Founder Vetting Report

Overview

A Founder Vetting Report is a paginated, twelve-section due-diligence document covering one founder. It is generated on demand from public OSINT plus the subject's own enriched profile, takes three to five minutes to produce, costs 200 credits (about $80 USD at the standard credit pack rate), and is delivered as a shareable HTML report with a printable PDF view.

It is intended for an investment partner, a corporate development lead, an acquirer, an executive recruiter, or a journalist who needs an institutional-grade synthesis on a single founder before a meeting, an investment, or a story.

The report is not a credit report, not a background check, and not an FCRA-regulated consumer report. It does not access sealed records, paywalled investigative archives, or non-public databases. Everything that appears in the report is recoverable by any sufficiently patient researcher with public-internet access; what we sell is the synthesis, the cited audit trail, the framework adherence declared on this page, and the speed of delivery.

The Three Frameworks We Adopt

MentionFox / Verifierce reports align with three published standards. They were not written for venture due diligence — they were written for the U.S. intelligence community, professional fact-checkers, and the United Kingdom's Defence Intelligence agency — but they translate cleanly to investment-grade research, and adopting them publicly is the single most credible thing a research vendor can do.

ICD 203 — Analytic Standards (Office of the Director of National Intelligence)

The U.S. Intelligence Community's Directive 203 defines nine tradecraft standards that any piece of finished intelligence is required to meet: properly described sources, proper expression of uncertainty, distinction between intelligence and assumptions, incorporation of alternative analysis, judgement of consequences, customer-relevant focus, logical argumentation, accurate reflection of source content, and clear language. We treat these as binding for every Vetting Report we produce. The full list is enumerated in section four below.

KSJ Three Models of Fact-Checking (Knight Science Journalism, MIT)

Deborah Blum's KSJ Fact-Checking Project at MIT documents three distinct working models: pre-publication line-by-line verification (the New Yorker model), post-publication rapid response (the Washington Post Fact Checker model), and structured public-knowledge correction (the BBC editorial model). A Vetting Report borrows from all three. Sonnet drafts each section under cited-evidence constraints (pre-publication discipline). The Corrections section below describes our post-publication response. And the methodology page you are reading enacts the third model — public statement of how we know what we claim to know.

UK PHIA Probability Yardstick (UK Defence Intelligence)

The Professional Head of Intelligence Assessment publishes a seven-band probability yardstick used across UK government intelligence reporting since 2018. It maps natural-language probability words ("highly unlikely", "realistic possibility", "almost certain") to numeric bands (under 5%, 10-20%, 25-35%, 40-50%, 55-75%, 80-90%, over 95%). The yardstick eliminates the well-documented problem that an analyst writing "may" and a reader interpreting "may" can mean wildly different things. Every probabilistic claim in a MentionFox Vetting Report is expressed using the seven yardstick terms, paired with a separate analytical-confidence rating (High / Moderate / Low) per ICD 203.

D&O Underwriter Risk Framework

The Directors & Officers liability insurance industry maintains a public-record framework for underwriting officer-level risk: executive compensation versus industry norms, related-party transactions, insider stock-sale patterns, prior litigation exposure, and material weakness disclosures. We adopt that framework verbatim in Section 6.5 of every Founder Vetting Report. A founder generates a HIGH / MODERATE / LOW D&O Risk Flag from the public record before any private indemnification structure is brought into diligence. For private bootstrapped founders most public-disclosure inputs return [insufficient public evidence]; that null result is itself a structural finding (no SEC obligation, no public exposure, lower D&O premium-equivalent risk on the public-record axis), not a gap.

The Twelve Sections of a Founder Vetting Report

The Nine ICD 203 Tradecraft Standards

Each Sonnet section prompt is constructed against these nine standards. Compliance is enforced at draft time (banned-words list, cited-URL minimums, mandatory uncertainty-flag phrasing) and reviewed at the executive-summary synthesis pass.

UK PHIA Probability Yardstick

Verbatim from the UK Professional Head of Intelligence Assessment guidance. Used in MentionFox Vetting Reports anywhere the underlying claim is probabilistic rather than asserted.

Every PHIA-band claim is paired with an Analytical Confidence rating expressed separately:

• High Confidence — multiple independent authoritative sources, recently re-verified.
• Moderate Confidence — one authoritative source or multiple secondary sources.
• Low Confidence — claim is inferred from indirect evidence or relies on uncited assumptions; the reader should weight accordingly.

Example phrasing as it appears in a real report: "Highly likely (~85%) the founder retains over 50% equity post-Series-B (Confidence: Moderate — based on standard dilution math plus partial round disclosures)." The phrasing is mandatory; vague hedges ("may", "could", "perhaps", "possibly") are banned in the Sonnet prompts.

Source Class Taxonomy

Per ICD 206, every citation declares its source class. Four classes, ordered by reliability:

Section twelve of every Vetting Report opens with a class-distribution summary of the form "47 citations: 18 Primary, 22 Authoritative-Secondary, 5 Aggregator, 2 Unverified." A report with too few Primary citations or too many Unverified ones is a self-disclosing weakness; the reader can decide what weight to put on it.

Honest Limits — what we do not do

Corrections Policy

Modeled on the BBC editorial corrections process. Three commitments:

1. Identification window. Errors flagged within thirty days of report generation are corrected on the canonical view URL within five business days. Errors flagged after thirty days are evaluated and corrected at our discretion.
2. Re-publication, not silent edit. Corrections do not overwrite the prior text silently. The report's view page preserves a redline diff between the original draft and the corrected draft, time-stamped, with a one-line explanation of the correction. Any reader who saw the original can audit the change.
3. Subject right of reply. The subject of any Vetting Report may submit a one-paragraph factual rebuttal to corrections@mentionfox.com. Verifiable rebuttals attach to the report alongside the original section. Where the subject and our research disagree on a public-record claim, both views are surfaced; we do not silently capitulate, and we do not refuse to publish the subject's view.

Data integrity floor — ALCOA. Every Vetting Report carries an ALCOA Methodology footer at the end: each factual claim is Attributable to a cited source, presented in Legible plain language, marked with the date it was Contemporaneously verified, sourced from the Original primary record where available, and Accurately reflects the underlying evidence with explicit uncertainty flags where evidence is thin. ALCOA is the U.S. Food and Drug Administration's data-integrity principle for regulated industries; we adopt it as the floor for synthesis research because it captures the same disciplines without the regulatory overhang.