Varcio FinOps Copilot

In Plain Words

What this product does, explained without assuming you work in cloud engineering — plus a glossary of every cloud and finance term used in this documentation.

This page is for readers who do not work in cloud engineering: a finance lead, a founder, a new joiner, anyone who has to understand what this product does without first learning what a Savings Plan is.

Nothing here is simplified to the point of being wrong. It is the same product, described without the vocabulary.

The problem, in ordinary language

Renting computers by the hour is easy to start and very hard to stop.

Any engineer can create something that costs money, in seconds, without asking anyone. Nobody has to switch it off. The bill arrives a month later as a single enormous number with no explanation of who caused it or what it bought.

So three things go wrong at once:

Finance cannot forecast

The number moves for reasons nobody can name, so next quarter's budget is a guess dressed up as a plan.

Engineering cannot act

The bill does not say which of the thousands of things they run is responsible, so there is nothing specific to fix.

Leadership cannot decide

Nobody can say whether the cost is reasonable for what the business actually got.

What the product does about it

It connects to your cloud accounts and reads the bill, then does five things with it.

It explains the bill

What you spent, on what, and which team caused it — including how much it could not attribute, so the report can be trusted rather than argued with.

That last part matters more than it sounds. A report claiming to explain 100% of the bill gets checked once, found wrong, and never believed again. One that says "89% attributed, and here is the missing 11%" is one a team can be held to.

It finds waste

Resources nobody uses. Resources far larger than the job needs. Discounts you are paying for and not using.

Each one comes with an estimate of what fixing it saves and how much work that is — so you can start with the things that are worth the most and cost the least effort.

It fixes things — but never on its own

Every change is proposed, reviewed, and approved by a person before it runs. Every one is recorded.

This is the part people are most nervous about, and the answer is consistent throughout the product: the software finds and proposes; a human decides.

It warns you early

When spend jumps unexpectedly. When the month is heading over budget. When a commitment is about to expire. When something insecure is also expensive.

It answers questions in plain English

You can ask what changed, why the bill went up, or what to do next — and it answers from your own data, not from generic advice.

That is Apex, and for most non-technical users it is the easiest way into the whole product.

Which parts of this documentation you need

A reading route for non-technical readers

  1. This page, including the glossary below.
  2. Core Concepts — the nine things the product talks about.
  3. Plans & Limits — what each tier includes.
  4. Then only the modules that matter to your job, from the module reference.

Every module page opens with what it is, who it is for, and closes with why it matters. Those parts are written for you.

The section headed How it works is the mechanism — it is there for engineers, and it is safe to skip. Nothing later depends on your having read it.

Glossary — cloud and finance terms

The words this documentation uses that come from cloud computing or cloud finance, explained without assuming either.

Product-specific terms — finding, opportunity, workspace, remediation class — are defined separately in the product glossary.

The basics

TermWhat it means
Cloud providerA company that rents out computers by the hour instead of selling them. This product covers the four large ones: Amazon Web Services (AWS), Microsoft Azure, Google Cloud, and Oracle Cloud.
EstateEverything your company runs across those providers, taken together. When this documentation says "the estate", it means all of it.
ResourceOne rented thing — a server, a database, a disk, an address. You are charged for each one, usually per hour, whether or not anyone is using it.
WorkloadA group of resources that together do one job, such as a website or a reporting system.
RegionWhich part of the world the data centre is in. The same resource costs different amounts in different regions.
KubernetesSoftware that packs many workloads onto shared machines. It makes better use of what you rent, and makes it considerably harder to tell who is spending what.

How you pay

TermWhat it means
On-demandPaying the standard hourly rate with no commitment. The most flexible and the most expensive way to buy.
CommitmentA promise to spend a certain amount for one or three years in return for a discount, often 30–60%. Reserved Instances and Savings Plans are the two common forms. The risk is that you keep paying whether or not you use it.
Spot capacitySpare capacity sold cheaply that the provider can take back at short notice. Good for work that can be interrupted, unusable for work that cannot.
EgressThe charge for data leaving the cloud. Data going in is usually free; data coming out is not, and it surprises people.
CreditsMoney the provider has given you that offsets the bill. While credits cover the bill your payable amount can be zero even though the underlying usage is not — and credits expire.
Blended, unblended, and amortised costThree ways of stating the same bill. Unblended is what you were charged that day; amortised spreads a one-off commitment payment across the months it covers; blended averages rates across accounts. They do not match, and mixing them is a common source of disagreement.

Waste and efficiency

TermWhat it means
WasteSpend that buys nothing: idle resources, oversized resources, disks with nothing attached to them, backups of systems that no longer exist.
RightsizingMaking a resource the size it actually needs to be. Most cloud waste is not unused machines but machines four times larger than the job requires.
Idle or orphaned resourceSomething still running, still being charged for, that nobody is using any more — usually left behind by a project that ended.
Time to live (TTL)An expiry date attached to a resource that was only ever meant to exist for a while. See Lifecycle.
Unit economicsCost expressed per unit of business — per customer, per order, per transaction — rather than in total. A bill that grows 20% is fine if the business grew 40%.

Attribution and accountability

TermWhat it means
TagA label attached to a resource saying who owns it and what it is for. Tags are how spend gets attributed to a team. Untagged resources are the reason most cost reports are argued with.
ShowbackTelling each team what their share of the bill was, without actually charging them for it.
ChargebackThe same thing, except the amount is genuinely billed to that team's budget.
DriftReality quietly moving away from what was agreed — a policy that stopped being followed, a tag that stopped being applied, a commitment that no longer matches what is running.

Control and response

TermWhat it means
AnomalySpend that jumped in a way the previous weeks do not explain. Usually a mistake, occasionally a launch, sometimes a security incident.
ForecastAn estimate of what the month or quarter will finish at, based on what has been spent so far and what usually happens next.
Budget and guardrailA limit you set, and what the product does as you approach it — warn somebody, block a change, or stop a resource.
FindingOne specific problem the product has spotted, with what it would save to fix, how confident the product is, and how much work it would take.
RemediationThe fix itself — resizing the machine, deleting the disk, stopping the cluster. In this product a remediation is always proposed first and executed second.
Approval gateA point where a person has to say yes before anything changes. Nothing that costs money or removes a resource happens without one.
PostureHow exposed the estate is — what is open to the internet, what is unencrypted, what is missing a patch. This product connects posture to cost, because insecure things also cost money.
Service-level agreement (SLA)A promise about how quickly something will be responded to or fixed.
FinOpsThe practice this product supports: getting engineering, finance, and leadership to agree on what the cloud costs, what it buys, and who decides. It is a working practice, not a piece of software.

Where to go next

On this page