@heynavtoor: THE VIRTUAL ASSISTANT INDUSTRY IS DONE. Two students at ETH Zurich shipped an MVP in four days. Now AI controls their C…
Summary
Browser-Use, an open-source framework for AI-driven browser automation developed by ETH Zurich students, challenges the traditional RPA industry by offering free, self-healing capabilities that mimic human interaction without relying on brittle HTML parsing.
View Cached Full Text
Cached at: 05/11/26, 04:36 AM
THE VIRTUAL ASSISTANT INDUSTRY IS DONE. Two students at ETH Zurich shipped an MVP in four days. Now AI controls their Chrome browser. Clicks. Types. Scrolls. Books flights. Fills CRMs. Scrapes leads. Submits applications. While they sleep. No Selenium. No UiPath license. No outsourced VA. No Zapier wiring. The numbers are insane: → Filipino virtual assistant: $8 to $12 an hour → US virtual assistant: $25 to $50 an hour → UiPath enterprise license: $20,000 to $100,000 a year → Browser-Use: $0. MIT licensed. Runs on your laptop. The science is wild: It does not parse HTML. It does not write CSS selectors that break the moment a website redesigns. It reads the page the way you read it. Screenshot in. Action out. The AI sees a search box, types the city. Sees a date picker, clicks May 15. The website redesigns tomorrow. Nothing breaks. The grift opportunity is wilder: Manus AI is built on Browser-Use. So is half of the AI agent layer raising rounds in 2026. Freelancers are shipping browser automations to local realtors, recruiters, and lead-gen agencies for $500 to $5,000 per workflow. One developer plus one weekend equals a business. The founders are Magnus Müller and Gregor Žunič. Two engineers from ETH Zurich. Y Combinator W25 batch. They open sourced the project in October 2024. Within 3 months they crossed 40,000 stars. Felicis led their $17 million seed round in March 2025. Paul Graham backed it. They run a 7-person team out of San Francisco. Here is what it does: → Drive any website with a single English prompt → Works with GPT-5, Claude, Gemini, DeepSeek, Llama, any LLM → Persistent login sessions, no captcha walls, no 2FA loops → Headless or watch-it-work mode → Multi-tab, multi-step workflows in one prompt → Self-corrects when an action fails → Pip install in 60 seconds → Free hosted cloud at http://browser-use.com or run it on your laptop Here is the wildest part: The company that built the framework powering Manus AI gave the framework away. MIT license. Free forever. Fork it. Run it. Sell automations on top of it. The numbers, today: 93,144 stars. 10,546 forks. 315 contributors. 123 releases. v0.12.6 released April 2026. Built since October 2024. MIT license. A $14 billion RPA industry charges enterprises six figures a year. Two engineers from ETH Zurich gave the same capability to anyone with a laptop. Your forms. Your tabs. Your workflows. No subscription. Free and open source. (Link in the comments)
Browser Use - The way AI uses the internet
Source: https://browser-use.com/ We launched Browser Use Desktop: a new browser experience, fully open source.Check it out 👀Agents at scale. Undetectable browsers. The API for any website.
### Browser Harness Open-source, thin, self-healing harness. Give your agent superpowers to complete any task on the web.### Stealth Browsers Anti-detect, CAPTCHA solving, 195+ country proxies. Zero config.### Browser Use Box Your 24/7 Claude agent. Claude Code and Browser Harness in a remote box. Drive it from Telegram, the web, or SSH.### Web Agents Extract, automate, test, and monitor — in natural language.### Custom Models LLMs purpose-built for browser automation.### Proxies Residential IPs across 195+ countries. Rotating, undetected.
Similar Articles
@nikunj: Man, /goal is just AGI if given the right tools.. Like what do you mean you went through all the entire database of 2k+…
A user describes an AI agent that autonomously fixed product images, frontend bugs, and descriptions from a database, used browser automation and web search, and ran for two hours while the user met founders, highlighting impressive AGI-like capabilities.
Beyond the hype: I just watched an AI agent automate a 4-hour research workflow in 18 minutes.
The author describes observing an AI agent autonomously complete a competitive intelligence workflow that normally takes 4 hours in just 18 minutes, highlighting its ability to navigate websites, process PDFs, and iterate search, suggesting a move beyond simple chatbots to autonomous execution.
@svpino: I'm yet to see an agent running inside a browser that doesn't feel like a hack. I tried a headless browser, but I can't…
Santiago (@svpino) discusses the challenges of running AI agents inside browsers, and @ego_agent announces 'ego lite,' a kernel-level rebuild aimed at making AI agents faster and more reliable.
@rohanpaul_ai: Robotics is slow because every change needs physical setup, people, space, and repeated field runs. Physical AI needs t…
Antioch introduces Antioch Agent, a browser-based robotics simulator that lets developers test robot software in a closed agentic loop without physical hardware, accelerating development cycles.
@browser_use: This agent is an AI UGC startup
An AI startup has built an agent that reads any internet content and automatically creates and uploads viral TikToks, running 24/7 on a VPS.