We're sunsetting PodQuest on 2025-07-28. Thank you for your support!
Export Podcast Subscriptions
cover of episode $17M Raised to Disrupt Online Interfaces with AI

$17M Raised to Disrupt Online Interfaces with AI

2025/4/17
logo of podcast AI Education

AI Education

AI Deep Dive AI Chapters Transcript
People
J
Jaeden Schafer
Topics
我今天要谈论一家名为BrowserUse的AI初创公司,它最近筹集了1700万美元资金。该公司致力于改进AI代理与网站交互的方式,这将深刻影响未来人们的日常网络使用体验,而用户可能对此毫不知情。BrowserUse的核心技术在于将网站转换为易于AI代理理解的HTML文本,从而提高AI代理的效率和准确性。 当前,许多AI代理依赖视觉模型处理网页截图,这种方式效率低下且耗费大量计算资源。BrowserUse通过提供易于理解的文本和HTML,显著提升了AI代理的效率,解决了AI代理在处理复杂网站时遇到的难题。 BrowserUse的成功得益于其技术被中国初创公司ButterflyFX采用,这使其获得了广泛关注和认可。该公司由两位数据科学硕士生创立,其开源项目在GitHub上获得了超过47000颗星,展现了其技术的受欢迎程度。 未来,网站将需要进行AI代理优化,以吸引AI代理访问并提升转化率。BrowserUse将成为这一趋势中的关键工具,它将被广泛应用于各种网站平台,例如Shopify、Squarespace和WordPress等,并催生一个新的产业。这将改变网站优化策略,从目前的搜索引擎优化转向AI代理优化,从而帮助企业提升效率和盈利能力。 BrowserUse的投资者认为Web代理是未来自动化任务的关键技术,这预示着AI代理优化将成为一个巨大的新兴市场。

Deep Dive

Chapters
This chapter explores BrowserUse, an AI startup that raised \$17 million to improve how AI agents interact with websites. It explains the current limitations of AI agents using vision models and screenshots for web navigation and introduces BrowserUse's solution.
  • BrowserUse raised $17 million in funding.
  • The company improves AI agent interaction with websites by converting them into easily readable HTML text.
  • Current AI agents often use vision models and screenshots, which is slow and inefficient.

Shownotes Transcript

Translations:
中文

Today on the podcast, I want to talk about an AI startup that has just raised $17 million that you'll probably interact with every single day in the future, and you probably will never know. And the company is called BrowserUse. So something that's really interesting here is exactly what they're doing in order to enable AI agents to more easily scrape websites, run on websites, understand what's going on on the website, and you as a

that may be using these AI agents in the future won't even know why these AI agents are doing so much better on some websites versus others. And it's all thanks to browser use. I want to break it down because I think this is a really interesting concept. But before we get into that, if you've ever wanted to start or grow an AI or sorry, any sort of business or your profession using AI tools, I'd love for you to be a member of the AI Hustle School community. So every single week I release a deep dive video. I have over 50 that I've recorded already where I break down exactly

exact tools, AI tools, how to use them, how my strategies I'm using AI to grow and scale my current businesses. And all of that, I published an exclusive school community for $19 a month. In the past, I've had this at $100 a month. And there is a current discount. So if you lock that in, when I raise the price in the future, it won't be raised on you. But we have over 300 members. And I break down all of the growth hacks I'm doing with AI, my co-host, Amy, how he made $20,000 with one of his side hustles, how he's scaling that up with AI,

this year, all of the different things that we're doing, the numbers, the money, how it all breaks down, stuff we don't share publicly, it's all in the community. So if that sounds interesting to grow and scale your business or career with AI tools, go check out the A Hustle School community. The link is in the description. All right, let's get into browser use. So

The thing that I found really interesting about this is this is a Y Combinator company. This was part of their winter 2025 batch, fairly new. And I think right now, of course, we have a lot of people talking about AI agents and what they are. And I think a lot of people have different definitions of what an AI agent are.

The reason for that, it's a little skewed is because a lot of people want to claim what they're doing is, you know, we have an AI agent, even open AI with their operator, I would say, when in reality, it's not like 100% full on agents that are completely autonomous to accomplish a task. Okay, so we'll get past that. Regardless, there's a lot of startups that are building tools to make these AI agents, whatever their current form is, or whatever their form is in the future, where these things are actually autonomous.

able to make them better. So browser use is essentially a lot of developers are using this, a lot of people are interested. They're essentially creating a solution that makes it easier for the AI agents to use and read websites. And the way that they're doing this is essentially they are, they're essentially going and taking your website and turning it

turning it into HTML text that's viewable and easy to scrape and for an agent to understand. Right now, the best AI agent models, and you can think really, I think like Chatshepti Operator and like a handful of other ones, Cloud has computer use. What they're essentially doing is they're using vision models. So they take a screenshot of the page, they send it to the model,

They might even do some of this processing on their back end where they're looking at it, they're reading it, they're deciding where to click. They pretty much map out the pixels on the screen and are like, hey, click here, do this. Like that's how they're working. It's like screenshot, send it over, figure out what's going on, figure out what they need to do, send the location to click.

And then send it back and then it does that. So it's doing it really fast, but that's essentially how these work. Now, the problem is, well, one of the big problems is that processing of taking that screenshot, turning it into text, understanding the context of what's going on and accomplishing the task. It takes computer processing time. And if you think about it, if you think of a website that a lot of people might be interacting with, maybe there's like, you know, today, maybe there's 100 people that are using Checkbox.

chat GPT operator to go to LinkedIn and help, you know, make a spreadsheet of all of the people that send them, you know, friend requests or, you know, connection requests, and they want to say hi to all the people, right? I don't know, it's a random use case. So let's say that 100, or let's say like 10,000 people are doing that every day with current system, they have to run like a lot of the exact same flows and exact same screenshots and converting that over a

into what these AI models have to do 10,000 times a day. And in the future, you might say, well, maybe there's going to be 10 million times a day. And so there's just a lot of the post-processing that can essentially be completely eliminated and the whole process can be sped up because if these websites go and decide to automatically have kind of this AI layer tool that scrapes their site and makes it really easy to essentially just turns it into like easily readable, understandable text and HTML for these AI models,

then it'll speed up how fast the AM model can run. Anyways, this is essentially what browser use is doing. I think this is a really cool concept

So they've just raised $17 million. They have, the round was led by Felicious, Astia Myers. They also had Paul Graham, right? The founder of Y Combinator, A Capital, Nexus Venture Partners. All of this, all of them have just come in and it's kind of just reported what they've been able to do. So this is a new round of funding. And the reason this company got like famous or kind of started to get some notoriety is because

There is a Chinese startup, which is ButterflyFX, and they used browser use in their mainest tool. So this was a really big viral tool. And so when this kind of went off, everyone was like, hey, what is actually powering this tool? How does this thing actually work? And so because of that...

That's kind of when browser use got famous and whatever. So the way this whole company started was that Magnus Muller and Greg Zinnick, these are the founders. They founded this last year. They were working through ETH Zurich's student project house accelerator. So they're both working on

Well, I guess Mueller had been working on some web scraping stuff in the past, and then he met Zunich in 2024. They're both getting their master's degrees in data science. And of course, like every startup story,

You get these kind of geniuses in the same room together. And they decided that they wanted to create something that was web scraping and data science, stick them together. And I would assume that the AI agent angle kind of developed as they realized the use case, but they kind of got started on something. And apparently they built the first demo for browser use in five weeks. And after that, it really took off. They open sourced it. So if you go over and look at GitHub, this thing has

Over 47,000 stars on GitHub. So a lot of people are evidently super stoked about it. It has 5,000 forks.

And yeah, this is something that I think developers love. What I think we'll probably see more in the future is I can imagine like, right, like you can imagine most of the world is running on like WordPress. Most websites are built on WordPress today. And there's a lot of others, right? You can say like Squarespace and whatever, all those other platforms. I imagine at some point, there's just going to be a plugin that you can get for most of those platforms where you can enable it. And it will just have browser use run through your website and essentially create a version that AI agents can access really well. The reason why this is so critical is because

yes, it makes the agents faster and whatever. But the reason why website owners want this is because they're saying like, this is going to be, it's going to be like today you have search engine optimization to try to get more users, right? You're like, how do I optimize myself to the top of Google? I predict in the future, we're going to have to focus on agent optimization because we're not even going to be doing a lot of the Google searches or quote unquote Google searches, the online searches. We're just going to be telling our agent like, hey, you know, go buy me this thing and book this thing. And I want to, you know, like,

You know, you might say, hey, AI agent, I want to start a podcast next week. Please go buy me all the relevant gear and like have it all sent to me. Okay, so it has to go do research. It has to go find the gear. It has to find the best place to buy the gear. It's doing all that research in itself. And companies are going to want to optimize for getting the AI agent to come to their website. There's going to be all sorts of sneaky things that they do, essentially, and

to make that happen. But I think that this browser use is going to be one of those. It's like the AI agents like, oh, you know, maybe this other website was cheaper, but I didn't quite understand there's something convoluted there. I'm just going to go to this other website that seems much more straightforward and I can understand it better. So there's going to be all of these. I think like as I'm looking to the future, it's going to be AI agent optimization is going to be a massive thing that these companies want to do. And so everyone essentially is going to want to get this tool running on their website and make it more

readable, accessible, and understandable to AI agents to increase the likelihood of it actually being, you know, the agent actually purchasing their tools or using their resources or getting their subscriptions or whatever. So this is what I think is very, very interesting. So

Something that's really interesting here is what they're kind of saying about this. So apparently they have about 20 companies that were in the current Y Combinator winter batch that use browser use. So evidently these brand new startups are already starting to use it. They can see the use on this.

They said, quote, there are companies coming to us saying, what can we do to make it easier for AI agents to navigate our website? There are sites, for example, LinkedIn, that change the way the website works all the time. So agents often fail on sites like those. According to them, they've been looking at getting into the AI agent space for the last few years and browser use just felt like the perfect thing for them to start working on. And their lead investor has been also looking at

investing in this for a number of years and was excited to add this to the portfolio. Their lead investor, Myers, said, we think web agents are the next frontier that really help with the end-to-end automation of human tasks. Web AI agents are this dynamic bridge between static pre-trained models that are mostly text-focused in the ever-changing digital landscape. Okay, so all in all, I think this is a fantastic business, raised $17 million, and I think they're going to do quite well

targeting end users. Right now, it's just like developers that are grabbing it and running it on their computers. That's fantastic. But I think if you look at a majority of all websites, you have a ton of Shopify websites, you have a ton of Squarespace websites, a ton of WordPress websites. And I think that you'll have the big companies like Shopify get plugins or make deals with companies like Browser used to get this built in. And everyone is going to be working on AI agent optimization in the future. I think this is going to be a massive new industry that nobody is talking about, but it is

It's going to be huge. You know, it's going to be the secret to some companies making hundreds of millions and billions of dollars is if they can nail an interesting concept to be looking at. Thanks so much for tuning into the podcast. If you enjoyed it or if you learned anything new, I would love it if you could leave a review wherever you get your podcasts on YouTube.

comment, subscribe. I would love to have you as a subscriber over on YouTube. And if you're interested in joining the AI Hustle School community, there's a link in the description, $19 a month. And there's over 50 videos showing you how to grow and scale your business with AI tools, how to make thousands of dollars and really take your business to the next level with AI. Thanks so much for tuning in and I will catch you next time.