Skip to main content
Evidence on every finding

Savings you can
audit.

Most cost tools hand you a number and ask you to trust it. ZopNight hands you the monitoring data that triggered the rule, the provider rate the saving was priced from, and a verdict measured after you apply. Where a figure cannot be proven, the answer is "unknown", not a rounded guess.

8finding categories
6verification probes
1-hopblast radius before apply
0fabricated discounts

Detect, price, check, then prove it.

The last step is the one most tools skip.

01

Detect

A registered rule catalog runs on a cron across AWS, GCP, Azure, Snowflake, Databricks and Kubernetes. Idle, rightsizing, schedule, compliance, discount, orphan, security and reliability. Findings are pre-computed, so the page opens instantly instead of calling clouds while you wait.

02

Price it honestly

Commitment savings come from the resource's real on-demand versus committed rate, read from the provider's own pricing feed, for each term we can price. A stale or nonsensical rate is dropped. An unpriceable term returns "savings unknown" rather than a number someone made up.

03

Check before you apply

One-click remediation runs a precondition, then an admin approval step for anything destructive, then the cloud action, then a validation. The blast radius shows every connected resource first.

04

Verify afterwards

We re-read the exact resource key the rule fired on and compare it to a baseline frozen when the finding was raised. Status is what you said you did. Resolution state is what we measured. They are two columns on purpose.

What it refuses to do.

A cost tool is only worth having if you can trust it when the news is bad.

No assumed uptime
A commitment saving needs measured uptime from activity data. Without it the answer is "savings unknown". Never an assumed 100 percent.
No credit for hours not run
A no-upfront commitment bills every clock hour of its term. The saving is bounded by measured run cost, so a discount is never credited against spend a commitment cannot touch. A break-even gate suppresses the recommendation entirely when the margin is under 10 percent.
No silence as proof
"The rule stopped firing" is not accepted as evidence of a fix. A narrowed IAM role and a genuine remediation produce the same silence. Rules declare the exact key to re-read.
Abstaining is allowed
Where no honest verdict exists, the recommendation says so. Six probes cover existence, power state, shape, duty cycle, setting and rate. A rule fitting none of them abstains by construction rather than guessing.

See what your estate is actually wasting.

Connect a read-only role and the first pass runs on your own accounts. Every figure comes with its evidence.

Multi-cloud automation· Production-ready in 30 min· SOC 2 · ISO 27001· 20–60% off the bill, first month· 4 platforms · 1 console· Multi-cloud automation· Production-ready in 30 min· SOC 2 · ISO 27001· 20–60% off the bill, first month· 4 platforms · 1 console·