Incore Bank’s AI Hits 99% Accuracy on KYC Checks

Know-your-customer (KYC) checks are expensive partly because a bank isn’t simply verifying an identity. Compliance teams have to collect information from documents, reconcile it across systems, investigate potential risk factors and build an auditable record explaining how they reached a decision. Incore Bank, a Swiss B2B transaction bank, is testing how much of that work artificial intelligence agents can do before a human needs to make the final call.

That test took shape as a proof of concept built with Kyndryl and Google Cloud for customer onboarding.The system achieved up to 99% accuracy, automatically extracting data from onboarding documents, and showed potential to shrink a process that can take months down to days, according to the announcement.

“At Incore Bank, we believe innovation must go hand in hand with trust, transparency and strong regulatory governance,” said Mark Dambacher, CEO of Incore Bank.

AI Agents Divide Up the KYC Work

The proof of concept applies Kyndryl’s Agentic AI Framework with Google’s Gemini models to coordinate multiple AI agents across a bank’s internal data and outside sources. The agents split up the work: some extract and check customer information, others gather supporting evidence, others identify risk factors, and others generate an explainable risk score along with a record compliance staff can review later, according to the same announcement.

Google Cloud has described a similar multi-agent approach for KYC, with one agent checking a document’s consistency and validity, another cross-referencing outside sources through web search, and each agent handling one piece of the process rather than one system trying to do everything at once.

The Incore test adds a specific performance number to that kind of architecture: up to 99% accuracy extracting information from onboarding documents. That figure does not mean the AI made KYC decisions correctly 99% of the time. It measures how well the system pulled information out of documents, one input among several that feed into the actual risk decision, an important distinction as banks work out which parts of this process they’re comfortable handing to software.

Most Banks Haven’t Automated This Part Yet

Incore Bank’s test lands at a moment when financial firms are spending heavily on AI but have barely touched this specific job. Eighty-five percent of financial services and insurance firms plan to increase AI budgets over the next 12 months, with productivity gains cited by 65% as a top reason, PYMNTS Intelligence found. Yet only 20% of those firms use AI for KYC, know-your-business checks and identity verification specifically, the second-lowest adoption rate among the tasks tracked, trailing only A/B testing.

That gap is closing fast on paper. Sixty-five percent of firms plan to adopt or expand identity verification and KYC automation within the next 12 months, PYMNTS Intelligence found in research done with Plaid, trailing only reconciliation automation at 70% as the area drawing the most near-term interest. KYC automation adoption reaches 90% when longer-term plans are included, not just the next year.

Incore Bank’s test is exactly the kind of experiment that gap is pushing banks toward: not replacing the compliance officer but automating the evidence-gathering that used to eat up most of that officer’s time, while leaving the actual decision and the accountability that comes with it, to a person.

For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.

The post Incore Bank’s AI Hits 99% Accuracy on KYC Checks appeared first on PYMNTS.com.