AWS
FinOps — savings recommendations
AI-written advice on what to stop paying for, based on your real inventory and spend.
Where Cost tells you where the money goes, FinOps suggests what to do about it.
How it works
We read your actual inventory — instances, disks, snapshots, elastic IPs, NAT gateways, databases — alongside what each service is costing, and produce recommendations per service in plain English with the estimated saving.
Results are cached, so re-opening the page is instant. Refresh to regenerate after you have changed something.
The usual findings
| Finding | Why it happens | Typical saving |
|---|---|---|
| Unattached disks (EBS volumes) | An instance was terminated, its disk was not | Small each, large in total |
| Old snapshots | Backups nobody ever deleted, accumulating for years | Often significant |
| Unassociated Elastic IPs | AWS charges for reserved-but-unused public IPs | Small but pure waste |
| Idle NAT gateways | Expensive per hour, sometimes forgotten in an unused VPC | Large |
| Oversized instances | Provisioned for a peak that never came | The biggest single item, usually |
| Instances running out of hours | Dev and staging running at 3 am on a Sunday | Up to 70% of their cost |
| Storage class | Old S3 objects on the most expensive tier | Moderate |
Before you act
Read each recommendation properly. Automated advice does not know your business.
- An "idle" instance may be your disaster-recovery standby.
- An old snapshot may be the only copy of something you are legally obliged to keep.
- An "oversized" instance may be sized for a month-end run.
Sensible order:
- Delete genuinely unused things — unattached disks, unassociated IPs. Safe and immediate.
- Review old snapshots with whoever knows what they are.
- Right-size after looking at real metrics over at least two weeks.
- Schedule non-production shutdowns — usually the biggest win for the least risk.
Doing it
You can act in the AWS console, or ask the assistant: "delete the unattached EBS volumes in ap-south-1". It will list them first and ask for approval — and deleting cloud resources always asks, even with auto-approve on. The safety switch.
Operators cannot delete AWS resources; an Owner or Admin must. See Roles.