What an app run costs, and the consent step

The cost is stated before anything spends

Every analyst app states its cost before you run it, in plain sight under the run button: "About 15 IU for the whole analysis, including the back and forth. You are charged per message." That line appears before you click anything, not buried in a confirmation you have already half-committed to, and it is the same figure the platform checks your balance against, so the number you are shown and the number that gates the run always agree.

The word "about" is doing honest work. Here is how the quote is built, and why the real charge usually comes in under it.

How the quote is derived

An analyst app run is a conversation with a specialist, and Intelligence Units are charged per message you send, not per analysis. A typical run takes about three messages: your opener (fired for you when you press run), your answer when the specialist asks you about the part your data cannot show, and one follow-up. The quote multiplies that expected back and forth by a deliberately conservative per-message figure, the same one the balance check uses, so a quote can overstate but can never understate what you end up charged.

That design is a promise: you consent to a number once, before the run, and the run does not then exceed it. A quote that guessed low would be a promise broken after the fact, with no way for you to see it happening, so Verinode rounds the estimate up instead. In practice most runs finish at or under the quoted figure. Every message's actual charge lands in your IU spend log as usual, and the published per-operation cost table lives on the public methodology page, covered in how IU costs are calculated.

The consent step

On the free Contributor tier, pressing run brings up a confirmation before any IU is spent, and it is worth reading, because it is a specific promise rather than boilerplate. It names exactly what the specialist will read and what it will do. Vendor Review's, for example: "I will go through your vendor spend and the jobs it landed on, and walk you through which vendors are running heavier than comparable work usually carries, where your spend is concentrated, and where you have room to renegotiate. I will tell you what I find and be clear about what your data does not show." Under the cost sits one more line of context: "This is a conversation, so the estimate covers the back and forth rather than one message. You are charged per message, and most run to about three."

Notice what these descriptions promise and what they refuse to. They promise to look and report; they never promise to find a specific number, because you might point the app at a thin book, and a promise to "show you the money you are losing" cannot be kept on data that does not contain it. Where an app has a hard boundary, the consent copy states it up front: Contract Review says it will not tell you whether a clause is enforceable, and Draft a Requisition says it will not run an applicant tracker or schedule interviews.

If your balance cannot cover the run, you are shown ways to add IUs or upgrade instead of the run starting; nothing spends first and apologizes later.

What never costs anything

• Estimate Analysis is free and shows no consent step. It is the one app computed statistically from your own past estimates, no AI model runs, so there is nothing to meter and nothing to consent to. Pick an estimate, press the button, read the result. See Estimate Analysis.
• Reopening a past result is free. IUs meter AI work, and the analysis was already paid for when it ran. Re-reading it under Past analyses is a stored read, not a new run.
• Exporting the PDF is free, for the same reason: the export renders exactly the finding that was saved, it never re-runs anything.
• A failed run does not double-charge you. If a run does not finish, the card says so, nothing is saved from it, and starting again does not charge you again for the run that failed.

HQ network apps show no cost prompt

If you use Verinode HQ, you will notice the network-level apps run with no cost line and no consent modal at all. That is not an oversight: HQ runs are not billed per operator, so quoting a spend would name a number that never lands on anyone. The operator-side quote-and-consent flow exists because your IU balance is yours, and nothing should draw on it without stating the figure first.

Best-practice example

A Contributor with a starter IU balance opens Vendor Review. The button says the run costs about 15 IU; the confirmation names what the specialist will read; they confirm. The run takes three messages and the spend log later shows the actual per-message charges, which total under the quote. The next week they re-read the result twice, export the PDF for a supplier meeting, and pay nothing further. When they later run Contract Review and their remaining balance cannot cover the quote, the run does not start on credit; they are shown the paths to add capacity, and the choice stays theirs.

Data sources