How this site is built — every claim sourced, every datum graded
A short note on the discipline behind this site: every kit datum carries a source URL + evidence grade (A/B/C) + verified_at timestamp. What that means for buyers reading specs here vs reading marketing decks elsewhere.
If you buy a research-grade reagent and the spec sheet says "94.1% transfection efficiency", you have to decide whether to believe it. Most spec sheets do not tell you where that number came from. This one does.
Every claim on this site about a SunVax kit carries three pieces of metadata: where the claim came from (a URL or a slide reference), an evidence grade (A for peer-reviewed primary source, B for conference / vendor / agency disclosure, C for cross-correlated inference), and a verified_at timestamp showing when we last confirmed the source still resolves to what we say it says. The 94.1% mouse-CD4-T number on SV101 is from SunVax's own 2026-05-03 slide 3 — Grade B (vendor disclosure, not yet peer-reviewed at this slide's level of detail) — and we verified it via direct slide read on 2026-05-17.
Why this matters for a buyer. Marketing pages typically present numbers without provenance. A scientist reading "95% efficacy" has no way to know whether that's a single best-case run, a typical run, or a confabulated figure. By requiring source + grade + verified_at on every claim, we make the same number falsifiable: you can click the source, check the original, and disagree if our interpretation is wrong. If our interpretation IS wrong, email back and we correct it within one business day. That has happened — once on Liu 2025 PMC11743637 where our initial draft had the dose/efficacy wrong on a referenced bispecific paper, and the audit pipeline caught it before publication.
What the four-grade system actually filters. Grade A means peer-reviewed primary source — PubMed, DOI-resolvable journal articles, NCCN guidelines, FDA documents. About 70% of our claims hit Grade A. Grade B is conference abstracts, vendor disclosures, agency announcements, named trade press — about 25% of claims. Grade C is inference from multiple Grade A/B sources; it appears with explicit "(Grade C inference)" annotation. Grade D is single-source unverified — forbidden on customer-facing surface; if a fact only has a Grade D source, we either upgrade it or remove it before publishing.
The freshness side. Every claim has a verified_at timestamp. Claims verified within 30 days are tagged FRESH; 30-90 days CURRENT; over 90 days STALE and flagged for refresh. If a source URL ever stops resolving, the audit pipeline catches it within 30 days and we either re-resolve to a current URL or downgrade the claim. The customer-visible refresh log on every quarterly issue shows exactly how many claims were re-verified in that cycle.
What this discipline does NOT do. It does not predict your experiment's outcome. It does not replace SunVax's empirical screening (which is the founder's documented methodology — see /insights/empirical-vs-ml-lnp). It does not certify the kit for clinical use (these are research-grade kits). And it does not promise that every claim we write is true forever — only that every claim was true as of its verified_at, with an explicit URL to check, and a 1-business-day correction SLA when it isn't.
If you ever want to audit any specific claim on this site, the email is on the contact page. If the URL has moved or the data has changed, we want to know.