World Model Readiness
Engraved team-functions instrument

For your team

Teams by Function

Module · Scaling output without scaling risk

The Marketing Team AI Check

AI lets your team produce ten times the content, which is a gift and a trap. The same speed that fills the calendar can flood your channels with off-brand copy, unlicensed images, and posts that read like a robot wrote them because one did. This module checks the five controls that let output scale without the brand paying for it: voice guardrails, human review before publish, asset provenance, channel-fit checks, and honest attribution of what the AI content actually earned.

Question 1 of 5 · Brand voice holds

Does AI-generated copy come out sounding like your brand, not the model?

A model defaults to a generic, over-eager voice that reads the same across every company using it. Guardrails mean a documented voice, examples, and a review that catches the drift, so your content sounds like you and not like everyone else's prompt output.

Question 2 of 5 · Human reads before publish

Does a person read AI content before it goes out the door?

AI produces confident text that can be subtly wrong, off-tone, or embarrassingly generic, and the failure is public the instant it publishes. A human in the loop before publish is the cheapest insurance there is. The alternative is your audience doing the proofreading.

Question 3 of 5 · Asset provenance tracked

Do you know the origin and licence of the AI-generated images and assets you publish?

AI images can reproduce copyrighted work, invent a face that resembles a real person, or come from a tool whose licence forbids commercial use. Provenance means you can say where an asset came from and that you are cleared to use it, before a rights holder asks.

Question 4 of 5 · Content fits the channel

Does your team check that AI content actually fits the channel before it ships?

A model will happily produce copy that is technically correct and completely wrong for the channel: LinkedIn earnestness on TikTok, a wall of text where a subject line belongs. Channel-fit means someone checks the format, length, and register suit where it lands, not just that the words are good.

Question 5 of 5 · Attribution is honest

Can you tell whether the AI content actually performed, or just filled the calendar?

It is easy to celebrate producing more and never ask whether it worked. Honest attribution ties AI-assisted content to real outcomes, engagement, conversion, pipeline, so you can tell the volume that earned its place from the volume that just added noise the audience learned to ignore.

For the statistics · one click each

Three questions for the public picture

These do not affect your score. They feed the anonymised, aggregated statistics; groups under 8 respondents are never shown.

How many people are on your marketing team?

1 to 2
3 to 6
7 to 15
Over 15

What share of your published content is now AI-assisted?

Under 10 percent
10 to 30 percent
30 to 60 percent
Over 60 percent
We do not track it

How does AI content get reviewed before it publishes?

It publishes automatically
A quick skim
A human reviews it
Reviewed against a checklist
It varies by person

Your context

Used to calibrate the report. Company size and sector remain in the anonymized dataset; your email does not.

What the five levels look like

Every dimension in this assessment is scored 1 to 5. This is what the levels mean, dimension by dimension. The graded report diagnoses where your own answers land and what to do about it.

Brand voice holds

  1. 1Sounds generic
  2. 2Voice varies wildly
  3. 3Loose guidelines
  4. 4Documented voice
  5. 5Voice enforced in review

At the low end: Generic AI copy makes your brand indistinguishable from every competitor using the same tool. Write down your voice with examples and hold copy against it before it publishes. What good looks like: Voice enforced in review keeps AI volume from diluting what makes your brand sound like itself. Refresh the examples as the brand evolves; a style guide frozen in time slowly goes stale.

Human reads before publish

  1. 1Auto-published
  2. 2Skimmed at speed
  3. 3Reviewed sometimes
  4. 4Reviewed before publish
  5. 5Reviewed against checklist

At the low end: Auto-publishing AI content means your first reader is your audience, and the mistakes are already live. Put a human approval step before anything reaches a channel. What good looks like: A checklist review before publish scales your judgement across the volume AI produces. Keep the checklist tight; a review that checks everything carefully is the one thing volume pressures first.

Asset provenance tracked

  1. 1No idea
  2. 2Assume it's fine
  3. 3Tracked informally
  4. 4Provenance recorded
  5. 5Provenance and licence cleared

At the low end: Publishing AI assets with unknown provenance is a rights complaint waiting to happen. Start recording which tool made each asset and under what licence. What good looks like: Recorded provenance with cleared licences means AI assets carry the same rights hygiene as anything you commission. Keep it current as tools change terms; a licence that was fine last year can shift.

Content fits the channel

  1. 1One size fits all
  2. 2Rarely adapted
  3. 3Adapted sometimes
  4. 4Channel-fit checked
  5. 5Fit checked and measured

At the low end: Pushing the same AI copy to every channel guarantees it fits none of them well. Adapt format and register per channel before you ship, even a quick pass helps. What good looks like: Checked and measured channel-fit means AI volume lands right where it goes rather than everywhere at once. Keep learning from the numbers; what fits a channel shifts as the channel does.

Attribution is honest

  1. 1Never measured
  2. 2Volume equals success
  3. 3Some tracking
  4. 4Performance attributed
  5. 5Attribution guides output

At the low end: If you never measure whether AI content performs, you cannot tell productivity from noise. Attribute output to real outcomes so more does not automatically read as better. What good looks like: Attribution guiding output means your team makes more of what works and quietly stops making what does not. Watch for last-click blind spots; the content that assists the win often gets none of the credit.