Activity
8K followers
-
Gregor Žunič shared thisIn October 2024, I was supposed to start my master's at Berkeley. Instead, a mistake at the border gave me two free months. Because of the mistake, I suddenly had two months to kill. I had just met Magnus at a dinner. We didn't have a giant market analysis or a business plan. We just thought: what if we build something cool nobody else has built before? So we built the first version of Browser Use in just 4 days. Instead of trying to control an entire computer, we focused specifically on the browser. We didn't know what it would become. We were just two people experimenting. In January 2025, Browser Use became the 𝗳𝗮𝘀𝘁𝗲𝘀𝘁-𝗴𝗿𝗼𝘄𝗶𝗻𝗴 open source repository in the world. Exploration precedes conviction. You don't need to know exactly what the company is before you start building. You just need to think end to end and solve a real problem.
-
Gregor Žunič shared thisThe best sales strategy is to provide so much value that paying feels obvious. I dislike transactional, extractive selling. I don't like when companies approach me and everything they do is just trying to extract expected value from the interaction. There is a better model: provide genuine value. Help someone. Make the relationship useful on day one. When you do that, paying doesn't feel adversarial. Open source is one of the best examples of this. You can create enormous value and capture very little of it, sometimes nothing. I think you should be okay with that. Not every person you help needs to become a customer. Make yourself useful, the revenue follows 🤑
-
Gregor Žunič shared thisOur product research strategy for the API is very simple: do people complain? Not every product decision requires a 40-person user study. When you are building infrastructure, strong user pain finds a way to reach you. It comes through support tickets, Discord, DMs and emails. Every single one of those messages is a piece of product research. And if nobody is complaining about a specific surface? Sometimes it’s simply good enough. Your support inbox is your roadmap.
-
Gregor Žunič shared thisPeople constantly ask me for the secret sauce to get into Y Combinator. There is no secret sauce 😄 I applied to YC 4 times and got rejected 3 times. For the first three times, my mindset was wrong. I was optimizing my startup around winning approval from a gatekeeper, rather than finding evidence from customers. The moment you stop optimizing for it is when the work actually gets good. Stop trying to get into YC. Be fiercely ambitious instead. Think end to end, and build something customers actually love. Most founders come up with an idea, apply to an accelerator, and if they get rejected, they drop it. Building a generational company is about the work, not the accelerator.
-
Gregor Žunič shared thisBuild something people want. Our new repo hit 7.5k Github stars in 2 days. And it shows that people really want SPEED for browser use. The motto never fails. If you just have a product that people want for multiple years you WILL succeed.
-
Gregor Žunič shared thisThree times this year I stopped and thought: how the hell is this my life? 1. Fundraising. Completely insane. Somehow even more ridiculous than Silicon Valley. The parody undersold it. 2. Living in a hacker house with a sauna in the backyard. Some mornings I walk outside and think: what sequence of decisions led to this being my normal life? 3. Standing in a room listening to Bill Gates speak. Having the strange realization that I wasn't watching this world from the outside anymore. I was somehow in the room. A few years ago, none of this was remotely normal to me. Now the weirdest part is how quickly the completely absurd starts feeling ordinary. Building a startup is a very strange way to live.
-
Gregor Žunič shared thisBreaking: Browser Use + Jev = Ultrafast ⚡ Findings flights took 7s and cost only $0.0039 🤯 > new action space every step > DOM state space > small LLM fallback to type (this video is at 1x speed btw) Built a tiny open source browser agent. https://lnkd.in/gK2jGjMH
-
Gregor Žunič reposted thisGregor Žunič reposted thisMost AI agents can't see or interact with the application they live in. Try it yourself: type "how do i set this up" or "what does this error mean" into your favorite app's AI assistant. It will very likely not be able to help you. We just launched the Frigade Assist API to fix that. It's one tool call your existing agent can make to connect with your application's UI. When a user asks how to do something, Frigade will generate a personalized step-by-step guide right on the screen. For more conceptual questions, the tool also supports passing back text-based answers to the parent agent so there's no longer any need to do RAG on your Help Center (which likely is outdated anyway). Works with the Vercel AI SDK and anything else that can call a tool.
-
Gregor Žunič shared thisWas fun to give a talk!Gregor Žunič shared thisWe love our customers at OpenRouter 💜 Last night, we launched our new event series: In Production. The idea is simple: bring together the people actually building with AI and give them the stage to show what they’re building, how they’re building it, and how they’re using OpenRouter in production. A huge shoutout to our first three founders: - Om Buddhdev from Olam Labs (YC S26) - Gregor Žunič from Browser Use - Hai Ta from Userlens (YC P26) Thank you for getting on stage, opening up your products, and showing us what building in production actually looks like. And we had a special guest - because who can demo your product better than the CEO? Alex Atallah, Co-Founder & CEO of OpenRouter, took the stage to share some of the new things we’ve launched and how they can help builders get even more out of OpenRouter. The energy in the room was electric. So many incredible builders, great conversations, and new connections. This is exactly why we created In Production - less talking about what AI could do, more showing what people are actually building with it. Huge thank you to CapitalG and Jane Alexander for hosting us in their beautiful office. And, as always, to my partner in fun, Anastasia Crew.
Gregor Žunič replied to a comment
1w
Yep exactly!
Gregor Žunič replied to a comment
1w
Thanks Nitin, took me three rejections to learn
-
Gregor Žunič liked thisGregor Žunič liked this🎉 Portfolio Highlight: Browser Use 🎉 Browser Use is the infrastructure that lets AI agents use the web. The web was built for humans, and vision-based agents that rely on screenshots tend to break on layout changes and dynamic sites. Browser Use turns buttons and page elements into structured text that LLMs can reliably act on, so agents can click, type, and extract data with far fewer failures. Today it offers Browser Use Agents, which complete tasks end to end, and browser infrastructure that gives teams remote browsers for their own agents. The open-source project has more than 116K GitHub stars and 12.3M monthly downloads, teams at Airbnb, Amazon, and Anthropic use it, and its integration into Manus put it in the global spotlight. In 2025, the company raised a $17M seed round led by Felicis. Founders Magnus Müller and Gregor Žunič met while pursuing master's degrees in data science at ETH Zurich and started the company through ETH's Student Project House. Magnus had been building web scraping tools for years, and together they combined scraping with data science to let an LLM drive a browser. Their early prototype took off on Hacker News, and open-sourcing it pulled in developers from around the world. With deep knowledge of how the web is structured and a rare ability to rally a community fast, they are the right team to build the foundation for agents on the web. Backed by Y Combinator
-
Gregor Žunič liked thisGregor Žunič liked thisPostcard from San Francisco: My American Experience 🇺🇸🌉 Written by: Urša Z. Summarising my ASEF Junior Fellowship experience in just a few words? Easy: professional growth and the American experience. 🔬 At the lab (University of California, San Francisco): Working with Prof. Andrej Sali at UCSF was incredibly stimulating. My research focused on understanding how mutations in a specific synaptic protein could contribute to autism. Using molecular dynamics simulations, I analyzed how mutated vs. unmutated proteins behave in cellular conditions. Beyond MD simulations, I expanded my toolkit with numerous computational and machine-learning tools – skills that will be invaluable for my PhD and future research career! ⚾ Life in the Bay Area: These 10 weeks were about so much more than research. I got the full-spectrum American experience: • Celebrated 4th of July with fireworks over the Golden Gate Bridge • Attended a Harry Potter-themed baseball game 🧙♂️⚾ • Watched surfers in Santa Cruz & took road trips along the coast • Experienced Waymo self-driving cars & massive warehouse shopping! 🤝 The People: What truly made this summer unforgettable were the connections. From meeting 1st-to-4th generation Slovenians at the Statehood Day celebration to exploring SF in a Tesla, road-tripping with fellow ASEF fellows, late-night student pizza discussions in Berkeley, and ice cream runs with lab mates. Looking back, I couldn't have asked for a better summer. Read more about it here: https://lnkd.in/d-QcFtEG A massive thank you to ASEF and Prof. Šali’s lab for this amazing opportunity, and to everyone who made these ten weeks in California truly unforgettable! 💙
-
Gregor Žunič liked thisGregor Žunič liked thisLast week we brought 500 developers to Ljubljana. 🇸🇮 Astra AI x Vercel: one evening about where software development is heading. Some highlights: 🧑🏻💻 Lars Grammel, creator of the Vercel AI SDK, on Software Factories: what happens when agents write most of the code, and what engineers do instead. 🚀 I talked about Scaling Astra towards 100M users: how a small team builds for that kind of scale when agents do a big part of the work. 📺 Guillermo Rauch joined us on the big screen for a live Q&A with Klemen Selakovič. 🎙️ To close, Boris Besemer and I sat down for a panel on "The era of 100x engineers", moderated by Frenk Dragar. We talked about how agent-powered development is changing how we hire, how we build and what "senior" even means now. Slovenia has the talent. That room proved it. Now we need to believe it and dream bigger. This AI wave is our shot. We can ride it, and we must. Huge thanks to Vercel, our speakers, the Astra team who made it happen, and everyone who showed up. 🙏 See you at the next one.
-
Gregor Žunič liked thisGregor Žunič liked thiswe thought browsers needed agents, when it was really agents that needed a browser
Experience & Education
Other similar profiles
-
Gasper Matic
Gasper Matic
Laboratory of photovoltaicsand optoelectronics, Faculty of Electrical Engineering Ljubljana
Explore more posts
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content