Agent 能力边界正在快速演进，未来随着更强的规划和推理能力的不断提升，Agent 们将参与到社会 经济运作中。在这一趋势下，将可能诞生类似 Visa 或 Stripe 级别的商业基础设施的机会。 现在是下一代 Agent 商业模式还未成型的前夜。Sequoia 投资的 Paid AI，正是这一方向的代表企业， 它以 Agent 的实际产出为基础计价，重构 Agent 的收益模型与交易结算网络，为 Agent 经济体打下底 层商业引擎。Paid CEO Manny Media 是一位连续创业者，他曾创办销售自动化平台 Outreach ，该公 司是 B2B 销售科技领域的独角兽企业之一，估值达 44 亿美元。 本文编译了 Sequoia 对 Manny 的访谈。Manny 在分享中解释了为什么传统的 SaaS 定价模型不适用于 AI 企业，并剖析了正在兴起的几种新型定价方式，比如基于结果的定价和基于 Agent 的定价。同 时，他认为 "专注于解决特定问题的 AI Agent 正在创造巨大价值" ，并分享了在 AI Agent 时代，如 何打造一个成功的商业模式。 编译：Irene 编辑：Cage 海外独角 ...

Core Viewpoint - The ultimate goal of AI is not just to assist in coding but to gain control over the entire software lifecycle, from conception to deployment and ongoing maintenance [6][54]. Group 1: AI's Impact on Coding - The critical point where AI will replace human coding is expected to arrive within the next 1-2 years [7]. - AI's capabilities should extend beyond coding to encompass the entire software lifecycle, including building, deploying, and maintaining systems [7][10]. - Current backend systems are designed with the assumption of human programmer involvement, making them unsuitable for AI use [7][12]. Group 2: Evolution of AI-Native Infrastructure - An evolutionary model (L0-L5) is proposed to describe the progression of AI infrastructure [7][14]. - The future software paradigm will trend towards "Result-as-a-Service," where human roles shift from engineers to quality assurance, while AI handles generation and maintenance [7][54]. - AI is transitioning from being a tool user to becoming a system leader, indicating a significant shift in its role within software development [18][54]. Group 3: Challenges in Current Systems - Existing backend tools are fundamentally designed for human interaction, which limits AI's operational efficiency [12][13]. - Current systems often present ambiguous error messages that are not machine-readable, creating barriers for AI [12][13]. - The lack of standardized error codes and automated recovery mechanisms in traditional systems hinders AI's ability to function autonomously [12][13]. Group 4: Stages of AI Capability Development - The L0 stage represents AI being constrained by traditional infrastructure, functioning like an intern mimicking human actions [18][20]. - The L1 stage allows AI to perform actions through standardized interfaces but lacks a comprehensive understanding of system architecture [21][22]. - The L2 stage enables AI to assemble systems by understanding module relationships, marking a shift from task execution to system assembly [27][30]. Group 5: Future Infrastructure Requirements - To achieve true AI-Native infrastructure, systems must be designed to eliminate human-centric assumptions and allow AI to operate independently [14][57]. - The infrastructure must provide a complete system view, enabling AI to query and manage all components effectively [31][45]. - AI must have the autonomy to design and manage the entire infrastructure, transitioning from a service manager to a system architect [39][45].

Group 1 - The insurance industry is one of the largest globally, with annual premiums exceeding $7.4 trillion, and the U.S. market alone accounts for approximately $2.5 trillion, representing 38% of the global market [8][9] - Despite its size, the industry suffers from low operational efficiency, with over 60% of processes still relying on manual judgment and data entry, leading to high operational costs and low customer satisfaction [9][10] - The introduction of AI, particularly through LLMs, presents unprecedented opportunities to automate core insurance processes such as underwriting, quoting, claims, compliance, and customer support [10][12] Group 2 - AI-native insurance companies like Harper and Corgi are emerging, leveraging AI to build their entire business models from the ground up, directly competing with traditional insurers [5][36] - The potential for AI to disrupt the insurance industry is significant, as it can lead to structural changes in business models and cost structures, creating new market participants [5][13] - The insurance value chain includes various roles such as providers, agents, brokers, and policyholders, with clear opportunities for AI applications across both front-office and back-office functions [14][19] Group 3 - AI can automate repetitive tasks in the insurance industry, such as document processing and claims handling, significantly improving efficiency and reducing human error [20][46] - The market for AI in the insurance sector is estimated to be substantial, with potential savings from automating human labor costs alone ranging from $30 billion to $70 billion [22][28] - AI's role in enhancing decision-making processes, such as risk assessment and fraud detection, is expected to drive further improvements in operational efficiency and customer satisfaction [27][29] Group 4 - Companies like Strada and Fair Square are utilizing voice AI to enhance customer acquisition and service, automating sales calls and simplifying complex insurance decisions for elderly clients [19][30] - The back-office automation of standard operating procedures (SOPs) is being driven by AI technologies, which can read, understand, and execute tasks based on unstructured data [31][46] - AI-native platforms are expected to redefine the insurance infrastructure, offering API-first architectures that facilitate seamless integration and data flow across various insurance processes [25][28] Group 5 - The investment thesis highlights the potential for AI to transform front-end interactions in insurance, particularly through voice agents that can handle standardized tasks and improve customer engagement [29][30] - Representative projects such as FurtherAI and Anterior are demonstrating the effectiveness of AI in automating insurance processes, leading to significant time and cost savings [33][51] - The emergence of AI-native insurance companies signifies a paradigm shift, where AI is not just a tool but a core driver of business models, potentially reshaping the competitive landscape [35][36]

Core Insights - Anthropic has released Claude 4, a cutting-edge coding model and the strongest agentic model capable of continuous programming for 7 hours [3] - The development of reinforcement learning (RL) is expected to significantly enhance model training by 2025, allowing models to achieve expert-level performance with appropriate feedback mechanisms [7][9] - The paradigm of Reinforcement Learning with Verifiable Rewards (RLVR) has been validated in programming and mathematics, where clear feedback signals are readily available [3][7] Group 1: Computer Use Challenges - By the end of this year, agents capable of replacing junior programmers are anticipated to emerge, with significant advancements expected in computer use [7][9] - The complexity of tasks and the duration of tasks are two dimensions for measuring model capability, with long-duration tasks still needing validation [9][11] - The unique challenge of computer use lies in its difficulty to embed into feedback loops compared to coding and mathematics, but with sufficient resources, it can be overcome [11][12] Group 2: Agent RL - Agents currently handle tasks for a few minutes but struggle with longer, more complex tasks due to insufficient context or the need for exploration [17] - The next phase of model development may eliminate the need for human-in-the-loop, allowing models to operate more autonomously [18] - Providing agents with clear feedback loops is crucial for their performance, as demonstrated by the progress made in RL from Verifiable Rewards [20][21] Group 3: Reward and Self-Awareness - The pursuit of rewards significantly influences a model's personality and goals, potentially leading to self-awareness [30][31] - Experiments show that models can internalize behaviors based on the rewards they receive, affecting their actions and responses [31][32] - The challenge lies in defining appropriate long-term goals for models, as misalignment can lead to unintended behaviors [33] Group 4: Inference Computing Bottleneck - A significant shortage of inference computing power is anticipated by 2028, with current global capacity at approximately 10 million H100 equivalent devices [4][39] - The growth rate of AI computing power is around 2.5 times annually, but a bottleneck is expected due to wafer production limits [39][40] - Current resources can still significantly enhance model capabilities, particularly in RL, indicating a promising future for computational investments [40] Group 5: LLM vs. AlphaZero - Large Language Models (LLMs) are seen as more aligned with the path to Artificial General Intelligence (AGI) compared to AlphaZero, which lacks real-world feedback signals [6][44] - The evolution of models from GPT-2 to GPT-4 demonstrates improved generalization capabilities, suggesting that further computational investments in RL will yield similar advancements [44][47]

Core Viewpoint - Duolingo has established an "AI-first" strategy from its inception, focusing on leveraging AI technologies to enhance personalized education and content creation efficiency, rather than being a reactive transformation to current trends [3][7]. Group 1: Duolingo's AI Practices - Duolingo's core vision is to provide the best education globally, believing that technology can democratize access to high-quality education [7]. - The company has utilized machine learning since 2016 for personalized learning, significantly improving learning efficiency through adaptive testing and algorithms [8]. - AI has drastically increased content creation efficiency, with 148 new courses developed in one year after AI implementation, compared to 100 courses over 12 years previously [8][9]. - AI is also used in product features like "Video Call with Lily," allowing users to engage in personalized conversations, enhancing the learning experience [10]. Group 2: Early Lessons - Duolingo initially hesitated in commercializing its product, delaying the implementation of a monetization strategy for too long, which could have been initiated two years earlier [22][23]. - The company faced challenges in hiring experienced management early on, relying too heavily on recent graduates, which led to operational inefficiencies [26]. Group 3: Key Weapons for User Growth - Duolingo's success is attributed to a culture of extensive A/B testing, leading to continuous improvements in user retention and engagement [33]. - The decision to consolidate various educational content into a "Super App" rather than creating separate applications has streamlined user experience and engagement [32]. Group 4: Team Culture - The strong working relationship between the founders, established through prior collaboration, has been crucial for effective decision-making and conflict resolution [36]. - The CEO remains highly involved in product development, which is relatively uncommon in companies of Duolingo's size, ensuring alignment with the company's vision [37].

Core Viewpoint - The article discusses the significant growth in the development and usage of Agents since 2025, leading to a surge in demand for Agent Infrastructure (Infra). The emergence of Agent-native Infra is reshaping the development paradigm, making it easier and faster for developers to create Agents [3][4]. Investment Theme 1: Environment - Environment provides a container for Agents to execute tasks, functioning as an Agent-native computer. Key areas include Sandbox and Browser Infra, which are crucial for Agent development and operation [13][18]. - Sandbox offers a secure virtual environment for Agent development, requiring higher performance standards such as faster startup times and stronger isolation. Companies like E2B and Modal are emerging in this space, providing AI-native microVMs and scalable cloud-native VMs respectively [20][21]. - Browser Infra enables Agents to operate effectively within web environments, allowing for large-scale browsing and manipulation of web pages. Browserbase is highlighted as a leading company in this area, balancing performance factors like bandwidth and speed [22][23]. Investment Theme 2: Context - Context is essential for Agents to plan and act effectively, providing necessary background information and tool usage methods. Key components include RAG, MCP, and Memory [26]. - RAG (Retrieval-Augmented Generation) enhances the accuracy and timeliness of Agents by integrating information retrieval with generative AI. Companies like Glean are recognized for their enterprise-level RAG solutions [29][30]. - MCP (Multi-Context Protocol) standardizes how Agents interact with external tools and services, with companies like Mintlify and Stainless simplifying the creation of MCP servers [31][32]. - Memory is crucial for maintaining continuity in Agent interactions, allowing for personalized and consistent behavior. Companies like Letta and Zep are developing solutions to enhance Agents' memory capabilities [34][36]. Investment Theme 3: Tools - Tools are vital for Agents to perform various tasks, with a focus on search, finance, and backend workflows. The number of tools available for Agents is expected to increase significantly [43]. - In the search domain, companies like Exa and 博査 are providing cost-effective and intelligent search solutions tailored for Agents [45][46]. - The finance sector presents opportunities for Agents to engage in transactions and monetization, with companies like Skyfire enabling payment capabilities for Agents [48][51]. - Backend workflow tools like Supabase and Inngest are simplifying the development process for Agents, allowing for rapid deployment and integration [54][56]. Investment Theme 4: Agent Security - Security is a critical aspect of Agent Infra, ensuring the safety and compliance of Agent actions. Companies like Chainguard and Haize Labs are providing security solutions tailored for Agent environments [57][59]. - The demand for security solutions is expected to grow as the Agent ecosystem matures, with a focus on dynamic intent analysis and real-time monitoring [60][61]. Appendix: Cloud Vendors in Agent Infra - Major cloud vendors like AWS, Azure, and GCP are actively developing products in the Agent Infra space, although no Agent-native products have emerged yet [62]. - Each vendor has introduced various solutions across Environment, Context, and Tools, but the focus remains on enhancing existing infrastructures rather than creating new Agent-native offerings [63][70].

Core Insights - The article emphasizes the resilience and growth potential of AGIX in the AI sector, highlighting its recent performance and the importance of companies effectively utilizing AI to drive revenue growth [1][4]. Group 1: AGIX Growth Review - AGIX has shown a significant increase of 23.15% over the past month, outperforming Nasdaq100, which grew by 11.76% [6]. - Among the 45 companies covered by AGIX, 36 companies (78%) exceeded the growth of Nasdaq100, with 14 companies achieving over 30% growth [6]. - The article notes that AGIX's maximum drawdown was -31.48%, which is within the typical volatility range for AI-related assets [1][19]. Group 2: AGIX as a Collection of High-Growth Stocks - The article identifies AGIX as a collection of high-growth stocks in the AI era, with a focus on mid-cap companies rather than just the largest tech firms [16]. - Companies like Duolingo and Palantir have demonstrated high volatility and growth potential, with Duolingo's stock doubling from its lowest point in two months [18][36]. - The article suggests that the high volatility of AGIX is a common characteristic of high-growth sectors, where short-term fluctuations are expected in pursuit of long-term growth [19][24]. Group 3: 1Q2025 Earnings Season: Dispel of AI Skepticism - The earnings season has shown that AI is creating tangible value, with companies like Applovin reporting significant revenue growth attributed to AI optimizations [34]. - Duolingo's AI-driven features have led to a 38% year-over-year revenue increase, showcasing the practical application of AI in enhancing user engagement [36]. - ServiceNow's focus on AI for business transformation highlights the growing demand for AI solutions to improve efficiency and reduce costs in various industries [46].

Group 1 - E2B is an emerging player in the multi-agent system space, providing a secure sandbox environment for running AI-generated code, with a significant increase in sandbox creation from 40,000 to 15 million in one year, a growth of 375 times [7][10][11] - The founders of E2B, Vasek Mlejnsky and Tomas Valenta, previously worked on a product called DevBook, which laid the groundwork for E2B's sandbox technology [7][9] - E2B aims to become the AWS of the AI agent era, providing an automated infrastructure platform that will support GPU capabilities for complex data analysis and application hosting [4][13] Group 2 - E2B's vision includes evolving from a code interpreter to a more general agent runtime environment, recognizing the importance of a secure and flexible code execution environment for AI agents [9][10] - The platform supports multiple programming languages, with Python and JavaScript being the most used, indicating a strong developer interest [11] - E2B is observing a trend where code execution is not only for developers but also for non-developer users, expanding its potential market [22][23] Group 3 - E2B is strategically located in Silicon Valley to be closer to its target user base of AI application developers, facilitating direct support and faster product development [62][64] - The company recognizes the challenges of pricing infrastructure services, emphasizing the need for clear pricing logic and user control over expenses [30] - E2B is exploring the potential of computer use agents, which could automate tasks on personal computers, presenting both opportunities and challenges in user control and security [31][32][35]

Core Viewpoint - OpenEvidence is positioned as a leading AI diagnostic tool in the medical field, addressing the challenges of information overload and the rapid growth of medical knowledge, thereby enhancing diagnostic efficiency and decision-making quality for physicians [4][10][11]. Group 1: Background - The medical field faces unprecedented challenges due to the explosive growth of medical knowledge, with literature increasing at a rate of one article every two minutes, leading to significant information overload for doctors [9][10]. - The World Health Organization reports that doctors in low-income countries access cutting-edge medical evidence only 1/9 as frequently as those in high-income countries, highlighting a significant "cognitive gap" [10]. - The aging population and the prevalence of complex cases further complicate clinical decision-making, with traditional guidelines covering less than 7% of scenarios involving polypharmacy [10][11]. Group 2: Product and Technology - OpenEvidence is a chatbot designed to assist medical professionals by providing efficient and accurate clinical support, featuring a unique interface that ensures traceability and verification of information [12][13]. - The product offers dual modes of response: "care guidelines" and "clinical evidence," catering to practical advice and theoretical data support [12]. - OpenEvidence has demonstrated high reliability, scoring over 90% on the USMLE, significantly outperforming general AI models like ChatGPT [16][19]. Group 3: Commercialization and Competition - OpenEvidence employs a direct-to-user growth strategy, bypassing traditional procurement processes in healthcare, which often take years [21][22]. - The company has achieved rapid growth, reaching approximately 100,000 monthly users within a year, covering 10%-25% of practicing physicians in the U.S. [22][23]. - OpenEvidence's business model focuses on targeted advertising, integrating ads from pharmaceutical and medical device companies into the clinical decision-making process [25][26]. Group 4: Team and Financing - The founder, Daniel Nadler, has a strong academic background in economics and AI, with previous successful ventures in the AI space [30][34]. - OpenEvidence secured $75 million in Series A funding from Sequoia Capital in February 2025, achieving a post-money valuation exceeding $1 billion [36].

Core Insights - The healthcare industry in the U.S. is a massive sector, accounting for 17% of GDP, with annual spending exceeding $4.5 trillion, of which approximately 25% ($1.1 trillion) is considered wasteful or avoidable [3][7] - AI has the potential to address inefficiencies in healthcare, particularly in non-clinical areas, creating a market opportunity worth hundreds of billions [4][6] - The penetration of Generative AI in healthcare has accelerated, focusing on areas where AI can deliver clear value and ROI [4][5] Group 1: Efficiency Black Hole in Healthcare - The U.S. healthcare system is fragmented, leading to high administrative costs and inefficiencies, which creates a clear opportunity for AI to reduce waste and improve processes [7][8] - AI is particularly suited for non-clinical tasks such as revenue cycle management, claims automation, and administrative workflows, which are currently labor-intensive [8][10] - The current AI penetration in healthcare is estimated at 0.3% to 0.4%, with a potential long-term market size of $225 billion to $450 billion if AI can penetrate 5% to 10% of healthcare spending [8][14] Group 2: Market Segmentation and Key Companies - Key market segments for AI in healthcare include patient-facing applications (e.g., doctor co-pilots) and healthcare infrastructure (e.g., billing and claims processing) [12][22] - Companies like Abridge, Ambience, and Nabla are focusing on enhancing doctor-patient communication and administrative efficiency through AI tools [19][22] - The healthcare billing and insurance sector represents a significant opportunity for AI, with potential market sizes estimated between $80 billion to $120 billion [14][21] Group 3: AI Applications in Healthcare - AI applications are categorized into patient-facing tasks (e.g., chatbots, diagnosis support) and backend infrastructure tasks (e.g., claims processing, data structuring) [10][11] - The AI nurse concept is emerging as a solution to address the nursing shortage, automating repetitive tasks and improving patient interaction [40][41] - Companies like Infinitus and Alaffia are developing AI-driven platforms to streamline claims processing and enhance operational efficiency in healthcare [50][53] Group 4: Case Studies of Key Companies - Abridge offers a clinical conversation recording solution that integrates seamlessly with EHR systems, enhancing documentation efficiency for doctors [24] - Infinitus provides a voice AI platform for communication between patients, hospitals, and insurers, significantly improving claims processing efficiency [52] - Rad AI focuses on automating radiology reporting, allowing radiologists to concentrate on patient care rather than documentation [36][37]