Extract data from KYC / AML documentation

KYC and AML documentation is the record a regulated business keeps to prove it knows who its customer is and has checked that customer against sanctions and financial-crime risk. Banks, fintechs, law firms, and corporate service providers assemble the file at onboarding and refresh it on a schedule: the customer identity and identification documents, whether the customer is a person or an entity, the beneficial owners behind a company, the screening results against watchlists, and the risk rating that follows. Compliance analysts, onboarding teams, and auditors all read the same fields. Internationally, the Financial Action Task Force sets the standard for customer due diligence, including its guidance on politically exposed persons, and financial institutions must identify the beneficial owners behind legal-entity customers, so the file carries a consistent shape even when the source forms differ from one provider to the next. Identity is only the start; the risk assessment is where the fields get dense. One politically exposed person flag, a beneficial-ownership breakdown where every owner above a threshold has to be named, and a screening result against sanctions and watchlists such as OFAC, UN, and EU lists all feed a risk level of low, medium, or high, which in turn decides whether enhanced due diligence applies. Beneficial owners arrive as a table, each with a name, an ownership percentage, and a country, and the percentages have to make sense against the entity. A watchlist match, its source, and the action taken are recorded for the audit trail. For Kestrel Holdings Ltd, screening against OFAC, United Nations, and European Union lists on 2026-05-18 returned no match, the risk level was recorded as medium, and beneficial owners Aurelia Voss and Tomas Reinholt were named with their percentages. Because handling this data is heavily regulated, extracting it is strictly a records task: the extraction reports what the documentation states and does not run screening, decide a risk rating, or offer any compliance advice. Talonic reads the KYC and AML documentation and returns the subject identity, the identification documents, the beneficial owners, the screening results, and the risk assessment as fields and tables. A file for the entity Kestrel Holdings Ltd, verified 2026-05-18 by document review and database screening against Refinitiv World-Check with no watchlist match, carrying a medium risk level and no politically exposed person flag, beneficial owners Aurelia Voss at 40 percent and Tomas Reinholt at 30 percent, and a 250,000 USD transaction limit, loads into a compliance system so an analyst reads the identity and the risk fields from data rather than the source forms. No screening or rating is performed by the extraction.

What gets extracted from KYC / AML documentation

How extraction works for KYC / AML documentation

KYC and AML files are assembled from onboarding platforms, identity-verification vendors, and manual review, and reach a compliance team as PDFs and scanned forms where identity, ownership, and screening results sit in different sections. Talonic classifies the documentation and maps it to the identity-verification schema in the Field Registry, which separates the subject identity and identification documents from the beneficial owners, the screening results, and the risk assessment. Subject type, individual or entity, is read from the source, the identification documents return as a table with their type, number, and issuing country, and the beneficial owners are kept as a table with each owner name, ownership percentage, and country. Politically exposed person status, the watchlist-match indicator, the screening source, and the recorded risk level are captured as their own fields, and any enhanced-due-diligence requirement is read as a boolean. Each value returns with a confidence score and a source-region pointer that meets DIN SPEC 91491, so an analyst can verify an ownership percentage or a screening result against the source file. Records-only by design, this extraction reports what the documentation states; it does not perform screening, set a risk rating, or provide compliance advice.

Sample extraction

A corporate KYC file with beneficial ownership and screening

{
  "document_number": "KYC-2026-0514",
  "document_date": "2026-05-14",
  "subject_full_name": "Kestrel Holdings Ltd",
  "subject_type": "corporate_entity",
  "subject_politically_exposed_person": false,
  "verification_status": "verified",
  "verification_date": "2026-05-18",
  "verification_method": "document review and database screening",
  "screening_source": "Refinitiv World-Check",
  "screening_watchlist_match": false,
  "risk_level": "medium",
  "enhanced_due_diligence_required": false,
  "currency": "USD",
  "transaction_limit": 250000,
  "governing_law": "United Kingdom",
  "beneficial_owners": [
    {
      "owner_name": "Aurelia Voss",
      "ownership_percentage": 40,
      "owner_type": "individual",
      "country_of_residence": "Germany"
    },
    {
      "owner_name": "Tomas Reinholt",
      "ownership_percentage": 30,
      "owner_type": "individual",
      "country_of_residence": "Denmark"
    }
  ]
}

Frequently asked