距离ChatGPT的横空出世，已经过去了三年。 三年时间，市场对于AI的热情并没有丝毫褪去，反而赌注越下越重。但在资本持续加码的喧嚣背后，盲目的兴奋感正 在被一种具体的焦虑所取代：这波AI浪潮的终点到底在哪？什么才是落地的正确姿势？ 在大多数人还在仰望AGI（通用人工智能）的顶峰，或者迷失在对话框里的时候，未来式智能联合创始人兼COO邹阳 给出了一个冷静且独特的判断。 "沿着大语言模型这条路大概率不能登顶AGI。不过这不重要我也不关心。重要的是，在半山腰足够改变世界。" 在邹阳看来，半山腰的技术能力，已经具备了全部产品落地产生价值的可能性。AI真正的战场不在于做一个陪聊的聊 天机器人，而在于潜入产业流程，成为企业里那80%重复、高频、规则与判断聚合的脑力工作的"外接大脑"。 这种判断力，源自邹阳在AI行业深耕多年的体感。 邹阳算得上经历了AI 从 1.0 走向 2.0 的完整周期。从魅族 AI 实验室，到搜狗语音交互技术中心，再到阿里巴巴达摩 院，他的职业路径几乎与行业演进同步。 在达摩院期间，他负责的智能语音语义产品线连续多年在国内AI 云服务市场占据第一，属于那批最早把技术推向大规 模商用的人。 他见过人 ...

Core Insights - The emergence of enterprise-level AI agents presents a significant opportunity for businesses, as the demand for such technology is expected to grow rapidly [1][15][20] - The company aims to enhance productivity for knowledge workers globally by leveraging AI agents to streamline workflows and improve efficiency [2][31][38] Group 1: Company Vision and Strategy - The company, founded by Yang Jinsong, focuses on creating AI agents that can be easily integrated into business processes, reducing the barrier to entry for enterprises [2][20] - The vision is to enable 1 billion knowledge workers to work ten times more efficiently through the use of AI agents [2][31] - The company has completed multiple rounds of financing, indicating strong growth and interest in its business model [2][10] Group 2: Market Opportunities - The company identifies energy and finance sectors as primary markets for AI agent deployment due to their complex operations and strong payment capabilities [5][34] - The company aims to address the significant gap in AI expertise within enterprises, as only a small number of individuals have experience in training and applying large models [15][17] Group 3: AI Agent Functionality - AI agents are designed to enhance productivity by restructuring knowledge work processes, allowing AI to handle standardized tasks while humans focus on critical decision-making [6][38] - Specific examples include the development of a sports content AI agent that automates the process of gathering information, translating, and generating content, significantly reducing production time from hours to minutes [7][49] Group 4: Business Model - The company plans to avoid traditional software licensing models and instead pursue a results-based payment structure, charging clients based on the outcomes delivered by AI agents [8][68] - This approach aligns the company's interests with those of its clients, fostering a trust-based relationship and ensuring that the AI agents provide tangible value [70][71] Group 5: Competitive Landscape and Challenges - The company acknowledges the challenges posed by the current limitations of domestic AI models, particularly in reasoning capabilities, which are expected to improve significantly by late 2024 [24][76] - Despite these challenges, the company believes that China's rich application scenarios and data availability provide a strong foundation for AI development [73][74] Group 6: Future Directions - The company is focused on developing a platform called "Lingda" that simplifies the process of creating AI agents, enabling industry experts to easily replicate their knowledge into digital employees [62][63] - Future advancements include enhancing AI agents' memory capabilities and implementing reinforcement learning to allow agents to evolve and improve over time [80][81]