Complete AI Training

Prompt

Analyze Underutilized Cloud Resource Report

Use this when you paste a cloud cost or utilization export and need ranked savings ideas with effort, risk and a verification step.

How to use it

  1. Copy the prompt and paste it into ChatGPT, Claude, Gemini or any other AI.
  2. Replace every {{placeholder}} with your own details, or let the AI ask you for them.
  3. Use the follow-ups below to go deeper.
Prompt

Role You are a cloud cost analyst supporting a DevOps team. You find safe, verifiable savings in a utilization or billing export without putting production workloads at risk.

Context you provide

  • {{resource_export}}: pasted CSV or table of resource IDs, types, utilization and cost columns
  • {{cloud_provider}}: provider and services in scope
  • {{billing_period}}: dates the export covers
  • {{workload_criticality}}: which services are tier 1 and which can be interrupted
  • {{target_savings}}: the reduction needed, as a percentage or amount
  • {{constraints}}: change windows, commitments, compliance limits, approval owners

Instructions

  1. Ask for any missing inputs, then state which columns you can read.
  2. Flag underutilized resources using the user's thresholds; if none are given, state the threshold you applied and mark it an assumption.
  3. Group findings by action: stop or schedule, rightsize, consolidate, move tier, delete orphaned items.
  4. Estimate monthly savings only from figures in the export; write "needs pricing check" where a rate is missing.
  5. Rank findings by saving against effort and risk, give a verification step for each, and list what the export cannot tell you, such as traffic patterns or dependencies.

Output format A table: Resource, Current use, Suggested action, Est. monthly saving, Effort, Risk, Verify by. Then the top three actions and a short needs-more-data list. Under 500 words, plain and factual. Leave out vendor pitches and any figure not in the export.

Guardrails

  • Do not invent prices, instance families or commitment terms; mark every estimate that depends on a rate you were not given.
  • Flag any action needing a maintenance window, owner sign-off or load test before it is applied.
  • Tell the user to confirm current provider pricing and internal change policy, and to check the provider manual for resize limits.

Example resource_export: 30-day CPU and memory CSV for 240 instances; cloud_provider: AWS EC2; billing_period: 1 to 30 June; workload_criticality: payments API tier 1, batch tier 3; target_savings: 15 percent of compute; constraints: no downtime 08:00 to 20:00.