以下文章来源于Monolith砺思资本 ，作者MONOLITH 可以说，目前的 Agent 更多还是惊艳的 Demo，不是可以规模化的产品。 Monolith 砺思资本办了一场「After the Model」技术沙龙，聊了聊：Agent 离规模化落地还有哪些难题？ 在活动中，一个被反复提及的观点是： Agent 需要是一个可持续工作的系统，而非单次任务的跑通。 这意味着，光有「模型智力」是远远不够的。想跨过工程这条鸿沟，必须还要「死磕」这几个硬指标： 稳定性、高吞吐量、成本控制、精确的状态管 理。 以下是活动的一些核心 Insight，供从业者参考。 Monolith砺思资本 . Monolith砺思资本是一家投资管理机构，覆盖一二级市场。遵循研究驱动的投资理念，投资技术与创新驱动的科技、软件、生命科学和消费领域。 对于个人极客来说，OpenClaw 是有趣的。但对于企业和商业环境来说，问题立刻暴露：昂贵（烧 Token）、不可控（安全边界模糊）、存在隐私问题， 且难以协作。 OpenClaw （原名 Clawdbot ）爆火。 ⬆️关注 Founder Park，最及时最干货的创业分享 Founder P ...

Core Insights - The industry has been significantly transformed by the rapid emergence of Cowork, Clawdbot, and Moltbook, indicating a new era in AI development [2][10] - Founders are actively sharing innovative applications of Clawdbot, showcasing its potential in creating remarkable new products [3][10] - There is a sense of urgency and excitement among founders, as they are fully committed to exploring the capabilities of AI [4][5] Event Details - Founder Park will host a closed-door online discussion aimed at founders and builders who have actively engaged with the recent AI developments [6][10] - The event will focus on various topics, including practical applications of Clawdbot, the maturity of AI technology stacks, and the future evolution of Clawdbot [11] Community Engagement - The initiative encourages collaboration among active AI entrepreneurs to share experiences and insights on the evolution of AI products [8][11] - The event is scheduled for February 4th, 12:00-14:00 Beijing time, providing a platform for discussion among industry leaders [11]

Clawdbot（现改名为 OpenClaw） 体验下来，持久的记忆管理系统很是让人惊艳——全天候保持上下文，无限期地记住对话并在此基础上持续深化互 动。 关于 Clawdbot 的记忆管理系统，AI研究工程师 Manthan Gupta 写了一篇文章详细复盘了Clawdbot独特的记忆机制原理。 所有的记忆，都是用 MD 文档进行存档管理的。 文章转载自「新智元」的编译版本。 ⬆️关注 Founder Park，最及时最干货的创业分享 超 19000 人的「AI 产品市集」社群！不错过每一款有价值的 AI 应用。 邀请从业者、开发人员和创业者，飞书扫码加群： 进群后，你有机会得到： 在聊记忆之前，咱们先得搞清楚模型在处理每个请求时到底看到了什么： 「它不搞那种基于云端、由大公司控制的记忆，而是把所有东西都留在本地，让用户完全掌控自己的上下文和技能。」 01 Clawdbot 里如何定义上下文与记忆？ 1.1 如何构建上下文？ [0] 系统提示词 (System Prompt) (静态+条件指令) [1] 项目上下文 (引导文件: AGENTS.md, SOUL.md 等) [2] 对话历史 (消息, 工具调 ...

以下文章来源于ThinkingHumans ，作者约尔 Yor ThinkingHumans . A builder, an observer, an investor, and a writer, with a community of fun facts, opinions, and technologists. 文章转载自「ThinkingHumans」，作者贺夏雨。 去年一年，AI 应用不少，表面上收入、用户数、融资额这些指标都在疯涨，但实际上已经越来越难反映真实情况了。 12 月低，Manus 以数十亿美元被 Meta 收购的消息，震动了 AI 圈。为什么是 Manus 跑出来了？PMF 是如何找准的？ 站在年初，我们再来回头看，Manus 作为一个很具代表性的产品样本，是怎么一步步找到了自己的位置，到底是哪些用户在哪些场景下一直使用、一直付 费？ 这篇文章，用一种完全基于一手事实的分析和推演方式，拆解了 Manus 的真实用户行为。 可落地的 Skill 搭建方法 从一个想法或一套 SOP，拆解成真正能跑起来的 Skill 写在前面：为什么看 Reddit 和 Discord 在这次分析中，我刻意 ...

X 上的网友制作了一张 Clawdbot 在 Github 的标星增长趋势对比图，几乎是直线上涨，现在已经到了 98K。 Clawdbot 背后的开发者 Peter Steinberger，是一个特别有意思的老哥。iOS 生态系统最早的一批开发者之一，开发经验可以追溯到 iOS 2.0 时代。在开 源社区贡献了非常多项目。 于是，「退休」三年后，Steinberger 重新回来了。Steinberger 在最新采访中提到， 早在去年五月份，我就有了做个人 Agent 的想法，那会儿 GPT-4 被 GPT-4o 替代，当时觉得模型还不够好，但我当时又觉得，这事大公司未来几个月肯定会做。 可是等到了 11 月，我发现市面上还是没有我想要的个人智能体。 于是，我决定自己搞。这基本就是 Clawdbot 项目的起点了。 Clawdbot（最近改名叫 Maltbot），目前最热门的 AI 项目，不光是在 X 上「一夜成名」，在谷歌上的搜索量更是直接超过了 Claude Code、Codex。 2011 年，Steinberger 创立了 PSPDFKit，用 13 年时间把它从开发者工具搞成了 B2B 软件巨头。以超 ...

AI-Native 知识内容平台 ThetaWave AI 宣布近日完成数百万美金的 Pre-A 轮融资，BAI Capital、高瓴创投及 MBA Fund 联合投资，老股东奇绩创坛 跟投。 ThetaWave AI 在 2025 年下半年突破一百万美元 ARR，目前现金流收入已能覆盖公司开支，用户次月留存 80%+ ， 三月留存 65%+。 不管怎么看，ThetaWave AI 都属于已经找到了 PMF，并且跑通了增长闭环的「优等生」产品了。 创始人李文轩大一辍学创业，ThetaWave AI 算是他的第三款产品，对于他来说，一直关注的都是「AI 带来的无限个性化内容供给的可能性」。而 ThetaWave AI 未来想做的，是想成为「每个人默认的获取知识的入口」，成为「第一个能够定义好新一代内容平台生产和消费方式的公司」。 以下是 Founder Park 与 ThetaWave AI 创始人李文轩的对话，经编辑整理。 体验产品：https://thetawave.ai/ 采访 | 万户 编辑 | 夏天 ⬆️关注 Founder Park，最及时最干货的创业分享 超 19000 人的「AI 产品市集」社群 ...

Core Insights - The article discusses the emergence of independent memory layers in AI systems as a necessary evolution for enhancing user experience and operational efficiency in AI applications [5][21][23] - It emphasizes that traditional methods like increasing context length and using Retrieval-Augmented Generation (RAG) are insufficient for addressing the complexities of memory management in AI [4][11][12] Group 1: Importance of Independent Memory Layer - The need for an independent memory layer arises from the limitations of existing AI models in maintaining continuity and context across interactions, which is crucial for effective collaboration and task management [9][10][20] - Memory is identified as a key factor influencing AI agents, with a focus on user profile maintenance, cross-dialogue memory, and a deeper understanding of user needs [3][21][22] Group 2: Challenges with Current Approaches - Current approaches like extending context length and RAG are seen as inadequate, as they do not address the dynamic nature of real-world data and user interactions [12][14][15] - RAG is criticized for being a passive method that does not support long-term collaboration or memory evolution, leading to inefficiencies in user experience [16][17][18] Group 3: Requirements for Effective Memory Systems - A robust memory system must manage different types of memories, ensuring they are accessible, editable, and auditable, akin to how the human brain organizes information [24][27][28] - The architecture of memory systems should balance cost and efficiency, addressing storage and computational demands while ensuring seamless integration into AI applications [25][26][30] Group 4: Future of Memory Management in AI - The article predicts that memory management will evolve into a critical infrastructure for AI, enabling models to become more than just tools, but partners in user interactions [22][23][49] - The concept of memory as an asset layer is highlighted, suggesting that memory systems should be transferable, reusable, and governable across different AI models and applications [40][41][48]

Core Insights - The article highlights the rising popularity of the Skill concept, which has quickly surpassed other concepts like Agent and MCP, becoming a sought-after feature across various products [2] - Skill aims to package expert work SOPs into reusable resource bundles that are easy to share and implement, requiring no coding or complex workflows, thus lowering the barrier for creation [3] Group 1: Skill Development and Opportunities - The collaboration between Founder Park and Kouzi to host a Skill recruitment competition encourages sharing best SOP practices, promoting skills as a fluid exchange resource [5] - The competition features two tracks: Workplace and Marketing Creation, allowing participants to develop Skills that enhance efficiency and creativity in their respective fields [7][10] - The Workplace track focuses on transforming daily work methodologies and complex processes into reusable Skills to improve workflow experiences [9] Group 2: Skill Application Areas - The Financial Professional Analysis track aims to create Skills that meet core needs for consistency, auditability, and strong decision support, such as financial report analysis and industry tracking [11] - The Geek track invites developers to create Skills that emphasize visual impact, technical innovation, and interactive experiences [12] - Skills should support various scenarios, including dynamic visual libraries and creative coding interactions, enhancing usability and engagement [14][15] Group 3: Competition Rewards and Timeline - Selected Skills will be featured in the Kouzi Skill Store and receive traffic support for broader user reach, along with potential monetization opportunities for developers [17][18] - The competition runs from now until February 9, 2024, with a structured timeline for submission, evaluation, and reward distribution [23][26]

Core Viewpoint - Toco AI aims to integrate certainty and engineering into AI coding, emphasizing that writing good, maintainable code is what brings social value, rather than just writing code itself [2][11][28]. Group 1: AI Architect and Software Engineering - Toco AI introduces an AI architect to address software architecture and maintainability challenges, moving beyond traditional coding capabilities [3][12]. - The company believes that architecture determines the upper and lower limits of software development, and thus emphasizes the need for a structured approach to software engineering [6][10]. - Toco AI's product is designed to embed best practices and modeling methodologies into AI coding, ensuring that software development is both rigorous and efficient [11][12]. Group 2: Market Position and Differentiation - Toco AI differentiates itself from competitors like Cursor and Claude Code by focusing on end-to-end software engineering solutions that incorporate standardized modeling practices [18][19]. - The company targets small to medium-sized businesses (SMBs) that may lack the resources for a dedicated architect, providing them with tools to build robust software architectures [16][17]. - Toco AI aims to become a standard in AI coding, similar to how Spring and Kubernetes have established themselves in their respective fields [22][61]. Group 3: Future of Software Development - The company anticipates that as AI coding evolves, the focus will shift from merely writing code to understanding and designing software architecture, which is crucial for long-term sustainability [28][30]. - Toco AI believes that the complexity of software will continue to grow, necessitating a strong architectural foundation regardless of the software's size [23][24]. - The integration of AI in coding is seen as a paradigm shift, where the emphasis will be on business logic and architecture rather than the specifics of coding languages [30][31]. Group 4: Challenges and Opportunities - Toco AI acknowledges that while AI can assist in coding, the ultimate decision-making and understanding of business needs will always rest with humans [32][33]. - The company is aware of the potential for job displacement due to AI advancements but believes that this will lead to a shift in focus towards higher-level problem-solving and architecture [37][51]. - Toco AI is committed to fostering a collaborative market environment, encouraging other companies to participate in the development of AI coding solutions [49][50].

Core Insights - ByteDance has launched "Kouzi 2.0," which includes new features like a skill store and long-term plans, positioning itself as "Workplace AI, use Kouzi" [1] - The evolution of Kouzi from a development platform to a coding tool reflects a strategic shift towards Vibe Coding, allowing users to develop their own skills [1][2] Development Journey - The Kouzi project has evolved over the past two years, initially resembling an early-stage startup rather than a well-planned product under ByteDance [2] - The team initially aimed to create a platform enabling everyone to gain programming skills through AI but shifted focus to a no-code chatbot construction platform due to challenges with existing coding capabilities [4][5] - Early user growth was driven by novelty, but the team recognized the need for sustainable value beyond initial engagement [6] User Insights and Strategic Shifts - The team discovered that high-frequency user scenarios primarily came from internal enterprise needs, leading to a pivot towards workflow solutions [7][10] - The introduction of workflows, initially seen as less appealing, became a crucial element in enhancing user engagement and value delivery [6][7] Product Evolution and Features - Kouzi has transitioned from a tool-focused approach to a partnership model, emphasizing long-term user relationships and ongoing support [20][22] - The introduction of "Kouzi Space" and the skill store allows users to upload and download skills, creating a repository of capabilities that can be leveraged for various tasks [11][17] Future Directions - The focus on "Vibe Coding" and the integration of skills aims to establish Kouzi as a "technical partner" for white-collar users, enhancing their productivity and efficiency [14][23] - The company is positioning itself to meet the needs of users looking to build their own systems rather than merely consuming existing tools, aiming for a deeper engagement with its user base [23]