Identity verification (IDV)
Digital identity verification confirms that a person is who they claim to be, typically by checking a government-issued document against a live selfie or a database record. Buyers across banking, mortgages, remittance, and brokerage call this IDV. It is the individual-identity check inside the wider KYC process, run once at onboarding and again whenever risk or circumstances change.
Digital identity verification sits at the start of onboarding, before a customer can open an account or transact, and it reopens whenever a risk trigger reopens the file. It is not the same as authentication, which confirms a returning customer is the same person who registered, usually through a password or biometric login rather than a fresh document check. It is also narrower than KYC, which layers risk assessment and ongoing monitoring on top of the identity check itself.
How does digital identity verification work?
Digital identity verification runs as a short sequence rather than a single check. The customer captures a document, usually a passport or driving licence, and the system reads the data on it, checks the document's security features, and compares the extracted details against what the customer entered. A live capture, a selfie or a short video, is then matched against the document photo to confirm the same person is presenting both.
The result feeds into the wider onboarding decision rather than standing alone. A clean match lets the customer continue immediately. A weak or failed match routes to a retry, a different document type, or a human reviewer, depending on how the institution has configured its risk appetite. Because digital identity verification runs in seconds, most legitimate customers never notice it; the friction only surfaces for the genuine edge cases, which is where a review step still matters.
What methods do digital identity verification checks use?
Identity verification checks combine a small set of methods rather than relying on one.
Document verification: the customer scans a passport, driving licence, or national ID card. The system reads the machine-readable zone or chip, checks fonts, holograms, and other security features against known templates, and flags a document that looks altered or expired.
Biometric and facial matching: a selfie or short liveness video is compared against the document photo. A liveness check confirms the person is physically present rather than holding up a photo or a screen, which is the specific weakness document-only checks cannot catch.
Database checks: the extracted details are cross-referenced against sources such as the electoral roll or a credit bureau record, confirming the person exists at the claimed address with a matching history, independent of anything they submitted themselves.
Institutions typically combine two or three of these depending on risk appetite and the products they offer, rather than treating any one method as sufficient on its own.
Identity verification vs KYC: what's the difference?
Identity verification is technically the document and biometric check that confirms a person's identity. KYC is the broader regulatory obligation: verify identity, assess money-laundering risk, and keep monitoring the relationship afterwards. IDV is one stage inside KYC, not a substitute for it.
In practice, the two terms overlap heavily in how buyers actually talk. Many buyers use "IDV" and "KYC" interchangeably, even when they mean the identity-document step specifically rather than the full regulatory process. The distinction matters most when scoping a project: an "IDV upgrade" might mean swapping a document-checking provider, while a "KYC upgrade" more often means changing the risk assessment and monitoring built around it.
How AI agents handle identity verification failures
A failed check is the point most identity verification flows lose customers, because a rejected document or a failed selfie match typically stops the flow with no explanation beyond "try again." An AI agent changes what happens at that moment rather than the check itself.
When a capture fails, the agent diagnoses the likely cause, such as glare across the document, a cropped edge, or an expired date, and tells the customer specifically what to fix rather than repeating the same instruction. If a database check comes back inconclusive, the agent requests the next-best evidence, a recent utility bill or a different document type, and resumes the case the moment the customer replies. Genuine high-risk cases, a suspected fraud pattern or a persistent mismatch, still route to a human reviewer with the case history attached, rather than being resolved by the agent alone. Gradient Labs runs identity verification failure-handling within a set of more than 20 pre-built financial-services guardrails, so the boundary between what the agent resolves and what it escalates is fixed in advance rather than decided case by case. The same pattern applies across onboarding: keeping your existing IDV provider and automating the work around it.
Frequently asked questions
Is digital identity verification the same as KYC?
No, though the two terms are often used interchangeably. Identity verification is the document and biometric check that confirms who someone is. KYC is the wider obligation that includes identity verification alongside risk assessment and ongoing monitoring. Many buyers who mean IDV say KYC instead, so the safest approach is to clarify which one a conversation is actually about.
What documents does identity verification accept?
Most providers accept a passport, driving licence, or national ID card, and some add a residence permit or a utility bill for address verification. Which documents are accepted depends on the provider and the jurisdiction, since document formats and the databases available to check them vary significantly by country.
How long does identity verification take?
A clean document and biometric match typically completes in under a minute. The time comes from capture quality rather than the check itself: a blurry photo, poor lighting, or a damaged document adds retries. Where a database check or a human review is triggered, resolution can take from a few minutes to several days.
Can identity verification be automated end-to-end?
The document and biometric checks themselves have been automated for years through specialist providers. What has typically stayed manual is everything around a failure: chasing a customer for a clearer document, reading an edge case a provider cannot parse, and deciding what happens next. That surrounding work is what an AI agent takes on, while the underlying provider check stays the same.