Current large language models (LLMs) have proven useful for analyzing financial data, but most existing models, such as BloombergGPT and FinGPT, lack customization for specific user needs. In this paper, we address this gap by developing FinGPT Search Agents tailored for two types of users: individuals and institutions. For individuals, we leverage Retrieval-Augmented Generation (RAG) to integrate local documents and user-specified data sources. For institutions, we employ dynamic vector databases and fine-tune models on proprietary data. There are several key issues to address, including data privacy, the time-sensitive nature of financial information, and the need for fast responses. Experiments show that FinGPT agents outperform existing models in accuracy, relevance, and response time, making them practical for real-world applications.
@article{arxiv.2410.15284,
title = {Customized FinGPT Search Agents Using Foundation Models},
author = {Felix Tian and Ajay Byadgi and Daniel Kim and Daochen Zha and Matt White and Kairong Xiao and Xiao-Yang Liu Yanglet},
journal= {arXiv preprint arXiv:2410.15284},
year = {2024}
}