Likeness finds every claim your copy makes about a real person — named, implied or pictured — and judges each one against your own release register in Python. Not against a model's opinion of what sounds risky.
Nothing opens in this browser — this one takes a job and does the work on our own graphics cards in Oregon. The free allowance below is how you try it: 60 passages, no card and no payment, on the key you are issued.
One passage of copy read is one unit, never one claim found. A per-claim price would quietly reward this tool for finding more of them.
The problem
Marketing writes that a performer is exclusive to you, or appears in a series, or endorses the line. The release in the drawer covers a single scene from eighteen months ago and says nothing about any of that. Nobody connects the two, because the person writing the copy has no access to the register and the person holding the register never reads the copy — and the first time they meet is in a letter.
Two separate steps on purpose. The model finds the claims on Tower 1; Python judges them against your register. The soft step does the reading and the hard step does the deciding.
One passage read is one unit, however many claims come out of it. Charging per claim found would pay this program to be alarming.
Judged against the releases you actually hold, by name and by date. Not against a general sense of what is risky in this industry, which is advice you can get free and cannot act on.
The judgement is arithmetic against your register: is there a release, does it name this person, does it cover this use, was it signed before the use began, has it expired. Five questions with yes-or-no answers, asked in Python. The model never gets a vote on whether a claim is defensible — it only reads the copy and says what was claimed.
How it works
Listing descriptions, site pages, newsletter text, ad creative — whatever is about to go out with a person in it. A passage is a unit; a page of copy is usually one.
Every assertion about a real person: named, implied by context, or carried by a picture reference. Printed as a list with the sentence it came from, so you can see what was read and how.
Each claim against your own releases, in Python. The output is a finding or a pass per claim, with which of the five questions failed and what document would answer it.
What it will never do
This is a program about claims, which makes it a program that could easily start making its own. Three of these four exist to stop that.
Pricing
A passage is a chunk of copy — usually a listing description or a page section. Sixty are free on every key, which for most studios is a month of new copy, and that is on purpose: this program is most useful running before every publication rather than once in an audit.
A month of new copy, checked before it goes out.
Every page and listing on a working site.
A whole site and back catalogue audited.
Per passage read, never per claim found, and never a share of anything. The one pricing decision in this product that is actually a design decision. Your licence key is emailed to you within two working days of payment.
Checkout is our own bitcoin server, not a card processor. The units you buy are added to your key: buy from the Buy more box inside the program and they arrive on their own once the payment confirms, or buy here and paste the invoice number from your receipt into that box.
Questions
An assertion a reasonable reader would take as fact: that someone appears in a title, is exclusive, endorses something, was present, or holds a role.
Implied claims count. "Her first time back since the series" claims a series appearance and a gap, and both are checkable against a register.
Because they fail differently. A model reading copy is good at noticing an implied claim and bad at being consistent about whether a release covers it.
Putting the judgement in Python means the same claim against the same register gets the same answer every time, and the answer shows which of five questions failed.
Then this program cannot do its job yet, and it will say so rather than guessing.
The register is what Release and Custodian are for. Likeness is the step after those exist; running it against nothing would produce a finding on every claim, which is noise.
No. The reading runs on Tower 1's own graphics cards in Oregon and the judging runs in Python beside it.
A key. One line of text, issued to you and shown exactly once — only its hash is kept here, so it cannot be looked up later and we could not tell you what yours is. Every key arrives with a free allowance already on it, which is the tier at the top of the pricing above, and the work starts the first time you send a job with that key on it.
Each job is metered in the unit named in the pricing, and a unit that fails — a file that will not read, a row the model held back — is not charged and is named in the job report rather than silently skipped. When an allowance runs out the next job is refused before any work starts, with the number it used printed on the refusal.
More units are bought in bitcoin. The checkout is a bitcoin server we run ourselves. Buy from the Buy more box inside the program and the units are added to your key on their own as soon as the payment confirms; buy from the plan buttons above and paste the invoice number from your receipt into that same box. Each payment adds its units once, to one key.
Not the archive. The claims about to be published are the ones where a finding is still cheap to act on.