新智元报道 编辑：KingHZ 【新智元导读】当AI的普及率和密度真正提升时，才是AI爆发的时代。AI的增长机会，不在于争夺顶端的0.5%，而在于满足99.5%普通人的需求与场 景。 在被称为「AI大年」的2025年，全球99.5%的普通人，仍然无法真正享受AI带来的生产力红利。AI的实际普及率和应用密度，远没有大众所想般乐观。 AI的资源与话语权，正加速向少数国家与科技巨头集中，并在用户层面演变为愈来愈烈的马太效应。 当我们在谈论AI的未来时，我们究竟在谈论谁的未来？ Bruce拥有新加坡国立大学AI博士、加州大学伯克利分校数学与计算机专业背景，曾师从图灵奖得主，先后在微软、领英等硅谷大厂任职，并在硅谷成功 创办过一款千万级下载量的社交产品。 作为一位活跃于以中美为主导的全球AI舞台、拥有跨国求学与创业背景的新加坡AI创始人，Bruce始终认为「技术不应是高墙，而应成为通向世界的桥 梁」。 2025年下半年，他创办了一款「全民AI」产品Agnes，在东南亚、拉美、中东等新兴市场迅速崛起。 App上线两个月新增注册用户达到300万，日活超过20万，增速一度超过早期的Instagram和Snapchat。产品在 ...

Core Insights - The article discusses the emergence of AI collaboration tools, highlighting that despite advancements, team collaboration remains inefficient and chaotic. The focus is on integrating AI into the collaborative process rather than merely enhancing existing tools like Office Copilot [1][2] - Agnes, a new AI collaboration product developed by a team of PhD graduates from top Singaporean universities, aims to redefine office collaboration by embedding AI into every step of the workflow, positioning AI as the "intelligent hub" of the team [1][4] Product Features - Agnes is designed to address real team needs such as task division, version management, and contextual memory, moving beyond traditional single-user efficiency tools to a multi-user collaboration framework [2][3] - The platform integrates all tasks, content, feedback, and versions into a single "AI-native collaboration workspace," eliminating the need for multiple tools and manual information transfer [3][9] - Agnes supports real-time collaborative editing, allowing team members to see modifications instantly, thus providing a smoother experience compared to traditional Office software [9] Technical Architecture - The underlying structure of Agnes includes a self-developed reasoning model (Agnes-R1) and a multi-agent framework (Code Agents) that facilitate complex task orchestration and context retention in collaborative environments [5][6][8] - The architecture is designed to manage structured processes, modular task breakdown, and complex workflow control, significantly reducing token consumption and improving accuracy in multi-step reasoning tasks [8][9] Market Position and Development - Agnes targets white-collar workers, students, and content creators, having already gained a user base and positive feedback in Singapore, where it is compared to "DeepSeek" [4] - The company maintains close collaboration with Singapore National University for research and development, focusing on AI trust protocols and multi-agent frameworks [4][11] - The product is still in development, with ongoing adjustments to enhance user experience, representing a shift from traditional office workflows to a new paradigm of AI-integrated collaboration [12][15]