Transforming Knowledge Management With Generative AI

Aboitiz Data Innovation faced a unique challenge: Design a wholesale architecture for a generative AI lab for a bank while ensuring accurate responses and maintaining strict information security protocols.

See Also: Safeguarding against GenAI Cyberthreats with Zero Trust

The new AI system can help revolutionize knowledge management in multilateral development banks, said Guy Sheppard, chief commercial officer at Aboitiz Data Innovation. The accuracy of the responses was a major concern, so the company ran Chat GPT-4, Mistral and Claude 3 in parallel to “assess response faithfulness and relevance, thereby maintaining user confidence and ensuring data accuracy.”

“The aspirations were not just for a proof of concept or a tightly bound document retrieval system. The challenge was to design a wholesale architecture for a generative AI lab, ensuring precise responses while maintaining strict information security protocols,” Sheppard said.

Generative AI solutions must be tailored to address industry-specific requirements, he said. To meet the bank’s knowledge management needs, the solution involved creating a centralized interface for both generic inquiries and specialized department-centric inquiries – ensuring that users could quickly access tailored information.

In this video interview with Information Security Media Group, Sheppard discussed:

• Running parallel large language models to maintain accuracy;
• Building a secure virtual private cloud to meet compliance needs;
• Ensuring widespread adoption by creating user-friendly interfaces.

Prior to joining Aboitiz Data Innovation, Sheppard led SWIFT’s Asia-Pacific financial crime compliance, intelligence and cyber teams. He has more than 20 years of experience in financial crime compliance, predominantly in analytics, entity resolution and correspondent banking KYC. He is a CyberEdBoard member.