台灣股票市場AI代理人(AI Agent)之創新商業模式與監理挑戰—以智動交易資訊公司之AI Agent平台為例

Abstract

本研究聚焦於人工智慧代理人（AI Agent）於台灣股票市場中作為輔助交易工具所面臨的商業模式建構與合規挑戰。隨著金融科技發展與生成式AI技術快速演進，新型AI代理人平台嘗試導入個股標的之自動盯盤與交易策略執行功能，協助一般投資人提升投資效率與操作便利性。然而，先前法規僅針對基金這類提供投資組合建議有制定所謂理財機器人專法，但對於個股交易流程的行為若應用AI技術尚未有明確具體可依循的專法，其適法性界定便成為第三方業者與主管機關溝通過程中的主要難題與監理挑戰。

本研究採用個案研究法、半結構式訪談以及結構式問卷，深入剖析國內致力打造AI Agent代理人平台之智動交易資訊股份有限公司，在尋求合規過程中的監理機關溝通回應與商業模式調整策略。研究結合主管機關政策演進整理、創辦人深度訪談、經營團隊與股東問卷，以及一般投資人使用調查，系統性盤點AI Agent於技術應用、商業運作、監理因應與市場接受四大面向的機會與限制。研究發現，目前以B2B2C模式與金融機構合作，將為最可行且具可擴展性的商轉路徑。

本研究最終提出一系列策略建議，包含主管機關在法規適用與產業輔導層面的調整方向，以及對平台創辦人在早期經營策略與未來國際布局的發展規劃，期能為台灣AI在金融交易場域的健康發展提供參考與助力。

This study focuses on the business model development and regulatory challenges faced by AI Agents used as AI-driven strategies and trading assistance tools in Taiwan's stock market. With the rapid evolution of fintech and generative AI technologies, emerging AI Agent platforms aim to enhance investment efficiency and operational convenience for individual investors by enabling automated stock monitoring and execution of trading strategies. However, while regulations have been established for robo-advisors that offer portfolio-based investment advice—primarily for mutual funds—there is still a lack of specific legal frameworks governing the use of AI technologies in individual stock trading. This regulatory gap poses significant challenges for third-party providers in navigating legal compliance and engaging with regulatory authorities.

This research adopts a case study approach, supplemented by semi-structured interviews and structured surveys, to examine the compliance journey and business model evolution of "Smart Trading Co., Ltd.," a domestic fintech company developing AI-based Agent platforms. The study integrates a comprehensive review of regulatory developments, in-depth interviews with the platform’s founder, and surveys of internal stakeholders (management and shareholders) as well as general investors. It systematically analyzes opportunities and constraints across four dimensions: technological application, business operations, regulatory response, and market adoption. The findings indicate that a B2B2C model—partnering with licensed financial institutions—currently offers the most viable and scalable commercialization path.

The study concludes with a series of strategic recommendations, addressing potential regulatory adjustments and industry support measures for authorities, as well as early-stage operational and international expansion strategies for platform founders. These insights aim to support the healthy development of AI-driven strategies and trading assistance applications in Taiwan.