About
Underwriting got better tools for deciding, and none for remembering
The last decade of insurtech went almost entirely into inference: better models, better scores, better predictions. Very little went into the record those predictions sit on. Ask most underwriting teams where a number in a file came from and the honest answer is that somebody typed it in from a browser tab, some months ago.
That is fine right up until the file is read by someone else. A regulator reviewing a filing, a reinsurer pricing a treaty, an examiner sampling declines, counsel on a disputed claim. Every one of them asks the same thing, and it is never what the score was.
PropertyRiskIQ is the boring half of the problem: assemble the property record from named sources, in one pass, and keep the evidence attached. Inference is more interesting. Evidence is what holds.
What we believe
Four positions
A citation beats a score
A score is a compression of evidence, and compression is lossy in exactly the direction that matters when a decision is challenged. Show the score if it helps, but never instead of the record.
Gaps belong on the page
The temptation in this category is to fill every section with something plausible. We would rather ship a layer marked as not connected than a confident number we cannot cite. Our own data sources page names the gap.
The best security control is not receiving the data
A product that needs your book needs a vendor risk review, a data processing agreement and six months. One that only needs an address needs an afternoon. That constraint shapes what we will and will not build.
Precision about limits is a feature
A distance band is not an ISO protection class. A data lookup is not a flood determination. Being exact about this costs nothing and prevents the failure mode where a tool gets used past what it can support.
How we work
Small on purpose, for now
PropertyRiskIQ is built by a small team inside an independent product studio. There is no sales organisation, which means when you ask a question about how a data source behaves you get an answer from someone who wrote the adapter.
We are onboarding a limited number of design partners rather than opening self-serve signup, because the product is better served right now by a handful of underwriting teams telling us where the file is wrong than by a larger number of users we cannot talk to.
If that sounds like your team, we would like to hear which addresses you would test it on. Write to hello@propertyriskiq.com.