A transparent PBM is only as transparent as its ability to notice when its own models drift and its own data leaks.
One screen, two verdicts, on a synthetic claims stream shaped like a pass-through PBM's pipeline (adjudication to lowest-net-cost routing to a partner-marketplace egress). It runs on load. Flip one preset to see a healthy pipeline turn leaky and drifted. Click a hop to see what inferred-diagnosis PII it carries; open a SEV to see who gets paged.
Preset (one click is the whole demo)
Everything below updates from this toggle. Sliders aren't needed; the verdict is the demo. (The PII method and the drift method each have their own focused tool in the nav above.)
Privacy: inferred-diagnosis PII per hop (click a hop)
Drift: are the savings, routing, and prior-auth metrics paging anyone?
Sources & method
Pipeline shape (adjudication, Drug Pathways routing, Connect-360 partner egress) follows SmithRx's own public framing of its platform and savings marketplace (company blog / Business Wire; company-self-reported). The hops here are a generic pass-through-PBM shape, not a map of SmithRx's real systems.
All prevalence numbers, claims series, and SEV events are SYNTHETIC, generated deterministically in this page. No SmithRx PHI, no real member data, no real savings figure is touched or implied. The high-sensitivity drug classes (HIV, oncology, behavioral health) are included because drug-to-condition inference is the privacy point, at a labeled, tunable prevalence.
Honest bound: this is the method on a synthetic pipeline shaped like yours, not a read of your real data. The real per-hop prevalence and the real metric series live behind your firewall; reading them there is the internal extension this scopes. The prior-auth analogy to insurer medical-claims litigation is framing, not a PBM precedent.