Screen a Vendor Against a Real Ownership Graph: Live
A while back we made a claim: onboarding a vendor is "who ultimately owns and controls it, and can we prove none of them are?" ("is this company sanctioned?" checks the applicant only). That is a graph traversal of the ownership chain. We wrote it up in Documents, SQL, Graphs and Time, then measured it on real data: a composed decision (resolve identity, traverse ownership, screen every owner) scored 100% against a naive single-node screen's 36.7%.
Now you can run it yourself.
Try it
grc.redevops.io: type or pick a company. The runtime resolves its identity, walks the ultimate-ownership chain, screens every owner against sanctions, and returns Go / No-Go / Abstain, with the full chain and the flagged owner shown. The ownership traversal is sub-millisecond; the whole decision, a few.
On real data, and we will say exactly what
No mock. It runs on 182,445 real GLEIF golden-copy entities (CC0) with 168,818 ownership edges, flagged against OpenSanctions (CC-BY-NC, non-commercial research use). One caveat, stated plainly: the sanctioned flag here is a derived exact-name match (not a licensed screening decision): enough to demonstrate the mechanism, and the benchmark's matching-robustness half deliberately stress-tests exactly that failure mode (real aliases and coincidental namesakes). A production deployment plugs in your own licensed sanctions and adverse-media feeds.
The thing the naive screen misses
Screen only the applicant's name and you onboard 45 of 50 sanctioned-linked vendors, because the sanction sits on an owner one to eight hops upstream. The demo shows it live: a clean-named applicant, a naive GO, and a composed NO-GO citing the exact sanctioned owner, with the path preserved as evidence. It also abstains when a name resolves to more than one entity, instead of guessing the wrong "Acme." That gap (100% vs 36.7%, and zero sanctioned entities onboarded instead of 45) is the whole point.
What it saves
The value is the analyst hours you stop spending tracing ownership by hand across registries, the questionnaire round-trips you skip, and (the expensive one) the bad onboarding decisions you never make. We built a transparent ROI calculator: every figure is derived from your own inputs and editable assumptions, with no magic multiplier, so you can put real numbers on it in a minute.
Want it on your data?
This is a reference build on public data. If it is useful for your program (your entities, your frameworks (SOC 2 / ISO 27001 / KYB), your licensed sanctions and adverse-media feeds, wired to your ticketing for remediation), we will build and deploy it to your specs, on the same governed Context / Mission runtime that decides what to screen, when to abstain, and how to replay the decision for an auditor a year later.