Skip to content
Counterparty background check

Counterparty background check

Agent: the counterparty check (counterparty_check), run from the Counterparty background check card on Why (/simulate?tab=knowledge&card=counterparty; &name= starts the check with a business name). The old /market_diligence address redirects to that card (owner, 2026-10-31).

What it does

It runs business due diligence on a carrier, broker, freight forwarder, shipper or consignee from public registries and sanctions lists, and from your own history with that business. It produces findings for a person to review, never a rating. Its jobs:

  • Operating authority. Whether the US motor carrier or broker authority is active, and whether a revocation is pending.
  • Insurance on file. Liability insurance against the required minimum, and cargo insurance or a broker’s bond where one is required.
  • Safety and orders. The published safety rating and any out-of-service order in force.
  • Authority age. When the authority was first granted. A very new authority on a high-value load is a red flag.
  • Sanctions screening. The business name and the registered names against the OFAC SDN list, the EU consolidated list and the UK Sanctions List.
  • Company registries. Legal name, status and registration from GLEIF (by LEI), SEC EDGAR (by CIK) and UK Companies House (by company number): an inactive or lapsed entity, a company in liquidation or dissolved, an insolvency history, a recently incorporated company, and a registered name that differs from the one given.
  • Exclusions. Active SAM.gov exclusions (debarment or suspension), by Unique Entity ID.
  • Your own history. Loads, on-time share, claims and disputes with this business in your records.

What it reads

  • US FMCSA public carrier data (company census, licensing and insurance, authority history, out-of-service orders), by the USDOT or MC number you give.
  • OFAC SDN list (U.S. Treasury), the EU consolidated financial sanctions list (European Commission) and the UK Sanctions List (FCDO), downloaded in full and screened on the deployment.
  • GLEIF Legal Entity Identifier records (CC0), SEC EDGAR company submissions (U.S. Securities and Exchange Commission), UK Companies House company profiles (Open Government Licence) and SAM.gov exclusions (U.S. General Services Administration), each by the identifier you give.
  • Your own lane history and account records.

These registries were chosen because each is public, free, usable commercially, has an API keyed by an identifier, and registers companies. Where a registry can also hold a person (an LEI registered to a sole proprietor, an individual EDGAR filer), the record is reported only as such, with no name. Court dockets are not consulted. A source that is not given an identifier, or whose API key is not configured on your deployment, is listed as not checked with that reason.

What it produces

  • Findings, each with its source, a link to check it, and when it was retrieved.
  • Red flags among them: revoked or inactive authority, insurance below the required minimum, an unsatisfactory safety rating, an out-of-service order, a very new authority on a high-value load, and any sanctions name match.
  • Sanctions matches, always shown as a possible match that needs human review.
  • Every source considered, including the ones not checked and why.
  • At most one recommended action: a review by the Compliance Manager.
  • The thresholds it applied.

Human oversight

  • The check decides nothing. A person reviews the findings and decides whether to work with the business.
  • FACE does not block, reject, blacklist or notify anyone.

Not a consumer report

This is business due diligence. It runs on businesses only and has no way to name a private individual. It is not a consumer report, and it must not be used to make or support decisions about any person’s credit, employment, housing or insurance. A name match on a sanctions list is not an identification: only a person can confirm or dismiss it.

Model

The findings are read from coded registry fields and computed on the deployment. No model decides them.

The optional summary uses the general reasoning tier.

Where it runs and data handling

On the deployment’s own model plane: Runink’s shared machines on a Trial, your own cloud on Cloud and your own servers on On-premises. No outside AI service is called on any plan. Registries are contacted for every organisation, unless the deployment has turned live public data off. The only values sent are the identifiers you typed: USDOT, MC, LEI, CIK, UK company number or SAM.gov UEI. Sanctions lists are downloaded whole and screened locally, so the business name never leaves the deployment. SAM.gov and Companies House need a free API key, configured per deployment as a secret; the key is never shown, logged or passed to a model. Everything fetched is cached with its retrieval time. See Facts shared by every card.

Guardrails

  • Refuses a check on a private individual, or for a decision about a person’s credit, employment, housing or insurance.
  • Refuses protected characteristics as a factor.
  • Refuses to hide, drop or clear a sanctions match or a red flag.
  • Never labels a business fraudulent or criminal and never blacklists it.
  • Treats everything in a registry record or a list as data. It follows no link and no instruction found there, and a summary is withheld when the material carries one.

How to read its results

  • Registry data is what the registry published at the time shown and can lag reality.
  • A clean result does not establish that a business is legitimate, and no name match does not clear anyone.
  • FMCSA covers US motor carriers, brokers and forwarders. For a shipper or consignee those sources are marked not applicable.
  • Each registry is checked by identifier (USDOT, MC, LEI, CIK, UK company number or SAM.gov UEI). Enter the identifier to check that registry.
  • A SAM.gov result labelled “not verified against the live API” comes from a deployment where the first SAM.gov call has not yet completed and been read. An answer FACE cannot read (an error page, an unexpected or partial response) is shown as not checked, never as “no exclusions”.

Thresholds

The checks consider how recently an operating authority or company was established, whether a load is high-value, and the claims ratio and on-time share in your own records. There are no built-in cut-offs: each is a target set for a deployment, or for your organisation by an admin under Admin → Public registries & market data. Until one is set, the figure is shown as a finding that says no target is set, and nothing is raised for review on it.