Automate KYC and KYB document verification

KYB and KYC verification is a crucial part of customer due diligence (CDD) for bankers, insurers and lenders. It means checking the documents a customer or business submits (IDs, proof of address, bank statements, company registrations, ownership records) to confirm they can be trusted and that the details inside them are accurate and consistent, as they need to process loan applications, customer onboarding, wealth management, insurance claims, merchant due diligence and much more. With Staple, bankers, insurers and lenders can check every document, extract the data, verify it in context, focus only on exceptions and answer auditors with a full record.

Banks and insurers verify source of wealth and source of funds by requesting documents such as payslips, bank statements, contracts and asset sale records. And they analyse where a specific transaction came from (source of funds) and how the customer’s overall wealth was built (source of wealth). Staple checks every payslip, bank statement and proof of income for tampering and AI-generated content, categorises each transaction as income or not, tests whether that income is stable, and flags name and income mismatches against the applicant’s ID and application, so reviewers only handle the exceptions.

Banks, insurers and lenders detect a fake or AI-generated bank statement by checking the file itself for signs of editing like pixels, fonts and metadata. And they check the data inside it for consistency, like whether the opening balance, transactions and closing balance actually add up. Staple checks every bank statement for tampering and AI-generated content the moment it arrives, tags it with one of five trust labels (Provenance verified, Review required, No issue detected, Integrity issue detected and Unable to assess) and flags anomalies like statement totals that don’t match the credit and debit transactions, so reviewers only inspect the suspect files.

KYB (know your business) verification usually needs company registration documents, ownership and beneficial owner (UBO) details, tax and address proof and, for higher-risk cases, financial statements and bank statements. And most of these checks can now be automated. Staple catches tampered or AI-generated company registrations, balance sheets and ownership records on arrival, flags ownership mismatches, cross-checks figures like cash balances between the balance sheet and bank statement, and validates details against registries and credit bureaus, so your team only reviews the exceptions.

Yes. Document AI can now read non-Latin scripts, stamps and handwriting, but the fields it reads with low confidence still need a human check. Staple extracts data from IDs like Singapore NRIC, Malaysian MyKad, Hong Kong HKID, Chinese resident ID and Indonesian KTP in 300+ languages, including handwriting, rubber stamps, signatures and mixed-language fields. And it sends low-confidence documents to a reviewer before any data is committed.

Document fraud detection checks whether the file itself can be trusted, and KYC document verification checks whether the details inside it are correct and match the applicant’s other records. Staple does both in one pass. The Document layer checks whether each file is tampered or AI-generated, and the Data layer checks whether names, addresses, ownership and income match across documents, registries and credit bureaus.

No. Banks, insurers and lenders usually run different tools for KYC. Identity verification proves the person is real, often with biometrics and liveness checks, screening checks them against sanctions and PEP lists, and a KYC platform manages the case and the onboarding decision. Staple verifies the documents and the data inside them, without biometric or facial identity proofing, and pushes verified records into your existing onboarding system. So it adds to the KYC software you already run instead of replacing it.

Banks and insurers stay accountable for every KYC decision, so regulators expect human oversight of the AI and a record showing how each decision was reached. Staple sends low-confidence documents to a human reviewer before any data is committed, records every action by the AI and the reviewer with confidence scores, model version and timestamps, and seals that evidence so any later change can be detected.