Enterprise-level Web Agents are here! Let AI "move bricks" for your business and make a fortune!

Have you ever considered that the web is completely beyond human processing capabilities? Imagine a business tracking price changes, inventory updates, and competitor activity across thousands of websites daily. This data changes minute by minute, making it impossible for traditional browsers and manual operations to keep up. When I saw the news that TinyFish had just secured $47 million in Series A funding, I realized that this wasn't just a round of funding, but the beginning of a new era: the era of enterprise-grade web agents. I've long considered the commercial applications of AI agents, but TinyFish's approach opened my eyes to a more realistic and disruptive direction: enabling AI agents to go beyond simply simulating human web browsing and instead execute complex business workflows with enterprise-grade scale, reliability, and compliance. This round, led by ICONIQ Capital, attracted participation from renowned investors such as USVP, Mango Capital, MongoDB Ventures, and Sandberg Bernthal Venture Partners. Particularly noteworthy is that Sandberg Bernthal Venture Partners was co-founded by former Meta executive Sheryl Sandberg, whose involvement adds significant strategic value to this project. However, what truly intrigued me wasn't the funding itself, but the fact that TinyFish has been deployed at scale in production environments at Fortune 500 companies like Google, DoorDash, and ClassPass, running millions of operations monthly. This means they've crossed the chasm from demo to real business value, a rare achievement in the AI ​​agent space. The founding team's backgrounds are also impressive, showcasing a perfect blend of technical depth and business acumen. CEO Sudheesh Nair, formerly President of Nutanix, brings extensive experience in enterprise product development and marketing. Co-founder Shuhao Zhang is a former Meta engineer who participated in the development of GraphQL and has deep technical expertise in building large-scale systems. Co-founder Keith Zhai is a former Wall Street Journal veteran reporter, whose media background brings a unique perspective on information acquisition and analysis to the team. This combination of technical, business, and media experience gives them a unique perspective to understand and address the real needs of enterprises for network automation. As Shuhao mentioned in the interview, "marketing and positioning are truly the most difficult parts" of building a company, and Keith's media background perfectly complements this shortcoming. The Technological Evolution from AgentQL to Enterprise-Grade Web Agents Understanding TinyFish's development journey has given me a deeper understanding of their technological expertise. The company has been quietly developing for 20 months, though it's only now officially emerging from stealth mode. Their first product, AgentQL, laid the crucial technical foundation for enterprise-grade web agents. AgentQL solves a long-standing problem for developers: how to enable AI agents to accurately identify and manipulate web page elements. While developing AgentQL, Shuhao Zhang observed a key trend: "Twenty months ago, I did see a trend and shift toward a more agent-based world. Back then, it was still GPT-3.5, but I was really seeing the reasoning capabilities and the ability to handle complex tasks. So, what was really missing was a more AI-native way for AI agents to access the network." This insight is crucial. Traditional web automation tools rely on CSS selectors or XPath, which often fail when faced with dynamically generated class names and ever-changing page structures. AgentQL, on the other hand, allows developers to use natural language to describe page elements, such as "a red submit button" or "a card titled with a specific content." I particularly appreciate the design decisions in AgentQL's technical architecture. The decision to analyze pages based on the DOM rather than screenshots was driven by profound technical considerations. Shuhao explained, "Language model training datasets contain far more HTML and DOM data than images. Furthermore, screenshots have physical limitations. When you have long pages, horizontal and vertical scrolling, and content hidden behind accordions, screenshots present a whole set of limitations." This technical choice reflects the team's deep understanding of the capabilities of AI models. The success of AgentQL has laid a crucial foundation for TinyFish. It has been integrated into major AI frameworks such as LangChain, LlamaIndex, and LFlow, serving hundreds of millions of API calls. More importantly, its innovations in Chrome extensions and developer tools allow developers to verify the accuracy of queries before deployment. This "what you see is what you get" development experience significantly lowers the barrier to entry and improves the reliability of final deployment. From AgentQL to an enterprise-grade web agent, TinyFish demonstrates the technological evolution from a low-level tool to a complete solution. Why Traditional Web Automation Approaches Are Outdated I've observed a clear trend over the past few years: the web is becoming increasingly complex, while our methods for accessing and processing web data remain stuck in the past decade. TinyFish made a profound observation in its latest blog post: "The story of the web has always been a story of scale." From its initial days of a few static pages, to millions of websites searchable via Yahoo and Google, to e-commerce and social platforms, and eventually entire enterprises, the web has grown at a rate far exceeding our ability to process it. The web has become a maze. Information is hidden behind login screens, content changes with scripts and personalization, and prices adjust minute by minute. As TinyFish puts it, "The web has become a maze, impossible to grasp at human scale." Traditional web scraping tools and automation scripts are struggling to cope with the modern web landscape. Websites use dynamic loading, anti-crawl measures, and personalized content, all of which frequently render traditional methods ineffective. Furthermore, enterprise applications require accuracy, stability, and compliance far beyond the needs of individual users. Sudheesh Nair, founder of TinyFish, has a deep understanding of this issue. He notes, "Today's web spans thousands of platforms and billions of pages, but companies are unable to fully tap its potential because the work required to create business value at scale is complex, manual, and limited by human capabilities." This observation is strikingly accurate. I've seen too many companies expend significant manpower manually gathering competitor information, tracking market prices, and monitoring inventory changes—tasks that are not only inefficient but also prone to errors. In their technical presentations, TinyFish specifically emphasized the complexity of the modern web environment. Shuhao Zhang mentioned that "class names in modern web frameworks are generated dynamically. Some sites change everything if you refresh the page. Content is also dynamic. So nth-child will also change order when you have a new banner, and the carousel will change." This kind of technical detail shows why traditional CSS selectors and XPath methods are no longer reliable.

I particularly agree with a point made by TinyFish: the web has moved beyond the capabilities of the browser. They wrote in their blog post: "Creating new opportunities and revenue depends on thousands of workflows running on thousands of websites, with billions of changes every day. No human analyst can keep up. Consumer tools, those built for individuals, one browser at a time, were never designed to carry this weight." What modern enterprises need is not better browsers, but intelligent systems that can understand and adapt to the complexity of the web. TinyFish's Enterprise Web Agent Revolution TinyFish's approach caught my attention because, from the outset, they clearly distinguished the fundamental differences between consumer and enterprise web agents. As they point out in their analysis, "Enterprise web agents are fundamentally different from consumer browser agents." Consumer-grade agents excel at handling simple, one-on-one tasks like scheduling travel itineraries or providing personalized recommendations based on browsing history. Enterprise web agents, however, need to automate complex business workflows that must be executed thousands or even millions of times without fail. Their enterprise web agents possess several key characteristics that suggest a true technological breakthrough. First, they are designed for results: these agents aren't designed to demonstrate technical prowess, but rather to achieve measurable business results, such as revenue growth, cost savings, or increased market share. Second, they offer complete workflow coverage, handling every stage of the entire process, not just isolated tasks. Third, they offer enterprise-grade reliability and compliance, meaning they can meet the security, governance, and uptime requirements of large global organizations. What impressed me most was their "planetary scale" capabilities. TinyFish's web agents can coordinate actions across thousands of platforms simultaneously, a scale unattainable by traditional automation tools. Imagine a single agent simultaneously monitoring price changes across thousands of e-commerce websites worldwide, analyzing competitor promotions in real time, and integrating this information into actionable business insights. This isn't just a technological advancement; it's a fundamental shift in the way business intelligence is gathered. From a technical implementation perspective, TinyFish utilizes advanced reasoning models to understand and adapt to changes in the network environment. Their system uses advanced AI models for reasoning and exploration, then encodes this knowledge for high-speed, deterministic execution at scale. This approach combines the flexibility of AI with the reliability of traditional automation. More importantly, their infrastructure is able to learn, adapt, and expand, which means that the system becomes smarter and more reliable with use.

I particularly appreciate TinyFish's considerations regarding security and compliance. Enterprise-grade applications cannot bear the same risks of data leakage or compliance violations as consumer-grade products. TinyFish's Web Agent has a built-in enterprise-grade security posture and governance framework, ensuring that all operations have complete logging and audit trails. As they emphasize: "The TinyFish agent is specifically designed to operate at the scale, reliability, and compliance required by enterprises." This deep understanding of enterprise needs is the key reason why they have been able to successfully deploy in Fortune 500 companies.

In the design of its technical architecture, TinyFish has demonstrated a deep understanding of modern AI technology. In his technical presentation, Shuhao Zhang mentioned, "Advances in generative AI and newly released reasoning models have made networks more complex and harder for traditional tools to access." However, these same reasoning models provide TinyFish's enterprise web agent with the ability to understand and handle the complexity of today's networks, enabling companies to securely scale their operations and transform complexity into a business advantage. Real-World Business Cases Prove the Value Even the best theory needs to be proven in practice, and I'm impressed by TinyFish's performance in this regard. They've already achieved large-scale deployments at leading companies across multiple industries, demonstrating the real-world business value of enterprise-grade web agents. TinyFish currently operates hundreds of thousands of enterprise-grade web agents, executing millions of operations monthly for Fortune 500 and high-growth enterprises. This scale itself demonstrates the maturity of the technology and the authenticity of the market demand. In the hospitality industry, the web agent TinyFish developed for Google solved a long-standing technical challenge. Thousands of hotels in Japan used outdated reservation systems that couldn't directly integrate with Google's search aggregator. Traditional solutions required these hotels to upgrade their entire IT systems, which was costly and difficult to implement. TinyFish's web agent automatically aggregated these hotels' inventory information, allowing consumers to find and book these rooms through Google Hotel Search without requiring any infrastructure upgrades. This case study perfectly illustrates how enterprise-grade web agents can create new business value without disrupting existing systems. In the transportation sector, a leading ride-hailing company uses TinyFish to collect millions of pricing variables monthly, enabling near-real-time dynamic market adjustments. This capability enables them to quickly respond to competitors' pricing strategies and optimize their own pricing models, ultimately improving market competitiveness and profitability. Imagine the immense human resources required to manually collect and analyze this data, and the difficulty in ensuring its timeliness and accuracy. Applications in the e-commerce sector further demonstrate the power of web agents. Global brands can simultaneously track competitor pricing across thousands of retail websites, monitor inventory changes, and capture promotional data. This real-time market intelligence enables companies to quickly adjust their pricing strategies, identify new market opportunities, and avoid missing out on crucial business information. More importantly, the collection and analysis of this data is fully automated, significantly reducing operational costs. TinyFish's customer base is also expanding. In addition to tech giants like Google and DoorDash, growing companies like ClassPass are also using their services. This demonstrates that the value of enterprise-grade web agents isn't limited to large enterprises; mid-sized companies can benefit as well. Specifically in the retail and travel industries, TinyFish focuses on a core use case: dynamic price monitoring, helping businesses track competitor prices, promotions, delivery times, and inventory levels in real time. Abhi Shah, Director of Data Science at DoorDash, offered a particularly compelling assessment: "TinyFish's platform manages the complexity of network interactions at scale. Beyond DoorDash, TinyFish also powers high-risk workflows for hotels, e-commerce platforms, and marketplaces, helping them capture changing network data, act faster, and translate continuous change into measurable results." This endorsement from real users speaks louder than any technical demonstration. From a business model perspective, TinyFish's success lies in their focus on solving real business pain points rather than pursuing technological novelty. Traditionally, these tasks have been handled by large offshore teams performing manual data entry or by custom software scripts, which often fail when website designs change. TinyFish offers a more robust and scalable solution, leveraging an AI-driven approach to address the rapidly changing nature of the network environment. ICONIQ Capital's decision to lead this round of funding gave me a lot to think about. As a top-tier VC focused on growth-stage investments, ICONIQ's investments are often driven by deep strategic considerations. Partner Amit Agarwal, in explaining his investment decision, mentioned a key point: TinyFish has already been productized by large-scale customers who have the development resources to build similar systems themselves. "They've operationalized it, productized it, and deployed it at scale for two large customers who have all the internal development resources to build something like this themselves," Agarwal said. This observation is crucial. Technology companies like Google and DoorDash are fully capable of developing their own network automation tools, but they choose to use TinyFish's solution. This demonstrates that TinyFish provides value beyond simple technical implementation. I believe this value is primarily reflected in three aspects: specialization, scale, and continuous innovation. Specialization is reflected in TinyFish's deep focus on the enterprise web agent space. Rather than attempting to create a general-purpose AI platform, they specifically address the specific network automation problems faced by enterprises. The ICONIQ investment team highly praised TinyFish's technical capabilities. Amit Agarwal stated, "TinyFish's innovative enterprise web agent replicates human behavior on the network at scale, with the resiliency and reliability required by enterprises. This is laying the foundation for a major shift in how enterprises and applications interact with the network, gather intelligence, and automate workflows. No one else has solved this problem, and TinyFish is already delivering results in customer production environments today." Scale comes from their investment in infrastructure. Building infrastructure capable of supporting hundreds of thousands of web agents simultaneously requires a massive technical investment, making this investment uneconomical for most enterprises. TinyFish has already built such an infrastructure, boasting planetary-scale processing capabilities, creating a significant competitive advantage. Continuous innovation is perhaps the most important factor. The web landscape is constantly evolving, with new anti-bot technologies, new website architectures, and new security measures emerging all the time. The TinyFish team specializes in addressing these challenges, and their solutions will continuously evolve as the web landscape changes. For enterprise clients, this means they can focus on their core business without worrying about maintaining and updating their web automation tools. From a market perspective, investors believe now is a crucial time for the explosion of enterprise-grade web agents. The AI ​​agent space is experiencing a gold rush, with large tech companies and startups alike vying to capitalize on the shift from large, static language models to dynamic agents capable of performing complex, multi-step tasks. TinyFish has established a technological leadership and customer base at this critical juncture, positioning them well to capitalize on this rapidly growing market. The new funding provides TinyFish with three to four years of growth funding, enabling them to continue investing in product development and expanding their go-to-market operations. CEO Sudheesh Nair explicitly stated that their goal is not just to help companies save costs, but to "help them make more money." This focus on creating incremental value, rather than simply optimizing costs, is a key reason for investor optimism. Key Breakthroughs and Future Challenges in AI Agent Technology From a technical perspective, TinyFish's success is inseparable from recent breakthroughs in large language models and reasoning capabilities. Previous automation tools relied on hard-coded rules and scripts, unable to adapt to the dynamic nature of the web. Today's AI models possess human-like reasoning abilities, capable of understanding web page structure, adapting to interface changes, and handling exceptions. However, as TinyFish observes, "Advances in generative AI and newly released reasoning models have made the web more complex and more inaccessible to traditional tools." I'm particularly interested in how TinyFish addresses the core challenges facing AI agents in enterprise environments. First and foremost is accuracy. Consumer applications can tolerate occasional errors, but enterprise applications demand extreme accuracy. A single pricing error or missing data can result in significant business losses. TinyFish ensures operational accuracy and consistency through its patented infrastructure, which learns and adapts while maintaining enterprise-grade reliability standards. Scale is equally critical. While individual users may only need to manage a few websites at a time, enterprise customers need to monitor thousands of platforms simultaneously. This represents not just a quantitative increase, but a qualitative shift. Large-scale deployments require consideration of complex issues such as resource management, error handling, and load balancing. TinyFish's "planetary-scale" capabilities demonstrate their extensive expertise in this area. Their system uses advanced AI models for reasoning and exploration, then encodes this knowledge for high-speed, deterministic execution at scale. Looking at the implementation details of the technical architecture, Shuhao Zhang faced many interesting technical decisions during the development process. For example, in the development of AgentQL, they chose to use the DOM rather than screenshots to analyze pages, based on a deep understanding of AI model training data and technical limitations. They also developed a sophisticated pre-processing system to handle the complex structure of modern web pages, including technical details such as nested iframes and shadow DOM. Security and compliance are another key challenge. Enterprises' online behavior must comply with various laws and regulations, including data protection and antitrust laws. TinyFish's Web Agent has a built-in enterprise-grade security posture and governance framework to ensure that all operations meet compliance requirements. Shuhao Zhang, in an interview, specifically emphasized the security risks associated with handling user identity and authentication status: "I absolutely do not recommend that users share their sessions with remote browsers. This is a very gray area." He suggested that companies should create independent identity and authentication systems for AI agents. I also noted TinyFish's innovation in handling network complexity. Modern websites use a variety of techniques to prevent automated access, including CAPTCHAs, behavioral analysis, and IP restrictions. TinyFish's web agent is able to adapt to these measures and maintain stable access. This adaptability isn't a one-time effort; it's a continuous process of learning and improvement. They even developed a "stealth mode" to counter anti-crawler detection, circumventing these restrictions by simulating the fingerprint characteristics of a real browser. However, challenges remain. Shuhao Zhang admitted that they haven't yet found a perfect solution for complex scenarios like infinite scrolling: "By definition, it's infinite. You always need to slice and dice it to fit the context window, and you need to remember where you stopped and start again." This technical honesty demonstrates their clear understanding of the boundaries of technology and points the way for future technological development. Profound Impact on Enterprise Digital Transformation

I believe the enterprise-level web agent trend represented by TinyFish will have a profound impact on enterprise digital transformation. As they stated in their company blog: "If you can transform the internet into analyzable data, this will fundamentally provide enterprises with advantages that others don't have." Traditional enterprise information systems primarily rely on structured data and API interfaces, but a large amount of valuable information on the web still exists in unstructured form. Enterprise-level web agents provide a new way to access and utilize this information. This change is not only technically significant, but also strategically so. A company's competitive advantage increasingly depends on the speed with which it acquires information and its analytical capabilities. Companies that can obtain market intelligence faster and more accurately will gain a competitive advantage. TinyFish's Web Agent enables companies to monitor the entire market environment in real time, a capability that is enormously valuable in a rapidly changing business environment. As Sudheesh Nair puts it, their goal is to help companies "make more money," not just save costs. From a cost perspective, Web Agents also offer significant benefits. Traditional market research and competitive analysis require significant manpower and often lack real-time updates. Enterprise-grade Web Agents can operate 24/7, at a fraction of the cost of manual methods, while offering greater accuracy and consistency. This increased efficiency allows companies to devote more resources to core business and innovation. I'm particularly optimistic about the application prospects of web agents in areas such as supply chain management, risk control, and market forecasting. Supply chain management requires real-time monitoring of supplier status, price changes, inventory levels, and other information. Risk control requires timely identification of external factors that could impact business operations. Market forecasting requires analyzing large amounts of market data and trends. These are all areas where web agents can play a vital role. TinyFish is currently focused on the retail and travel industries, but their technology is readily applicable to other sectors. More importantly, web agents have the potential to transform the way companies access external information. Traditionally, companies have relied primarily on purchasing third-party data services or commissioning research firms. However, web agents enable companies to obtain the latest and most accurate information directly from the source, reducing intermediaries and improving the timeliness and reliability of information. This ability to directly access first-hand information will become a key source of competitive advantage for companies. TinyFish mentions a key point in its technological vision: "At its best, technology doesn't demand your attention. It fades into the background, making way for the vital work of humans." This philosophy reflects their deep understanding of the value of technology. The best enterprise-grade technology should be invisible, allowing users to focus on business goals rather than technical details. This is precisely the core value of enterprise-grade web agents. Challenges and Future Developments While I am extremely optimistic about the prospects for enterprise-grade web agents, this field still faces several significant challenges. The first is technical challenges. The online environment is constantly changing, and new anti-scraping technologies and security measures are constantly emerging. Web agents need to continue evolving to adapt to these changes. While TinyFish has made significant progress in this area, this is a never-ending technological race. As they say, "Transforming the complexity of the web from an obstacle into an opportunity." Legal and ethical issues are another significant challenge. While much of the web's information is public, automated access can still raise legal and ethical issues. Legal regulations regarding web crawlers vary across countries and regions, and companies need to ensure their practices comply with all relevant laws and regulations. TinyFish needs to strike a balance between technical capabilities and regulatory compliance. In particular, industry standards and best practices need to be established regarding data privacy and user identity protection. Intensified competition is also a real challenge. With the rapid growth of the enterprise web agent market, more and more companies will enter the field. Large technology companies may develop their own solutions, and specialized software companies may launch competing products. TinyFish needs to continue innovating to maintain its competitive advantage. However, as things stand, they have already established a significant first-mover advantage and technological barriers. From a team-building perspective, TinyFish faces typical challenges for tech startups. Shuhao Zhang mentioned in an interview that "the most difficult part for a founder is definitely positioning and business," reflecting the common challenges tech founders face in market development. However, co-founder Keith Zhai's media background provides significant reinforcement to the team in this area. I believe TinyFish's success strategy should focus on several key areas. First, continue to deepen its technological moat, particularly its ability to handle complex network environments and large-scale deployments. They need to maintain their technological leadership in AI reasoning capabilities, network adaptability, and enterprise-grade reliability. Second, expand their customer base from their current large enterprise market to the mid-sized enterprise market. Third, build an ecosystem and establish partnerships with other enterprise software vendors to make Web Agent part of a larger digital solution. From a product development perspective, TinyFish is evolving from low-level tools like AgentQL to a complete enterprise web agent solution. They plan to officially launch the company in the next month or two, at which time they may announce more product details. From a technical architecture perspective, they are building a complete technology stack, including runtime infrastructure, application-layer business logic processing, and observation, monitoring, and authentication systems. From an industry development perspective, I predict that enterprise-grade web agents will become a standard component of enterprise technology stacks. Just as enterprises today use CRM and ERP systems, future enterprises will widely use web agents to obtain and analyze external information. This market could reach tens of billions of dollars, providing enormous growth opportunities for early adopters like TinyFish. Ultimately, I believe TinyFish represents more than just a new technology solution; it represents a fundamental shift in how enterprises interact with the online world. In an era where information is the key to competitive advantage, companies that can better understand and leverage network information will gain a sustainable competitive advantage. As TinyFish puts it: "Focus on what impresses you. For everything else, there's TinyFish." Their $47 million in funding is just the beginning of this transformation; the real value creation lies ahead. Transforming network complexity into business opportunity is the core proposition of the enterprise Web Agent era.