"早期我们曾尝试围绕大模型技术搭建平台赋能业务，但平台本身过重的开发投入、僵化的形式、业务 高昂的迁移成本使得业务难以快速使用，而没有使用就无法获得反馈来优化模型效果，模型就一直达不 到业务使用要求。"天弘基金人工智能部负责人平野坦言，"后来我们逐渐摸索出了以轻量化、服务化为 核心，嵌入业务系统进行赋能的大模型服务新形态。" 天弘自研FinAgent服务框架，设计了以"服务化"为核心的三层技术架构，在保证底层能力可复用的同 时，深度嵌入到每个业务环节中，目前已经成功落地20多个金融智能体，在包括投研在内的多个业务领 域已经产生了实实在在的价值。 人工智能发展如火如荼，金融行业人工智能的渗透尤为迅速。据中国人民银行刚刚发布的2024年度金融 科技发展奖项目名单显示，公募基金的项目历史上第二次进入一等奖名单，此次由天弘基金"基于大模 型的FinAgent金融智能体系统"摘得。 创新大模型服务形态，解放投研生产力 当前，公募行业纷纷探索大模型落地的场景，也有不少中小机构仍然踟蹰不前。而天弘基金在大模型应 用领域已经探索出了一条从1.0到3.0的渐进式发展路径。 AI技术在接手重复性、标准化工作的同时，也有助于释放投 ...

Core Viewpoint - The integration of AI in the financial industry is accelerating, enhancing research and decision-making capabilities for fund managers, as exemplified by Tianhong Fund's TIRD system, which has improved its performance in the equity market [2][5][6]. Group 1: AI's Role in Fund Management - Tianhong Fund's TIRD system provided timely sell signals during the human-shaped robot stock surge, aligning with the fund's internal assessments, demonstrating AI's effectiveness in risk management [5][19]. - The fund's proactive approach to AI integration reflects a shift in the public fund industry towards digitalization and improved investor returns rather than mere product scale [7][9]. - The TIRD system aims to streamline the research and investment decision-making process, ensuring that actions align with investment goals and styles, thereby enhancing overall fund performance [15][16]. Group 2: Challenges in Traditional Fund Management - The public fund industry has historically relied on individual fund managers' capabilities, leading to inconsistent performance and a lack of systematic investment strategies [9][10]. - Issues such as a lack of collaboration between research and investment teams, and the absence of a unified decision-making framework have hindered effective investment outcomes [12][11]. - The reliance on star fund managers has resulted in a short-term focus, making it difficult to implement long-term, sustainable investment strategies [11][12]. Group 3: TIRD System Features and Benefits - The TIRD system incorporates a digital platform for research and investment, allowing for real-time data analysis and decision-making, which enhances the efficiency of the investment process [15][17]. - Key functionalities include automated stock selection strategies, risk alerts, and performance tracking, which collectively improve the decision-making process for fund managers [17][19]. - The system promotes knowledge sharing and collaboration among team members, facilitating a more integrated approach to investment research and strategy development [15][21]. Group 4: Future of Asset Management - The future competitiveness of fund companies will hinge on their ability to leverage AI and digital tools for enhanced efficiency and depth in investment strategies [31][30]. - As passive investment strategies gain traction, the challenge for active fund managers will be to maintain their value proposition by consistently delivering alpha through improved decision-making processes [28][29]. - The industry is moving towards a model where human-AI collaboration will define success, rather than reliance on individual star fund managers [31][30].