Kyndryl, Incore Bank and Google Cloud Test Agentic AI for KYC Risk Assessment

Kyndryl, Incore Bank and Google Cloud have completed a proof of concept using agentic AI to support bank customer onboarding and Know Your Customer (KYC) risk assessment.

Announced on 31 August 2026, the project used Kyndryl’s Agentic AI Framework together with Google’s Gemini models to automate parts of the customer-risk assessment process while preserving explainability, auditability and human oversight.

AI agents applied across onboarding and ongoing review

According to Kyndryl and Incore Bank, the proof of concept introduced a semantic layer built around Kyndryl’s policy-as-code capability and guardrail agents. Multiple AI agents worked across structured and unstructured data, internal systems and trusted external sources to extract and validate customer information, collect evidence, identify risk factors, generate explainable risk scores and create auditable decision records for compliance professionals.

The architecture was designed to support both initial onboarding and ongoing customer reviews, including monitoring for changes in customer circumstances. The companies said the approach is intended to reduce repetitive manual work without removing human control from regulated risk decisions.

Up to 99% extraction accuracy reported in the proof of concept

Kyndryl said the proof of concept achieved up to 99% accuracy in automated extraction of data from customer-onboarding documents. It also demonstrated the potential to reduce onboarding time from months to days.

The result is a proof of concept rather than a production-wide deployment, and the companies have not described it as a replacement for compliance professionals. Incore Bank said the project was designed around strong regulatory governance, transparency and human oversight, with automation focused on evidence gathering, consistency and risk assessment support.

Incore Bank is a Switzerland-based B2B transaction bank regulated by FINMA and provides services across traditional and digital assets.

The project illustrates how agentic AI is moving beyond document extraction into more integrated KYC workflows, where policy controls, risk scoring, evidence trails and ongoing review can be orchestrated within a single governed process.