KYB (Know Your Business)

KYB (Know Your Business) is the process of verifying the legal identity, ownership structure, and legitimacy of a company before onboarding it as a customer or partner. It covers checks such as confirming registration status, identifying ultimate beneficial owners, and screening the business against sanctions and watchlists. Where KYC verifies individuals, KYB verifies the legal entity and the humans who control it.

How it works

A KYB check starts with document collection: certificates of incorporation, articles of association, shareholder registers, tax identification records, and proof of address. A compliance team, or increasingly an automated pipeline, then extracts structured data from these documents and cross-checks it against public registries and sanctions lists.

The hard part is rarely the first layer. It's tracing ownership through holding companies, trusts, and nominee shareholders to find the actual person who benefits from the business (the ultimate beneficial owner, usually anyone above a 25% ownership threshold). This can involve several layers of corporate structure, each represented by its own set of documents in different formats and sometimes different languages.

Manual KYB is slow because analysts have to read PDFs, scanned registry extracts, and handwritten shareholder agreements, then manually reconcile names and percentages across documents. Document AI speeds this up by extracting key-value pairs (company name, registration number, ownership percentage) directly from unstructured filings and flagging inconsistencies, such as a shareholder register showing 40% ownership while a separate declaration claims 30%. This is where cross-document validation becomes central to KYB: no single document tells the whole ownership story, so the system has to compare claims across files and surface contradictions for a human to resolve.

Why it matters for document processing

KYB sits at the intersection of regulatory pressure and document chaos. Financial institutions, payment providers, and increasingly B2B SaaS companies must run KYB checks before opening accounts, and the underlying documents rarely arrive in a consistent format. A UK certificate of incorporation looks nothing like its French or Singaporean equivalent.

This makes KYB a strong test case for extraction systems: high stakes, unpredictable layouts, and a legal requirement to catch discrepancies rather than paper over them. Teams building KYB automation need reliable key-value extraction, some ability to handle unseen document types without retraining, and a human-in-the-loop step for the ownership chains that genuinely are ambiguous. Getting this wrong has consequences beyond bad data: regulators can fine firms for inadequate due diligence, and onboarding a shell company tied to sanctioned individuals is a real operational risk, not a hypothetical one.

Related terms

• KYC (Know Your Customer)
• KYC onboarding
• Cross-document validation
• Key-value extraction
• Human-in-the-loop (HITL)

FAQ

How is KYB different from KYC?

KYC verifies an individual's identity, usually a customer opening a personal account. KYB verifies a business entity: its registration, ownership structure, and the real people who control it. In practice, KYB almost always requires running KYC checks on each identified beneficial owner, so the two processes overlap rather than run separately.

Can KYB be fully automated?

Most of it can. Document extraction, registry cross-checks, and sanctions screening are well suited to automation. But ambiguous ownership structures, unusual corporate forms, and edge cases in beneficial ownership thresholds still need a trained analyst to make the final call. Full automation without human review tends to create compliance risk rather than remove it.