company due diligence

The problem

Vetting a new vendor, supplier, or counterparty means cross-referencing registries, sanctions lists, and adverse media by hand — hours of work, and one missed shell-company signal or ownership discrepancy can be expensive.

What it does

This is a structured method the AI follows — not a data service with its own feeds. It guides the model to:

• Work a company or counterparty across international jurisdictions using only public sources.
• Check registration status, entity type, and incorporation details against official registries where reachable.
• Screen the entity and its key people against sanctions lists, watchlists, and PEP references with a two-tier verification test.
• Weigh digital-footprint coherence, domain age, and adverse-media signals for fraud indicators.
• Produce a report with a mechanical GREEN / AMBER / RED verdict from a fixed decision table.

Honest by design

Every claim carries a certainty tag — verified / reported / unverified — and the skill never invents registry data. A control it could not run is marked unverified, never quietly passed off as "no risk found." That anti-false-GREEN rule is the difference between a check that was done and one that only looked done.

Audited & signed

The assessment logic — the legal/ethics gate, the source method, the sanctions test, the verdict table — ships under a detached GPG signature over a SHA-256 manifest, with verification instructions included. Cross-check the signing key's fingerprint against this listing (an independent channel):

3E5C 01D0 6619 A997 5C83 DDA7 7273 DD27 D9C1 E2B1

Alter a single rule and verification fails. You are buying logic you can prove hasn't been tampered with.

Why this beats prompting it yourself

A one-off prompt manufactures confidence. This enforces a legal/ethics gate — public sources only, data minimization, business purpose, GDPR-aware on natural vs legal persons, and no disguised surveillance of private individuals — plus a synthesis framework that ties every conclusion to a verification tag and a mechanical verdict. The output is auditable, not improvised.

Use cases

• Screening a foreign supplier before an initial deposit or signing.
• Background-checking a B2B customer before extending credit terms.
• Preliminary due diligence on a target for investment or acquisition.
• Sanity-checking a freelancer or sole trader who claims specific business credentials.

Known limitations

Public-source data only. It is a method the AI executes, not a licensed data provider — coverage depends on the sources reachable in your setup, and UBO (ultimate beneficial owner) access varies by jurisdiction. Not a credit agency, not KYC/AML, and not a substitute for formal legal advice or licensed screening.