Fubon Financial Holdings is advancing its use of artificial intelligence (AI) across its subsidiaries, reflecting a broader trend in the financial sector. According to a survey by the Financial Supervisory Commission (FSC), nearly half of financial institutions intend to implement or expand AI applications in the future. More information on this survey can be found at https://pse.is/8a2wrh.
In July, Fubon Financial Holdings established a "Generative AI Application Promotion Team" that brings together expertise from five subsidiaries: Fubon Life, Taipei Fubon Bank, Fubon Insurance, Fubon Securities, and Fubon Asset Management. Currently, almost 20 projects are underway, focusing on areas such as virtual assistants for staff support and workflow optimization. The company aims for full AI integration among employees.
A recent development is the launch of the “Knowledge Retrieval Engine,” which consolidates regulatory and product documents from four subsidiaries—life insurance, banking, property insurance, and securities. Using natural language processing (NLP) semantic search technology, it provides employees with a unified portal for document queries. In October, new features were added to enable image analysis within documents and rapid summarization of large datasets using generative AI. This tool also allows comparison across multiple documents and table generation. The engine has been rolled out to 12 frontline business units across subsidiaries.
Taipei Fubon Bank has operated an AI-powered voice customer service platform since 2021 that integrates wealth management and consumer finance services. Currently, over 80% of customer support is handled by AI automation. "AI voice analysis enables audits to evolve from sampling to comprehensive coverage, ensuring service quality and identifying potential risks to safeguard customer rights," according to the company statement. The system processes about 560,000 calls monthly—equivalent to the work capacity of roughly 145 people—and serves around 900,000 users each month via web text response services with a satisfaction rate of 93%. The bank plans to further develop AI solutions that combine big data with decision-making processes.
Fubon Life is incorporating generative AI technology into its operations. The company states: "The upgrade of the 'Underwriting Smart Assistant' utilizes generative AI to automatically summarize medical records, successfully reducing the time taken for individual underwriting from 50 minutes to 25 minutes." Its “AI Occupational Code Recommendation System” has processed more than 1.2 million policies since going online in 2023 with a reported success rate of 90%. Other ongoing projects include tools for product development and claims processing efficiency.
For customer service enhancements at Fubon Life: "We have introduced a ‘Text-based Customer Service Chatbot’ and leveraged Copilot technology to enhance knowledge base integration and response accuracy." Additionally, an AI voice recognition system now enables customers to apply for travel insurance using voice commands.
Fubon Insurance has improved its customer hotline through AI voice robots since introducing this technology in 2023. Six self-service functions are now available by phone—including claims inquiries and policy validity checks—resulting in faster communication for customers. In August this year, Fubon Insurance launched a fully automated “Smart Phone Interview” feature that handles required verification calls related to insurance applications and amendments.
At Fubon Securities, data governance supports expanding AI applications for more accurate customer segmentation and personalized communications via multiple channels like apps and messaging platforms. This year’s upgrades include leveraging generative AI models so their intelligent customer service system can better understand client needs rather than simply answering questions passively. An upcoming release will see an app featuring an investment assistant capable of providing Q&A services on investments as well as stock suggestions based on investor preferences.
Natural Language Processing (NLP), mentioned as part of these initiatives, uses machine learning techniques for analyzing textual data.