Show, don't tell.
The sample report
We are running the AI Spend Assessment on Recovea's own AI workload and publishing the whole thing here: the findings, the numbers, and whatever they say about us.
The sample report is being produced. It is not published yet.
This page exists ahead of the artifact on purpose. It is the link in our outreach, and a link that 404s is a worse first impression than a page that tells you where things stand. When the report is written it lands here, at this address, in full.
No summary of it exists yet either, and we are not going to describe findings we have not finished measuring. What we can tell you is exactly what the document will contain, because it is the same deliverable every customer gets.
What will be on this page
The same four things an assessment delivers, with our own workload as the subject and nothing edited out because it is unflattering.
The written findings, ranked
Our own AI spend attributed by application, project, model, and route, then ranked by what actually moves the bill.
The report itself, downloadable
The document, exactly as a customer receives it, with the method and the assumptions behind every number stated in it.
The receipts export
A hash-chained receipts export of our own workload, with the instructions to re-derive the chain offline yourself.
The uncomfortable parts
Including whatever the meter says we are wasting. An audit you cannot fail is a sales pitch, and running one on ourselves is the only version of this page worth publishing.