FinPersona: An LLM-Driven Conversational Agent for Personalized Financial Advising

Takehiro Takayanagi, Masahiro Suzuki, Kiyoshi Izumi, Javier Sanz-Cruzado, Richard McCreadie, Iadh Ounis

Abstract: Conversational financial assistance is gaining increasing attention as a subdomain within the broader field of conversational AI, attracting interest from both the AI and finance communities. The application of conversational agents in finances is a particularly challenging subfield to explore, as financial scenarios require personalization and emotional support, especially during periods of market turmoil. In this paper, we present FinPersona, a Large Language Model (LLM)-driven conversational agent for personalized financial advising. FinPersona enables a non-expert user to explore investment options through conversation and dialogue driven by knowledge of the individual user and their circumstances. Through FinPersona, we aim to illustrate the salient features of a working prototype, as well as raise awareness of the many challenges unique to the financial domain. A video demonstration of the system is available at: https://youtu.be/tvJAYEg73Ro