI platform in your CUR.
The query pulls a month of Bedrock, AWS Marketplace AI models, and Claude Platform spend from your Cost and Usage Report — grouped by account, product code, legal entity, and IAM principal. Billing metadata only.
SELECT line_item_usage_account_id, line_item_product_code, legal_entity, line_item_iam_principal, SUM(line_item_unblended_cost) AS total_cost FROM your_cur_table WHERE ( -- Native Bedrock line_item_product_code = 'AmazonBedrock' OR -- Marketplace Claude (Anthropic, PBC) line_item_product_code LIKE '%anthropic%' OR -- Claude Platform (CCU billing) line_item_product_code LIKE '%claude%' ) AND line_item_line_item_type = 'Usage' AND year = '2026' AND month = '4' -- adjust to your target month GROUP BY 1,2,3,4 ORDER BY total_cost DESC
line_item_iam_principalThe key field. Populated = attribution enabled. Null or empty = spend is unattributable at workload level.
legal_entityIdentifies Marketplace Claude (Anthropic, PBC) vs native Bedrock — different billing paths, different attribution mechanics.
line_item_product_codeSurfaces all three AI platform paths in a single query — Bedrock, Marketplace models, and Claude Platform.
What We Evaluate
Four dimensions. One Governance Score.
The query results tell us where your AWS AI governance stands today — and exactly what it would take to move from your current state to Level 4: Confident.
IAM Principal Coverage
Is caller identity enabled in your Cost and Usage Report? This single CUR setting unlocks workload-level attribution across your entire Bedrock and Marketplace AI estate — and is the foundation everything else is built on.
What % of AI spend rows have line_item_iam_principal populated?
Attribution Quality
Of the spend with an IAM principal, how much flows through dedicated per-workload roles vs broad shared SSO access? The answer determines how much of your AI bill can be allocated to a specific team, product, and budget today.
What % of principals are dedicated finops-ai-* roles vs AWSAdministratorAccess?
Platform Coverage
Which AWS AI billing paths are active — native Bedrock, AWS Marketplace models, Claude Platform? Each has a distinct attribution mechanism, and each needs to be in scope for governance to be complete.
Which legal_entity and product_code values appear in your CUR?
Model Governance
How many distinct foundation models are in use, across how many accounts? Are any active without an approval record? Understanding your model footprint is the starting point for setting estimates and a weekly governance cadence.
How many distinct model ARNs appear? Which accounts have the most exposure?
Governance Score
Where are you today?
The assessment shows you exactly where you are — and the fastest path to Confident.
No IAM principal. All AI spend is a single line item reviewed once a month.
IAM enabled but broad shared roles. Team-level view only. No workload split.
Some dedicated roles. Partial attribution. No confirmed estimates.
Full workload attribution. Weekly spend cards. Estimate vs actual. Ready to scale.
Monthly review cadence: manageable.
Compute, storage, and network costs change slowly. A monthly invoice review is workable. Drift is gradual and visible in trend data.
Weekly review cadence: essential.
A single code deployment or new prompt template can 10x token consumption overnight. By the time a monthly review surfaces the spike, four weeks of spend have already happened.
What You Receive
Five deliverables. Within 24 hours.
Specific to your environment. Built from your own CUR data.
Your maturity level (1–4) across all four dimensions, with specific evidence from your query results. Not a generic benchmark — scored against your data.
The exact dollar value of ungoverned AI spend in your environment over the assessed period. This is the number that drives the internal business case.
A dimension-by-dimension breakdown showing which governance controls are missing, what they cost you in attribution coverage, and what estimate discipline is possible today.
A specific recommendation for the first workload to onboard — selected from your data based on spend, attribution status, and implementation complexity.
A 30-day roadmap from your current maturity level to Level 4: Confident — including which workloads to onboard first, how to set the first estimates, and how to shift from monthly invoice reviews to weekly spend card approvals.
Take the first step toward Confident AI spend.
Request the assessment and we will send you the Athena query and setup instructions within one business day. You run it. We score it. Your governance roadmap arrives within 24 hours.
or email us directly
support@finopscenter.comFinOps Center deploys entirely inside your AWS estate. Learn more about AI Spend Governance →