Skip to content

Free diagnostic

Find out what your estate is really costing you.

Two weeks of measurement across your AWS bill and your Kubernetes clusters, then a readout: where the money goes, which savings survive scrutiny, and what to do first. No cost, no commitment, and the document is yours whether you hire us or not.

We are not going to promise you a large number.

Automated cost tools sum every finding they produce, count the same dollars twice when two detectors reach them from different directions, and recommend actions you cannot carry out. On the last estate we reviewed, reconciling those overlaps removed roughly three quarters of the headline figure. What was left was small enough to be believed and specific enough to act on — which is the only kind of number worth having.

How that review went

What you get

Three things, and they are all documents

A measured baseline

We read your bill and your clusters instead of modelling them. Every figure in the readout traces back to a line on the bill for a named month, so your own engineers can check it without us in the room.

A reconciled range, not a headline

Findings that reach the same dollars from two directions are counted once. Advice you cannot execute — resizing a node your autoscaler owns — is rewritten as the action you can. What survives is a low and a high end, each with its basis written down.

A sequenced plan you keep

What to do first, what it depends on, and what to leave alone until the fleet stops changing shape — plus an explicit list of what we could not evaluate and what data would close each gap.

How it runs

Four steps, about three weeks, almost none of it yours

  1. 01A 30-minute call

    What you run, what hurts, and whether a diagnostic would tell you anything you do not already know. If it would not, we say so on the call and that is the end of it.

  2. 02Read-only access

    A read-only role in the billing account, and optionally a metrics agent in the clusters. Fourteen days of measurement is what makes percentiles mean something; less than that is guessing with extra steps.

  3. 03We do the analysis

    Off your calendar. Nobody on your side is asked to run queries, export spreadsheets or sit in working sessions while we think.

  4. 04A 45-minute readout, then the document

    We walk through the findings and you push back on them — that conversation usually removes something, which is the point of having it. The written version follows within three working days.

Access

What we need, and what we will never ask for

What we need

  • A read-only role in the billing account — cost and inventory APIs, nothing that can change a resource.
  • Optionally, our metrics agent in the clusters: one deployment, read-only Kubernetes RBAC, no privileged containers.
  • Fourteen days between access and readout, so the percentiles describe a real fortnight rather than a quiet Tuesday.
  • About an hour in total from one engineer who can answer “why is this here?”.

What we never ask for

  • Administrator credentials — not in the cloud account, not in the cluster.
  • Write access to your repositories.
  • A migration onto a platform of ours. There is nothing to adopt.
  • Your source code.

Honest limits

What this is not

Not a dashboard trial

There is no login to forget about after a fortnight. The deliverable is a document, and it keeps working when you close it.

Not a number built to impress a CFO

If the defensible figure turns out to be small, the report says it is small. We would rather lose the follow-on engagement than hand over a figure that falls apart in your first internal review.

Not a security audit

We report what the estate's posture shows us in passing — public buckets, unused keys, groups open to the world — but that is a by-product of the inventory, not a certification, and we will not pretend otherwise.

Not unlimited in scope

One estate: one payer account and the clusters attached to it. Anything larger stops being a free diagnostic and becomes a quoted engagement, which we will tell you before starting rather than after.

Ask for the diagnostic

Write to us with roughly what you run — cloud provider, how many clusters, rough monthly spend. That is enough for us to say whether it is worth doing, and we would rather tell you no by email than on a discovery call.