Browser Use Secures $17M to Enhance AI Navigation on Websites

The term "AI agent" might still be up for debate, but a bunch of startups are diving headfirst into creating "agentic" tools that automate all sorts of tasks online. One of these startups, Browser Use, is making waves in the developer and investor communities with its innovative solution that makes websites easier for AI agents to "read."

Browser Use recently announced to TechCrunch that it secured a hefty $17 million in seed funding, led by Felicis' Astasia Myers, with contributions from Paul Graham, A Capital, and Nexus Venture Partners. This funding round hadn't been publicized before.

As part of Y Combinator's 2025 winter batch, Browser Use has been turning heads lately. Its visibility skyrocketed when the Chinese startup Butterfly Effect used it in their viral Manus tool.

Magnus Müller and Gregor Žunič launched Browser Use last year from ETH Zurich's Student Project House accelerator. Müller, who had been tinkering with web-scraping tools for years, crossed paths with Žunič in 2024 while they were both pursuing their master's degrees in data science. They brainstormed the idea of blending web scraping with data science to prompt a browser to carry out tasks.

In just five weeks, Müller and Žunič whipped up a demo for Browser Use, which quickly gained traction. They then decided to open-source it.

At its core, Browser Use transforms the buttons and elements of a website into a more understandable, "text-like" format for AI agents. This transformation helps agents grasp the different options available and make decisions on their own.

"A lot of agents use vision-based systems and try to navigate websites via screenshots, which often leads to things breaking," Müller explained. "We convert websites into something agents can understand, allowing us to run the same tasks repeatedly at a lower cost."

The demand for AI agents that can interact more smoothly with websites is on the rise, and Müller believes Browser Use can serve as a "fundamental layer" to meet this demand. He noted that over 20 companies in the current Y Combinator winter batch are already using Browser Use for their projects.

"We've got companies asking us, 'How can we make it easier for agents to navigate our website?'" Müller said. "Take LinkedIn, for example; it's constantly changing how its website functions, which often causes agents to fail."

Astasia Myers from Felicis mentioned that the firm has been eyeing the AI agents space for years, and Browser Use seemed like the perfect fit to expand their portfolio. She highlighted the company's founding team and its open-source-first approach as key factors in their decision.

"We believe web AI agents are the next big thing in end-to-end automation of human tasks," Myers told TechCrunch. "Web AI agents act as a dynamic bridge between static pre-trained models, which are mostly text-focused, and the ever-changing digital landscape."