What this prompt does
This prompt turns an AI assistant into a structured AWS cost-optimization consultant. Instead of vague "reduce your bill" advice, it walks the model through eight ordered phases — quick wins, right-sizing, Savings Plans, Spot, storage, data transfer, governance, and a prioritized action plan. You feed it your [monthly_spend] and [account_count], and it frames every recommendation against that scale, so a $25,000-a-month, four-account estate gets different priorities than a small single-account setup.
The structure works because it mirrors how a real cost review actually runs: you chase orphaned resources and idle load balancers first (cheap, fast, no risk), then move into right-sizing [primary_services] where you trade a little analysis for steady savings, and only then commit to Savings Plans at a [coverage_target] you can defend against baseline usage. The [waste_categories], [spot_eligible], and [s3_buckets] variables keep the output anchored to your environment rather than generic theory, and the final phase forces the model to rank everything by estimated monthly impact.
When to use it
- A monthly bill review lands and leadership wants real dollars saved, not just a report
- You inherited an AWS estate and suspect waste from old deployments or oversized dev environments
- You are deciding between Compute, EC2 Instance Savings Plans, and Reserved Instances and need a coverage recommendation
- You want a Spot diversification strategy for batch, CI/CD, or dev workloads without risking interruptions
- S3 storage classes and lifecycle policies have never been tuned across your
[s3_buckets] - You need cost governance — AWS Budgets alerts at
[budget_thresholds]and a[tag_strategy]
Example output
Expect a phased report: a quick-wins list with named waste categories, a right-sizing table showing before/after instance types and projected savings, a Savings Plan recommendation with a coverage percentage and rationale, Spot and storage sections, and a closing prioritized action plan that ranks each item by estimated monthly savings and effort. The action plan is the part you hand back to stakeholders.
Pro tips
- Set
[monthly_spend]and[account_count]accurately — the model calibrates which optimizations are worth the effort to your actual scale - Be specific in
[waste_categories]: naming "oversized dev environments left running" pulls sharper findings than "general waste" - Pick a
[coverage_target]you can commit to long term; over-covering with Savings Plans locks in spend you may not use, so I keep it near baseline, not peak - List only genuinely interruptible workloads in
[spot_eligible]— putting stateful production services there invites trouble - The model cannot see your actual Cost Explorer data, so treat its dollar figures as estimates and validate against your real usage before acting
- Iterate phase by phase: run the quick-wins section, implement, then come back for right-sizing rather than trying to action all eight at once
