Skip to content

Claims-verified indemnity

Structure 2 of Inference Re: what happens once claims start accumulating against attested deployments. This is where Volume 1’s underwriting fundamentals — rating & pricing and the loss ratio — meet the evidence stack.

Parametric triggers get the market started without loss history. But every policy written on an attested deployment does something no AI policy has done before: it pairs claims outcomes with an independently verified exposure population. As those claims accumulate, “the first verified loss ratio emerges — enabling indemnity pricing with a real denominator.” GROUNDED · Brochure

In Volume 1 terms: the loss ratio is losses ÷ premium, and it is the hub of the flywheel. What was missing for AI risk was never the numerator — losses will arrive on their own schedule — it was a denominator anyone could trust. See the missing denominator.

Once behavior is continuously attested, price can follow behavior:

  • “Credit for cleaner attestation records.” GROUNDED · Brochure An insured whose receipts show low drift, few scope-exceedance blocks, and healthy review completion earns a defensible discount — the AI-liability version of a good-driver record.
  • This is the same mechanism as schedule rating from Volume 1: a documented credit for a demonstrated risk characteristic. The difference is the documentation — a notarized attestation history instead of a questionnaire. STABLE
  • The renewal conversation writes itself: the insured’s own evidence is the submission. (Customer success runs on this — see 22-04.)

Experience rating and credibility weighting, in producer language

Section titled “Experience rating and credibility weighting, in producer language”

Two Volume-1 concepts do the heavy lifting here:

Concept Plain version In Inference Re
Experience rating Price a risk from its own verified history rather than the class average. STABLE Possible only once attested deployments accumulate paired attestation-and-claims data.
Credibility weighting How much weight the insured’s own experience gets versus the book rate — more data, more weight. STABLE The exposure-unit definition and credibility-weighting approach are a cohort deliverable, to be published as an open working paper. GROUNDED · Brochure

The producer’s takeaway: a brand-new attested deployment is priced mostly on the book (low credibility); a deployment with a long, clean receipt chain earns its own rate. Nobody gets to assert their way to a discount — the receipts carry the weight.

“Every policy written generates the data that prices the next one. The product compounds.” GROUNDED · Brochure That is Volume 1’s flywheel restated at market scale: evidence → better selection and defensible credits → lower loss ratio → capacity → more attested premium → richer experience base → repeat. The claims-verified structure is the flywheel’s second lap.

02 AI liability ILS

Drill this page →17 bank questions stand behind what you just read. Check it while it’s warm.