A readiness read, in full.
This is an anonymised composite, built only to show the shape of the work. It describes no real firm. Any real AIQ score is private to the company it belongs to and is never published.
Anonymised composite, the Meridian pattern. Describes no real firm. Public information only. Any real AIQ score is private.
The question a diligence room asks first
When this composite cites a model that "drives retention," can a single point of that retention be shown as real, measured, and attributable to the model itself, rather than to a pricing change, a new onboarding flow, and a strong quarter that all landed at the same time? On the public story alone, the claim reads as confident and unproven. A diligence team does not argue with confidence. It asks for the cohort where the model was switched off, and prices what it cannot see.
The read, headline first
Read on public signals, this composite would have started nearer AIQ 42, not because the work is weak but because the evidence behind the claims was never assembled as evidence. On a defined remediation path it reaches 70, against an observed ceiling of 86 for its profile. The claims do not change. What changes is that each one can now survive a question. The number is shown here only because this firm is invented. For a real company it stays private.
Where a diligence room would press
Six evidence blocks, each tagged to how a serious investor actually reads an AI claim.
Is there a single accountable owner for the AI, or is it distributed across vendors and operations with no centre of gravity? The composite names capable engineers but no owner of the claim.
Which systems are in production today, not pilots or generic automation, which workflow does each touch, and what could be shown live? The story implies more than the public record supports.
Does an owned model change a client outcome, or does the intelligence largely sit in a partner or an off-the-shelf tool that a competitor could wire up next quarter?
Can one efficiency or retention gain be quantified with a before and after, and decomposed to isolate the model from scale, mix, and ordinary discipline? If the model were set to zero, would the headline change at all?
What proprietary, consented, model-ready data does the book generate, and what owned capability could be built on it that a competitor buying the same tools could not replicate?
Across products and markets, which lines actually run the AI in production versus aspiration, and does the narrative match the file underneath it?
Indicative readiness read
On the readiness instrument, and on public signals alone, this composite would sit in the lower band, and that is a feature of an outside-in read, not a criticism of the business. The band looks low mainly because, from the outside, the AI presents as supporting colour rather than an evidenced value driver. The above-floor signal comes from an honest public posture. The load-bearing items, a named system, a measured banked gain, a decomposition, a governance owner, sit near the floor, likely because they have never needed to exist. A closer-than-you-think band is reachable quickly, precisely because the underlying business is real. The gap reads as evidence and governance, not capability.
What would close it, three exhibits
- A measured banked-gain exhibit. One named deployment with a system-recorded before and after, plus a decomposition that separates the model's contribution from scale and from partner effects. This is the exhibit that moves the headline from asserted to evidenced.
- A production-system register. A one-page inventory of the AI systems actually live, each with an owner, a workflow, a go-live date, and a recorded demo. This turns broad adoption posture into shown evidence.
- A data and governance pack. A stated data-moat thesis over the proprietary book, paired with an accountable owner and model-risk controls fit for the relevant regulators. This converts latent defensibility into an investor-ready asset.
A note on independence
This composite references public-style signals only. It is not a rating or certification of any company, and it was prepared in a personal capacity, not on behalf of any employer. A real read is private to its owner, and below-the-line results are never displayed. The structure here is the structure of the work, and the work is constructive: see the gap from the outside while you can still close it from the inside.
When does your AI claim go to print, or when is your next LP or investor AI-diligence cycle?