Kyndryl, Incore Bank, and Google Cloud Automate Bank KYC with Gemini Agentic AI

Kyndryl, Incore Bank, and Google Cloud have successfully completed a proof of concept using agentic AI to automate the Know Your Customer (KYC) process. By leveraging Google’s Gemini models, the collaboration demonstrates how financial institutions can slash onboarding times from months to days while maintaining the rigorous auditability required by global regulators.

What was announced

The proof of concept (PoC) focused on one of the most persistent bottlenecks in the financial sector: the gathering, verification, and reconciliation of evidence for customer risk assessments. Traditionally, KYC processes rely on manual handoffs between teams, requiring professionals to cross-reference data from multiple unstructured documents, internal systems, and external sources. This manual intensity often extends onboarding timelines to several months, frustrating customers and increasing operational costs.

To solve this, the partners applied Kyndryl’s Agentic AI Framework alongside Google’s Gemini models. The solution introduces a semantic layer using Kyndryl’s "policy as code" capability and specialized guardrail agents. Unlike standard generative AI, this agentic approach uses multiple AI agents that work autonomously across structured and unstructured data to extract and validate customer information. These agents identify risk factors, provide explainable risk scoring, and generate comprehensive, auditable decision records for compliance professionals.

The results of the PoC showed that the solution could achieve up to 99% accuracy in automated data extraction from onboarding documentation. By automating these repetitive tasks, the system demonstrated the potential to reduce the total onboarding lifecycle from months to a matter of days. This allows compliance professionals to pivot their focus toward judgment-intensive cases and high-risk exceptions rather than administrative data entry, all while maintaining human-in-the-loop oversight and regulatory transparency.

"At Incore Bank, we believe innovation must go hand in hand with trust, transparency and strong regulatory governance. The proof of concept we completed with Kyndryl and Google Cloud shows how cloud-enabled agentic AI can help banks assess both prospective and existing customers more efficiently while maintaining the rigorous oversight and controls required in a regulated environment. By reducing manual effort and supporting consistent, transparent decision-making, the solution has the potential to strengthen customer risk management across the entire customer lifecycle."

Mark Dambacher, CEO Incore Bank AG.

The companies involved

Kyndryl is a global provider of mission-critical enterprise technology services, specializing in the design, management, and modernization of complex IT environments. The company frequently works with large-scale organizations to integrate emerging technologies like artificial intelligence into legacy infrastructure. Its proprietary Agentic AI Framework is designed to provide the governance and "policy as code" guardrails necessary for highly regulated industries like banking.

Google Cloud, a subsidiary of Google, provides the infrastructure and the Gemini family of generative AI models used in this initiative. The company has established a significant presence in the financial services sector, offering specialized tools for data analytics, security, and regulatory compliance. Its Gemini models are central to the "agentic" capabilities that allow for complex reasoning and data extraction in this PoC.

InCore Bank AG is a Swiss-based business-to-business bank that provides outsourcing, private banking, and digital asset services to other financial institutions and institutional clients. Operating in a jurisdiction known for strict regulatory standards, InCore Bank AG focuses on maintaining high levels of transparency and governance while adopting digital innovations to streamline its service delivery.

What FF News has reported before

FF News has closely tracked the rise of autonomous and agentic systems within the financial sector as banks move beyond simple chatbots. Recently, we covered how Starling Bank Debuts Agentic AI Assistant to Automate Small Business Banking and Combat Fraud, signaling a shift toward AI that can execute complex workflows rather than just answering queries. The integration of AI into regulated marketplaces was also highlighted when Sionic Debuts Instant Bank Pay and AI Fraud Detection on Microsoft Marketplace. Furthermore, the trend toward autonomous back-office functions gained momentum as Amex Ventures Invests in Fazeshift to Scale AI-Powered Autonomous Accounts Receivable. These developments reflect a broader industry move toward "agentic" solutions that handle end-to-end tasks with minimal human intervention.

What this means

This announcement marks a significant shift from "generative" AI to "agentic" AI in the back office. For years, KYC has been the "unsolvable" problem of banking, resistant to automation due to the messy, unstructured nature of compliance documents. By proving that agents can achieve 99% accuracy, this PoC puts traditional KYC outsourcing firms and legacy software providers under immense pressure. The competitive advantage for banks that can onboard institutional clients in days rather than months is massive. However, the industry now faces a critical question: will global regulators accept "policy as code" as a legal equivalent to a human compliance officer’s manual verification in a court of law?