Hong Leong Bank handles 67% of customer support queries with Gemini-powered AI chatbot and achieves 99% accuracy

As a 120-year-old institution serving millions of customers across Malaysia, Hong Leong Bank needed to modernize its digital banking experience while maintaining the human touch. Legacy systems created integration challenges, and high volumes of customer queries strained traditional support channels. The bank also needed to process trade operations documents efficiently, with submissions involving up to 600 pages across 20 documents each.

Hong Leong Bank built its AI ecosystem on Google Cloud with Gemini 2.5 Flash as the foundational model, deployed as a private instance on Vertex AI for data security. The bank launched AI chatbots for both customer-facing and internal support, Document AI for trade operations, and Gemini Code Assist for software development. Cloud Run handles high-volume AI and Document AI workloads, enabling reliable processing of large document submissions.

The Gemini-powered chatbot now handles 67% of customer support queries autonomously, achieving 99% AI accuracy. Development time accelerated by 25% through Gemini Code Assist. The platform processes document submissions of up to 600 pages for AI data extraction and classification. Over 300 data scientists and developers now work on the AI platform across the organization.

• Running LLMs as private instances on Vertex AI enables banks to maintain strict data governance while deploying cutting-edge AI.
• A unified AI platform approach (chatbots, Document AI, code assist) creates compounding efficiency gains across the organization.
• Using Cloud Run for high-volume AI workloads eliminates manual infrastructure management and enables 100% focus on AI development.