While there may not be a universally accepted definition of an "AI agent" just yet, numerous startups are eager to develop "agentic" tools designed to automate various online tasks. One notable company, Browser Use, has garnered significant attention from both developers and investors with its innovative solution that enhances website "readability" for AI agents. In a recent update to TechCrunch, Browser Use revealed that it has successfully secured $17 million in seed funding, led by Astasia Myers of Felicis, alongside contributions from Paul Graham, A Capital, and Nexus Venture Partners. This funding round had not been reported prior to this announcement. As a participant in Y Combinator's winter 2025 batch, Browser Use has increased its visibility in recent months, especially after Chinese startup Butterfly Effect utilized its technology in the widely circulated Manus tool, boosting the company's profile even further. Founded last year by Magnus Müller and Gregor Zunic through ETH Zurich’s Student Project House accelerator, Browser Use arose from Müller’s extensive experience in web scraping and his collaboration with Zunic, whom he met while obtaining their master’s degrees in data science in 2024. Müller mentioned that they conceived the idea of merging web scraping with data science to enable a browser to execute tasks more efficiently. The duo developed a demo for Browser Use within just five weeks, which quickly gained traction, leading them to open-source the project. Essentially, Browser Use converts the buttons and elements of websites into a more digestible, "text-like" format for AI agents, facilitating better understanding and autonomous decision-making by these agents. Müller explained, “A lot of agents depend on vision-based systems and navigate websites through screenshots, which can lead to failures. ” He emphasized their approach converts websites into a format that agents can comprehend, allowing the same tasks to be performed repeatedly at a lower cost. As more AI companies aspire to enable their agents to interact with websites more seamlessly, Müller believes Browser Use has the potential to become a foundational layer that fulfills this demand.

He noted that over 20 companies in the current Y Combinator Winter batch have utilized Browser Use for their specific needs. “Companies are approaching us asking, ‘How can we simplify navigation for agents on our website?’” Müller remarked. He highlighted that certain sites, like LinkedIn, frequently alter their functionalities, posing challenges for agents. Myers from Felicis shared that the firm has been exploring the AI agent sector for several years and recognized Browser Use as a prime opportunity to expand their portfolio in this area. She credited the founding team's expertise and their open-source-first approach as pivotal factors in their decision. “We believe that web AI agents represent the next frontier, significantly enhancing the end-to-end automation of human tasks, ” Myers told TechCrunch. “[W]eb AI agents serve as a dynamic bridge between static, pre-trained models that are predominantly text-focused and the constantly evolving digital landscape. ”