You built something real.
Tell it with numbers, to buyers.
An AI agent reads the running system, writes the report twice (technical and plain) with every number labelled, drafts one post per channel in your voice and checks that no number changes on the way. You rewrite and publish.
What it costs to have your work told.
Public ranges of content written by professionals, and our service launch price.
Range source: whito.co.uk (July 2026), read 8 Oct 2026. Bars not to scale.
The report first. Then the posts.
Two ways.
Do it with your AI agent, or I write it about your work.
The kit
Do it yourself
$19
The service
I do it for you
€290 €490
Launch price.
Before buying, read the free guide.
9 pages: the problem, what it costs, the method, the report, a case with the error found, the posts per channel, the limits. Enough to tell your next work without numbers that move.
What you are probably thinking.
"Do I have to give access to the system?"
Read access: logs, dashboards, the running site. The agent changes nothing; what it cannot verify it labels as declared.
"Does it write in my voice?"
It first reads three to five of your published posts. Then the posts stay drafts: you rewrite them, and after the rewrite is when to run the check again.
"What if I have no results to show?"
Tell what there is: the problem, the choices, the mistakes, the cost. Traction is written only with the number and its source.
"What does the check check?"
That every number in the post appears in the report, with the exact digits; and that an estimate is said as one.
"Do I need to code?"
No. Your AI agent, Claude or ChatGPT, uses the kit; you answer three questions and rewrite the posts.
Where it comes from.
It is the method I use to tell my own systems: a 16-section report on the automated newsroom of one of my projects on 27 September, the launch post on the 28th. Running the published post through the check, I found a spend written ten times bigger than the real one.
Read more: how to write a case study of technical work.