Core Insights - The CEO of Superblocks, Brad Menezes, believes that the next billion-dollar startup ideas are hidden within the system prompts used by existing AI unicorns [1] - Superblocks recently announced a $23 million extension to its Series A funding, bringing the total to $60 million, focusing on tools for non-developers in enterprises [1] Summary by Sections System Prompts - System prompts are lengthy instructions (often over 5000-6000 words) that guide foundational models like OpenAI or Anthropic in generating application-level AI products [1] - Each company uses unique system prompts tailored to specific domains and tasks, which are not always publicly available [1] Sharing Insights - As part of launching its AI programming assistant Clark, Superblocks shared 19 system prompt files from popular AI programming products [2] - Menezes noted that system prompts constitute only 20% of the core technology, with the remaining 80% being "prompt enhancement," which includes additional instructions and actions taken during response generation [2] Components of System Prompts - The research on system prompts includes three parts: role prompts, context prompts, and tool usage [3] - Role prompts help maintain consistency in LLMs by providing a defined persona, while context prompts set the necessary background for actions [4] - Tool usage allows the model to perform tasks beyond text generation, guiding it in executing various functions [4] Market Opportunity - Menezes sees an opportunity for Superblocks to enable non-programmers to build applications by handling more tasks, such as security and accessing enterprise data sources [5] - The company has already secured notable clients, including Instacart and Paypaya Global [5] - Superblocks adopts an internal practice where software engineers are not allowed to write internal tools, promoting the development of user-driven intelligent agents [5]

Group 1 - The event titled "AI Making Money" is scheduled for June 10, 2025, in Hangzhou, targeting professionals from large companies, startups, and entrepreneurs [1] - The gathering aims to facilitate idea exchange, experience sharing, and networking among participants [1] - Registration for the event is limited and will close at 8 PM the night before, with a first-come, first-served policy [2] Group 2 - The company is currently recruiting for a new internship program [5] - The company is seeking creative entrepreneurs from the post-2000 generation [6]

Core Insights - The acquisition of Expensya by Medius is considered one of the largest startup acquisitions in Africa, with the deal valued at slightly over $120 million [1] - Founders Karim Jouini and Othmani have launched a new venture, Thunder Code, a generative AI-driven software testing platform that has secured $9 million in seed funding [1][2] - Thunder Code aims to address the inefficiencies of traditional manual testing through AI-driven "agents" that can simulate human tester behavior and continuously learn from feedback [3][4] Company Overview - Thunder Code was founded by the same team behind Expensya, leveraging their experience and expertise in AI to create a more efficient software testing solution [1][3] - The company has quickly gained traction in markets such as the US, Canada, France, and Tunisia, focusing on web application testing with plans to expand to mobile, desktop, and API testing by the end of 2025 [4][5] Investment and Funding - The seed funding round included notable investors such as Silicon Badia, Janngo Capital, and strategic angel investors from the AI sector [2] - Former employees of Expensya also participated in the investment, highlighting a strong belief in the new venture's potential [3] Market Potential - The software testing market is projected to exceed $100 billion by 2027, currently dominated by traditional platforms like Tricentis and BrowserStack [5] - Thunder Code's use of AI is expected to provide a competitive edge in a crowded startup landscape, allowing for rapid execution and value creation with fewer resources [4][5]

图片来源： a16z Z Highlights ： a16z (Andreessen Horowitz) 是一家风险投资公司，以其多元化的投资领域著称，被投资公司包括 Airbnb 、 Meta 和 Twitter 等。本次访谈由 a16z 于 2025 年 5 月发布， Guido, Yoko 以及 Matt 三位合伙人分享了关于传统编程和最近的 vibe coding 编程模式的看法。 AI 编程生态跃迁：万亿美元市场的效率革命 Matt Bornstein: 我们基本可以确定，目前编程是 AI 领域的第二大市场。如果我说错了请纠正 —— 纯聊天机器人应该排第一，编程排第二，这是单纯看数 据的结果。 Yoko Li ： 但消费者市场是很多不同领域的集合。 M att Bor nstein ： 完全正确。 Guido Appenzeller ： 这是我的市场定义方式。如果看真正同质化的市场，编程可能确实是第一。 Yoko Li ： 编程市场比陪伴型应用更大吗？ Guido Appenzeller ： 是的。 Yoko Li ： 我也这么认为。 Guido Appenzeller ： 现阶段确实如此。可能要看 ...

Core Insights - Toma, an AI voice startup co-founded by Monik Pamecha, has raised $17 million in funding, primarily from a16z and other notable investors, focusing on the automotive dealership sector [1][3]. Group 1: Company Development - Toma's AI voice assistant is now utilized by over 100 dealerships across the nation, assisting with service appointments, parts orders, and sales inquiries [3]. - The company initially focused on banking and healthcare but pivoted to the automotive sector after recognizing the less stringent regulations and high demand for efficient communication solutions [1][2]. - Toma employs a subscription model for its services, with additional fees for enhanced features as the AI handles more dealership operations [4]. Group 2: Market Insights - The automotive industry faces challenges in managing unpredictable call volumes, making it difficult for dealerships to maintain consistent service quality [3]. - Toma's AI training process involves 1-2 weeks of training on dealership customer calls to develop contextual understanding, which is crucial due to the significant operational differences among dealerships [3][4]. - The initial testing revealed a call answer rate of only 45%, highlighting the need for improved communication strategies within the industry [1]. Group 3: Funding and Growth - The recent Series A funding round was timely for Toma, as the company had only recently hired its first dedicated sales personnel after relying on the co-founders for operations [5].

Core Viewpoint - The article discusses the upcoming WWDC25 event and emphasizes the opportunity for participants to engage in discussions about Apple's ecosystem and monetization strategies through Vibe Coding [1]. Group 1: Event Overview - WWDC25 will take place on June 9, 2025, from 23:00 to 03:00 [5]. - The event will be held at Dongsheng Building in Beijing, with a limited capacity of 50 participants [7]. Group 2: Activities and Engagement - Participants will have the chance to watch the WWDC25 Keynote live and engage in a night snack party [6]. - An open mic session will allow attendees to share their app monetization experiences and insights on Apple's business prospects [3]. - Networking opportunities will be available during the event, fostering connections among like-minded Vibe Makers [2][6]. Group 3: Event Logistics - The event is free to attend, but registration is required due to the limited number of spots [7]. - Attendees will be provided with snacks and drinks to enhance the experience during the information-rich night [4].

Group 1 - Liam Fedus, a former OpenAI executive, is raising over $100 million to establish Periodic Labs, focusing on materials science AI, with a valuation of approximately $1 billion [1] - A group of lesser-known former OpenAI employees has also left to start a new company focused on reinforcement learning, backed by Benchmark, although the company's name remains undisclosed [1] - The trend of OpenAI employees leaving to start or join new AI companies has drawn comparisons to the influence of the "PayPal Mafia" [1] Group 2 - Former OpenAI executives and employees are leaving to focus on sectors like edtech, audio software, and advanced AI models, indicating strong investor interest in AI-related ventures, particularly those associated with OpenAI [2] - Thinking Machines Lab, founded by ex-CTO Mira Murati, is reportedly raising over $1 billion at a valuation of $10 billion, with terms still subject to change [2] - Over a dozen of the nearly 30 employees listed on the startup's blog previously worked at OpenAI, highlighting the trend of ex-employees forming new companies [3] Group 3 - Former OpenAI chief scientist Ilya Sutskever has raised several billion dollars for a new AI research lab, achieving a valuation exceeding $30 billion, making it one of the highest-valued AI startups globally [3] - Sutskever's company ranks just below xAI Corp. and Anthropic, both founded by early OpenAI associates, while ChatGPT's developer remains the highest valued at $300 billion [3]

Core Insights - The article discusses the evolution and differentiation of AI Agents, emphasizing the need to distinguish between genuine innovative companies and those merely capitalizing on the concept of AI Agents [1][19][22]. Group 1: AI Agent Framework - The operation of AI Agents is broken down into three layers: perception, decision-making, and execution, highlighting the importance of each layer in the overall functionality of AI Agents [10][14][15]. - The "white horse is not a horse" concept is introduced to analyze the diversity of AI Agents in the market, categorizing them into pure, neutral, and free forms based on their operational characteristics [17][18]. Group 2: Technological Evolution - The article identifies the internalization of Agentic capabilities as a necessary evolution for LLMs, with examples from OpenAI's o4-mini and Anthropic's Claude 4 showcasing different design philosophies [30][38]. - Engineering integration is increasingly contributing to model capabilities, with tools like Prompt Engineering revealing significant potential for Agent products [2][31]. Group 3: Multi-Agent Systems - The limitations of Single-Agent systems are discussed, including memory constraints and the complexity of tool interactions, leading to the conclusion that Multi-Agent systems are becoming essential for overcoming these challenges [79][80]. - Multi-Agent architectures offer advantages in complexity, robustness, and scalability, allowing for parallel exploration of solutions and improved adaptability to human collaboration [82][83]. Group 4: Future Directions - The article suggests that the future of AI Agents will involve a competition between "experience universality" and "deep reliability," with hybrid architectures likely becoming a common choice [40][41]. - The emergence of protocols like MCP (Model Context Protocol) and A2A (Agent-to-Agent Protocol) is highlighted as a significant development in facilitating communication and tool integration among AI Agents [61][70].

Core Viewpoint - The article highlights the explosive growth of Anysphere, a startup developing the AI programming assistant Cursor, which has raised $900 million and achieved an annualized revenue of over $500 million, leading to a valuation of $9.9 billion [1][2]. Group 1: Company Growth and Financials - Anysphere plans to announce on June 5 that it has raised $900 million to fuel its growth, with an annualized revenue surpassing $500 million [1]. - The company's valuation will reach $9.9 billion, significantly up from $2.5 billion reported in January 2023 [1]. - The recent funding round was led by Thrive Capital, with participation from Andreessen Horowitz, Accel, and DST Global, bringing total funding to over $1 billion [1]. Group 2: Product and Market Position - Cursor, launched in 2023, features an AI code editor that analyzes programmer actions and suggests code completions, along with a chatbot for code-related queries [1]. - The majority of Anysphere's revenue comes from subscription services, with a recent shift towards enterprise clients, now constituting an important revenue segment [2]. - Over half of the Fortune 500 companies are reportedly using Cursor in some capacity [2]. Group 3: Competitive Landscape - Anysphere's rapid growth has made it one of the most notable startups in the industry, with over 1 million daily users [3]. - The company is facing increasing competition, including a potential $3 billion acquisition of its competitor Windsurf by OpenAI [3]. - Anysphere's revenue surged, achieving a recurring annual revenue of $100 million in just 14 months, a record previously held by cloud security firm Wiz [3]. Group 4: Future Plans and Investments - The company intends to invest most of the new funding into research and development, focusing on improving and building proprietary AI models [4]. - Anysphere is accelerating the procurement of high-performance chips, such as GPUs, to support AI model training and operation [4]. - The company plans to expand its team, increasing from over 60 employees in April to more than 100 [5].

图片来源： Rosebud 日记应用 Rosebud 已获得 600 万美元种子轮融资，该应用致力于通过 AI 技术完善其作为自我反思与个人成长导师的功能。 本轮种子轮融资由 Bessemer Venture Partners 领投， 776 、 Initialized Capital 、 Fuel Capital 、 Avenir 、 Tim Ferriss 等投资方跟投。 Rosebud 运用人工智能技术分析用户的日记内容，识别长期行为模式，并提供见解、提问和指导。该公司表示其目标是像人类导师那样，帮助用户更深入 地了解自我、培养新习惯并追踪目标达成情况。 Rosebud 透露，自推出以来，用户已在平台累计书写 5 亿字，使用时长超过 3000 万分钟。 Rosebud 由 Y Combinator 校友、社交应用 Secret 联合创始人 Chrys Bader ，与加州大学伯克利分校认知科学毕业生 Sean Dadashi 于 2023 年共同创立。 二人在男性互助小组相识，基于自身接受心理治疗和教练辅导的经历萌生了创建 Rosebud 的想法。 " 每个人都是独特的，不仅使用不同的口头语言，更有着差 ...