How AthenaHQ and Browser Use Are Building the Next Layer of the Internet | E2132

Today's show: Alex interviews two cutting-edge startups shaping the future of AI: AthenaHQ, pioneering "GEO" (Generative Engine Optimization) to help brands rank in AI-powered search engines like ChatGPT and Gemini, and Browser Use, building infrastructure that lets AI agents take real actions on the web—from form-filling to workflow automation. Both teams dive deep into how fast AI is evolving, how brands and developers can stay competitive, and why adaptability and technical precision are key in the era of intelligent search and autonomous agents. Timestamps: (0:00) Episode Teaser(1:54) Athena HQ's Generative Engine Optimization (GEO) strategy(10:05) Squarespace - Use offer code TWIST to save 10% off your first purchase of a website or domain at https://www. Squarespace.com/TWIST(14:51) Will GEO Market Adoption Forever CHange Marketing?(19:50) Fidelity Private Shares℠ - Visit ⁠https://fidelityprivateshares.com⁠! Mention our podcast and receive 20% off your first-year paid subscription.(22:58) How Being AI Model Agnostic Gives You An Edge(29:54) Superpower - Superpower - Visit superpower.com/twist to get $50 off your membership. This offer is only for the first 100 twist listeners who sign up. (35:13) Browser Use and Making Smarter AI Agents(45:03) Will AI Agents Do Your Grocery Shopping For You? Subscribe to the TWiST500 newsletter: https://ticker.thisweekinstartups.com Check out the TWIST500: https://www.twist500.com Subscribe to This Week in Startups on Apple: https://rb.gy/v19fcp Links from episode: AthenHQ: https://www.athenahq.ai/ Browser Use: https://browser-use.com/ Follow Andrew: X: https://x.com/andrewyan200 Follow Magnus:X: https://x.com/mamagnus00 Follow Lon: X: https://x.com/lons Follow Alex: X: https://x.com/alex LinkedIn: ⁠https://www.linkedin.com/in/alexwilhelm Follow Jason: X: https://twitter.com/Jason LinkedIn: https://www.linkedin.com/in/jasoncalacanis Thank you to our partners: (10:05) Squarespace - Use offer code TWIST to save 10% off your first purchase of a website or domain at https://www. Squarespace.com/TWIST (19:50) Fidelity Private Shares℠ - Visit ⁠https://fidelityprivateshares.com⁠! Mention our podcast and receive 20% off your first-year paid subscription. (29:54) Superpower - The best founders know: better health = better business. Visit http://superpower.com/twist to skip the waitlist. Great TWIST interviews: Will Guidara, Eoghan McCabe, Steve Huffman, Brian Chesky, Bob Moesta, Aaron Levie, Sophia Amoruso, Reid Hoffman, Frank Slootman, Billy McFarland Check out Jason's suite of newsletters: https://substack.com/@calacanis Follow TWiST: Twitter: https://twitter.com/TWiStartups YouTube: https://www.youtube.com/thisweekin Instagram: https://www.instagram.com/thisweekinstartups TikTok: https://www.tiktok.com/@thisweekinstartups Substack: https://twistartups.substack.com Subscribe to the Founder University Podcast: https://www.youtube.com/@founderuniversity1916

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Speaker 00

  One thing I was curious about is just the pace of AI development and how that impacts your ability to help people. Is it worth investing in trying to rank well across all of the AI models and services out there? Or do you guys focus a bit more on like, okay, look, here are the three places people are actually doing AI search?

• 0:15Speaker 01

  We are model agnostic. We focus on the underlying architecture of AI search. What that means is that individual models, whether it's ChatGPT, Gemini, or DeepSeek, et cetera.

• 0:28Speaker 00

  Sure.

• 0:29Speaker 01

  They rise and fall in popularity, right? And it's very much like a footrace. At one point, Gemini 2.5 is ahead in the lead, but then other models will race ahead. So we're not betting on any particular horse here. We're helping our customers be resilient.

• 0:46Speaker 02

  This week in startups is brought to you by Squarespace. Turn your idea into a new website. Go to squarespace.com/twist for a free trial. When you're ready to launch, use offer code TWIST to save 10% off your first purchase of a website or domain. Fidelity Private Shares. If you want the all-in-one equity management platform, Fidelity Private Shares has you covered. Visit fidelityprivateshares.com and mention this podcast for 20% off your first year subscription. And? Superpower. The best founders know better health equals better business. Visit superpower.com/twist to join and skip the waitlist.

• 1:19Speaker 00

  Hey, everybody. Welcome back to Twist. This is Alex. Today, we are talking to not one, but two companies from the Twist 500. What is that? Well, it's our search for the top 1% of startups out there. And we currently have about 350 companies on the list with another 15 coming later today. But if there's a name that we're missing that you absolutely think we need to consider, well, shoot me an email, alexwlancho.com, or send us a tweet at alex@jason@lawns, and we'd love to hear from you. Today, we are talking to Athena HQ and Browser Use. We're going to start with Athena HQ. It's a startup that wants to help other companies understand how well, or frankly, how not well, they are doing at showing up inside of AI search queries.

• 2:03Speaker 00

  Now, we're all super familiar with SEO, search engine optimization. And I think it's time to get accustomed to what Athena HQ calls GEO, or Generative Engine Optimization. Listen, if you care about search, you care about web traffic, or just how brands are adapting to the AI era, this one's for you. Let's go. Hey everybody, welcome back to Twist. This is Alex and every time we have a Twist 500 interview, I tell you that I'm excited about it, that we have a great founder and a cool company and an interesting product. And I always mean it, but today I mean it extra because I'm incredibly excited about this company. Its name is Athena HQ. It was part of the recent Y Combinator batch and it's working on a problem that I am incredibly interested in because I have a long career in media. That's where I lost all of my hair.

• 2:47Speaker 00

  And so I am very accustomed to the idea of search engine optimization or SEO, as you probably know it. Now Athena HQ wants to bring that concept into the era of generative AI and they've rebranded SEO to be GEO. I want to learn more about that and the company itself. So please join me in welcoming Andrew Yan to the show. Andrew, hey, how you doing?

• 3:09Speaker 01

  Doing well, good to chat, Alex.

• 3:10Speaker 00

  Thank you for being here. First, starting off, you were a Google Search PM and I want to just ask you as a former fan of Google Search, what happened to it? Because now it's just ads and AI slop and I'm a little bit disappointed and I was going to blame you, but I figured I'd ask first. What, what happened over there?

• 3:29Speaker 01

  You can blame, put all the blame you want on me. It was all me.

• 3:32Speaker 00

  I'm glad I finally caught you.

• 3:34Speaker 01

  It's pretty clear that Google has gone through many different stages of its life, kind of like a newborn child to a toddler phase and now in his teenage years. And now we're in this adolescent stage is how I like to think about it, of figuring out the identity of Google, so what Google Search means in the age of AI search. When people are expecting a ChatGPT style answer, which is generated for them, which goes against the original newborn phase of Google of the temblu links style of search. So this is a existential question for Google is, does Google make the shift from temblu links to the AI style search without cannibalizing in search ads revenue and?

• 4:15Speaker 01

  That's the big elephant in the room is how do you do that? And we're seeing steps already with AI overviews being scaled up across more and more query types as well as AI mode, the preview mode being launched out about a month ago, early March. So I think it's safe to say that we're heading in the direction where Google is trying to avoid that innovator's dilemma.

• 4:39Speaker 00

  Yes, exactly.

• 4:40Speaker 01

  Reinventing itself for this new age, using different types of, different modalities of search beyond just text search, but also image, video, and way beyond.

• 4:50Speaker 00

  I didn't bring that up just to be annoying and rude, once again, to the Google Search team, as I usually am, but instead to kind of underscore the shift that's going on in search, because for the longest time, built into Chrome, whether you went there to Google.com, Google was the absolute kind of first step into most of the internet. And now it really does seem that the world has shifted and people now expect much more of a personalized response directly to their query as opposed to a list of possible answers to it. And also, they're just more accustomed to a non-search interface for asking questions. And so to me, it really does feel that AI search, generative search, whatever you want to call it, is catching on very rapidly. And I think that kind of lays a foundation underneath Athena HQ, because my presumption,

• 5:32Speaker 00

  Andrew, is that the number of queries that are going through generative AI search engines broadly is going up very steadily and therefore creating a lot of demand for your company's ability to help people better rank and show up inside of those AI search queries. Is that fair?

• 5:49Speaker 01

  Yeah, definitely. I believe ChatGPT just crossed the 520 million mark recently.

• 5:54Speaker 00

  Yeah, it's kind of crazy. Gemini has like 350 million actives. I think, I think ChatGPT crossed 800 million.

• 6:00Speaker 01

  I might be wrong on that number. Yeah, it might have gone up even more since I last checked.

• 6:03Speaker 00

  That's the problem with AI. It's moving so quickly, it's a little bit hard to actually keep things under control. So let's talk about the core mechanics of what your company does. As I understand it, SEO, I'm going to tinker with my links, my keywords, my internal site architecture, and generally try to be understandable by search engines and therefore findable, indexable, and shareable. In the realm of GEO or generative engine optimization, what can you do to help people rank better inside of AI search queries? And how difficult is that work?

• 6:35Speaker 01

  That's the magic question. If there's a magic wand, I'll just wipe it and then solve it immediately for all the websites in the world.

• 6:42Speaker 00

  Excellent.

• 6:42Speaker 01

  How we approach it and where it starts is measurement is how do you find out where you're doing well on AI search and where you're not doing well on AI search? And after finding measurement, we help our customers optimize your content for AI. Specifically, specific structures of content or citations of content that used to be working for SEO and strategies that worked for SEO don't necessarily carry into the GEO world.

• 7:12Speaker 00

  Okay, so essentially what we have done before is not going to get us into the new era. So we need new tools, new methods, new tricks even.

• 7:20Speaker 01

  Exactly. And we found that being number one on Google search, we've seen companies oversee the number one ranked website on Google search, even though they might be a smaller player, but they have a GEO strategy. And we're helping some of our smaller customers with this is just because you're number one on Google doesn't mean that you are safe on AI search.

• 7:41Speaker 00

  I wouldn't think there'd be much of a translation at all, but let's just start with measurement. You mentioned, you know, knowing where you're showing up and in what context. The way that I can think about doing that is by running a bajillion queries through OpenAI's API, for example. Is there a less tedious way of figuring out where people rank inside of generative AI search? Or is it kind of a brute force, go out there and do every single query and then see where you land?

• 8:06Speaker 01

  So there are definitely optimizations you can take to this sampling based approach to measurement. But fundamentally, a sampling based approach is necessary because AI search is probabilistic, right? And that's the beauty and the weakness of gen AI.

• 8:23Speaker 00

  Actually, Andrew, for people who don't know, can you explain probabilistic versus deterministic in a search context? Because you'll do a better job than I will.

• 8:30Speaker 01

  Yeah, happy to. So let's say you search up best basketball shoes, right? On Google, it's fairly deterministic. You can see the same responses pretty much every time. On ChatGPT and Gemini or these different, this new architecture, you're not guaranteed to see the same thing. And if you search the same thing 10 times, you might get 10 different answers, sending from 10 different sets of websites that are informing those answers.

• 8:53Speaker 00

  I mean, SEO was tough and kind of a black box. That just sounds torturous. Okay, so you said a sampling based approach. I presume that if you were doing GEO for Alex Incorporated, you would do a bunch of individual queries through different AI models in and around the topic that Alex Inc. works in. And then I guess, how do you know when you have done enough sampling to know with confidence that a brand is showing up X% of the time in Z context and so forth? Like, how many data points do you need?

• 9:22Speaker 01

  It's a good thing that my co-founder and I studied math and statistics in undergrad. That's a great question. It depends on your company size and the number of product lines you have.

• 9:32Speaker 00

  Okay.

• 9:33Speaker 01

  If you think about a company, like a conglomerate company like Nike, which cares about basketball shoes, running shoes, athletic clothing, you have many different product lines. Each one of them is essentially its own cluster that you need to be tracking.

• 9:48Speaker 00

  Okay.

• 9:48Speaker 01

  And that could be to the order of millions of prompts that we need to be tracking on their behalf or analyzing on their behalf. Versus if you have a very poignant solution, you might get away with only a couple of thousand. So it really depends on the scope of company.

• 10:05Jason Calacanis?

  All right, founders, let's talk about your website, huh? It might be a little bit of a sensitive subject. You might be a little embarrassed about it. Let's fix that right now. If you're launching something new, give your brand the refresh it deserves. You're working so hard on your company and your website looks terrible. Do what I do, use Squarespace. It's the all in one platform to make a stunning professional website. And it's ridiculously easy to use. It's going to work for everything. If you're selling products, it's got all that e-commerce built in. If you're selling services, it works perfectly. And they have a new AI product called Blueprint. Want more control? Well, of course, you can choose one of the award winning templates and use the intuitive drag and drop tools to make it your own. Easy peasy.

• 10:47Jason Calacanis?

  Squarespace equals the most functional and beautiful website on planet Earth.

• 10:53Jason Calacanis?

  Go ahead, Squarespace.com/twist to get a free trial. And when you're ready to launch, go to Squarespace.com/twist to get 10% off your first website or domain purchase. That's Squarespace.com/twist.

• 11:05Speaker 00

  And for folks who are curious, kind of what we're talking about, I was reading the case study you guys have about Figma and Canva on your site. So what I have here on screen for folks who are on the audio is essentially, it's quasi pie chart showing AI search share of voice inside of the Canva Figma Adobe Express domain. And here you can kind of see how different brands might stack up against one another. So that's kind of what we're talking about. Once you have sorted out, let's not use Nike because it's too big, strictly a company that makes shoes for the game of pickleball. Pickleball shoe company. Once you've done the proper amount of searching there, you have a good sample, you feel confident about knowing about it. How much can you influence AI search? Because in Google, there's a whole industry around SEO, helping people rank up. Some people are charlatans, some people are actual experts.

• 11:49Speaker 00

  A little hard to tell from the outside, but I'm just curious. If my brand is struggling very hard to do well in, I don't know, Gemini, OpenAI, and you know, anthropic models, how quickly can you guys actually influence that? And how much of it is just time and luck?

• 12:05Speaker 01

  So we can definitely influence it. And we're seeing on the faster end, our suggested content being picked up within a week on the fastest end, which is incredibly fast. And this is thanks to the new indexing method that perhaps the startup pioneer that started off with. But now we saw chat to be incorporating search into GA in October last year. And more and more engines clawed incorporating search as well recently. Where it starts for this hypothetical pickleball shoe company is first figuring out if you are tracking the right prompts and figuring out, are you targeting the right prompts with the right content?

• 12:48Speaker 01

  If you're not, then we help you fix your content itself, or we help you address content towards the right prompts. So that's number one is making sure that you have the right content that is addressing the questions that users are asking. And we have our own proprietary data sources to.

• 13:05Speaker 01

  Estimate the volume of these prompts and of these topics.

• 13:09Speaker 00

  What can you tell me about these proprietary data sources? And are these things that Athena HQ is sourcing itself? Or is this something that you're purchasing from a third party that might have a useful data set for you guys to use?

• 13:20Speaker 01

  Yeah, there's not much I can say about that beyond that, except that we are able to track 1% of all traffic from ChatGPT and Perplexity and get that volume information for our customers.

• 13:34Speaker 00

  1% is a lot. That's a lot of queries because these, I mean, you know, ChatGPT alone does roughly one bajillion queries every day. So 1% is a lot. Is that, that almost feels like too much information. Tell me, tell me why I'm wrong and that's actually a useful number.

• 13:51Speaker 01

  Useful is all in the eyes of beholder. It's about where do you fit this information into your workflow? And we could have the most powerful data set. We could have OpenAI's data if we, if we had it and it still might not be useful if we don't integrate it in the right way to our customers workflows. So how we do it is we build a product around marketers workflows. And we make sure that marketers are able to get the most value out of Athena beyond just an analytics platform, but actually helping them on each step of measuring, taking action, and then being able to measure the impact of those actions.

• 14:28Speaker 00

  Okay, this all makes pretty good sense to me. And the reason why I said at the top of this that I'm actually very excited is because my personal view is that AI search is going to become the de facto within five years. Therefore, it's going to do bajillions of queries and therefore everyone's going to want to have a tool like they had in the SEO era to help them move forward. So to me, you guys are going right down the direction towards where I see the future going. My question is, does the market agree? Because I'm not sure if I'm thinking ahead of where the average marketer is or is GEO already something that people are working into their, let's say, 2025 operational plans?

• 15:05Speaker 01

  Yes, depends on the company. There are definitely early adopters who are. Way ahead of the curve here and already anticipating the repercussions of not adapting to G E O. And to do not adapting to AI search and using G E O. The kind of beacon on the hill, so to speak, here is HubSpot, right? Which lost eighty percent of all of search traffic year on year, organic search traffic year on year.

• 15:31Speaker 00

  What? I missed that HubSpot lost eighty percent of its search traffic?

• 15:35Speaker 01

  Oh, yep. They made a post about it.

• 15:38Speaker 00

  Oh, that HubSpot was famous for being good at that. That's pretty embarrassing.

• 15:42Speaker 01

  So that's, and that tells us that this is shift going on. The winners of the last war, the Chegs, the Yalps, the Nerd Wallets will not necessarily be the winners of this next era.

• 15:55Speaker 00

  Yeah.

• 15:55Speaker 01

  We see yourself with Chegg, right? Which lost, I believe, fourteen billion dollars in market cap. Because, because of AI search, well, mostly because of AI search.

• 16:04Speaker 00

  For folks who don't know, Chegg was originally a textbook rental service that then pivoted into essentially a homework helping service, if you will. And then ChatGPT came along and did that better for cheaper. And Chegg's share price last time I checked, Andrew, was down like ninety-five percent or something like that. It was, it was pretty brutal. I think that shows the risks of not having a good G E O strategy. But my, my, my guess is that AI search query volume today couldn't make up that eighty percent that HubSpot lost or that Chegg lost or a similar company. But probably it's going to become a bigger share and might be able to help people avoid falling into an S E O death drop.

• 16:45Speaker 01

  So the way how we look at AI search and how to think about AI search from a marketer's perspective is the buyer's journey itself is changing. Before you measure buyer's journey performance in terms of how many clicks are we getting from Google? How many, how much traffic are we getting? But that's no longer the right metric to be benchmarking against because customers are doing more research upstream of your website. They're learning about your website on ChatGPT, about your company on ChatGPT, on Perplexity.

• 17:15Speaker 00

  Yeah.

• 17:16Speaker 01

  I'll share an anecdote here, Alex.

• 17:18Speaker 00

  Hit me.

• 17:18Speaker 01

  We actually chose when we were deciding which corporate card to go with. We researched on ChatGPT.

• 17:24Speaker 00

  Right.

• 17:25Speaker 01

  And that informed a good portion of our decision. ChatGPT did not make the decision at the end of the day, but that was a big influence on which corporate card we chose. And I can imagine many similar situations around the world where there are these decisions, whether it's evaluating products or evaluating life decisions that people are making informed by AI.

• 17:47Speaker 00

  I just realized why that's the case. Well, first of all, what did you choose and why was it RAM?

• 17:52Speaker 01

  Unfortunately, we went with Vex.

• 17:53Speaker 00

  No way. Oh, Enrique is going to be so happy. The difference is, and I was trying to put my finger on this, but I think you just helped me because if I go to Google, I type in best corporate card or best corporate spin management suite or whatever, I'm not expecting Google to actually tell me which one is best. I'm expecting you to tell me which one ranks the highest, which is probably a comp, but not exactly the same thing. But I would actually type into ChatGPT or similar, hey, ChatGPT, comma, what is the best corporate card? And so it's definitely talking to me more directly in a way that's searched previously wasn't. So it actually feels more probabilistic, but more deterministic in how it tells me its probabilistic answer. Does that make sense?

• 18:36Speaker 01

  Yeah, it definitely does. And I think what you're getting at, Alex, is actually Gemini and ChatGPT as buying assistants. I have them go out, do research for me, doing deep research. And then I ultimately make pull the trigger, but they're the ones collecting all the data for me.

• 18:50Speaker 00

  Yeah, no, it's, I am not an AI first guy. I'm an AI adjacent guy. But increasingly, I find myself more and more and more and more each and every quarter, it takes up more and more of my workflow. Okay, let's talk about the customers you do have because one thing I was impressed by is how expensive the service is. A lot of things that we see in the world of AI today cost 15, 20, 40 bucks a month. According to the website, you guys cost, I think it's a minimum of 400 a month, monthly, a little bit less if you pay annually. But that's much larger than I often see. So one, good job on not underpricing your product. But my question is, how is the market reacting to a price at that level?

• 19:30Speaker 01

  The marketers who understand the value have no question about it. They understand that if you frame the question in this way of, what does it mean if your pickleball, Alex, your pickleball shoes company makes it to number one on ChatGPT? Right? And what's the alternative to being number one on ChatGPT?

• 19:50Jason Calacanis?

  This advertisement is paid by Fidelity Private Shares. All right, founders, we all know cap tables, due diligence, and of course, managing investors is a huge headache, but there's a very simple solution for you. Today, we're talking with Kristin Kraft, an old friend of mine, and she works at Fidelity Private Shares, a new group over at Fidelity. You've heard of Fidelity before, and they have a mission to help startups simplify equity management. They're going to save you money. They're going to give you better service. Welcome to the program, Kristin.

• 20:20Speaker 04

  Thank you so much, Jason. It's great to see you again.

• 20:22Jason Calacanis?

  Yeah, great to see you as well. Maybe just from a product perspective, what are you trying to accomplish with the product?

• 20:28Speaker 04

  So, Jason, we are super excited about our cap table management and data room platform. We want to make it super simple for founders and startup operators to manage all sort of ownership and equity in the company and essentially prepare to raise. We want to make sure that everybody goes into these fundraising conversations well-prepared, they're ready to share their cap table, and that they're ready to go through due diligence as they're trying to close their round. So from a product perspective, that is where we're laser focused, and that product is really well-built, really strong attention to detail in the way that Fidelity is known and beloved for.

• 21:06Jason Calacanis?

  So if you want an all-in-one equity management platform, Fidelity Private Shares, they've got you covered. Visit fidelityprivateshares.com. That's one word, no spaces, no dashes. fidelityprivateshares.com. And hey, I mentioned This Week in Startups, they'll give you 20% off your first year subscription. Once again, fidelityprivateshares.com. And tell them that you heard about it here on This Week in Startups.

• 21:29Speaker 01

  The ROI is pretty clear in terms of

• 21:32Speaker 00

  Huge.

• 21:32Speaker 01

  If you're, yeah, if you're CPC on Google Ads, is let's take Vanta for Secure Frame, right? For Vanta for Secure Frame, I believe that term itself, on the high end, if I remember correctly, that's to the order of like hundreds some dollars per click.

• 21:46Speaker 00

  For Vanta, the compliance management

• 21:49Speaker 01

  Software.

• 21:50Speaker 00

  Software service. Yeah, yeah, I think they're actually, they're a twist sponsor. Shout out Vanta.

• 21:53Speaker 01

  There we go. So that gives you a sense of the value of Athena that Athena can provide. First thing, you know, we always press on value. We are not the budget provider here. We are the The option that gives you the best value for what you're paying for.

• 22:09Speaker 00

  Okay, I'm gonna rephrase my entire question. Now that I know that you can spend over a hundred bucks on a single click for a Vanta Google search, why haven't you added two zeros to your price tag because you're undercharging dramatically for this? I mean, honestly, I had it backwards. Maybe 400 bucks a month for the light packages, a fraction of the value you're offering, which is good because it'll help you with retention. But like, I mean, if people are this dependent on you to remain relevant in the AI search era that you and I both think is coming, why not charge more?

• 22:38Speaker 01

  We could. We're focusing on the value part right now and delivering the most value to our customers. It's all about how do we help our customers succeed? It's not really about how can we charge the most amount of money and squeeze the most amount of money out of our customers. That's not the name of the game for us. We're playing the long-term game.

• 22:58Speaker 00

  I thought this was capitalism, you know, just saying, kidding. One thing I was curious about is just the pace of AI development and how that impacts your ability to help people because while I'm really glad that we got O3 and O4 Mini and O4 Mini Hi and we got Gemini 2.5 Pro, 2.5 Flash, 2.5 whatever, et cetera. There's a lot of different models and given that they're each a little bit different, I presume that customers of Athena HQ want to make sure they're doing well in all of them. Is it worth investing in trying to rank well across all of the AI models and services out there? Or do you guys focus a bit more on like, okay, look, here are the three places people are actually doing AI search. We'll help you there and nothing else is as commercially pertinent, if you get me.

• 23:43Speaker 01

  Yeah, that's a great question. So we are model agnostic. We focus on the underlying architecture of AI search. What that means is that individual models, whether it's ChatGPT, Gemini, or DeepSeek, et cetera.

• 23:59Speaker 00

  Sure.

• 24:00Speaker 01

  They rise and fall in popularity, right? And it's very much like a footrace. At one point, Gemini 2.5 is ahead in the lead, but then other models will race ahead. So we're not betting on any people or horse here. We're helping our customers be resilient, regardless of if there's a new model that comes out, for example, Grok gaining a lot of popularity recently. Helping our customers adapt to that.

• 24:24Speaker 00

  Okay, so essentially, you guys have a technology that works regardless of the model in question. It's not like you have one engine for ChatGPT's models and one for Google's models. It's more one product that they can interface with. Okay, that makes a lot more sense, and I appreciate that. One other thing about the SEO world that I think we're all aware of is that there was sometimes tension between the SEO community and Google. And I think you've probably seen this throughout time. And so I'm kind of curious, how is your relationship with the OpenAI's of the world? Do they care that you're trying to help people do better? Are they in favor of it? Are they opposed? I'm curious what that relationship is like.

• 25:03Speaker 01

  Yeah, I mean, we're not, we're not trying to be antagonistic by any means. The goal for us is to help companies tell their story on AI.

• 25:13Speaker 01

  And if you're a company and you have stale content, you have old pressing pages, you have an old name. That's bad quality content that if you're OpenAI and if you're Google, you don't want there anyways, because it's just not relevant to the, to the end user. So we help companies with these as well as we're coming out with a feature soon to detect hallucinations.

• 25:39Speaker 00

  Oh, interesting. So defense instead of offense.

• 25:42Speaker 01

  Exactly, helping with both.

• 25:44Speaker 00

  Okay, I dig that. But what you're describing is the equivalent of white hat SEO. Making better content that is more findable and better for both Google's results and end users. Everyone's in favor of that. There are other SEO methods that have been less popular because they're a little bit more nefarious or tricky. Is there any risk of that type of activity working in a GEO era? Or is it harder to trick AI models because they kind of do their own thinking versus kind of Google's great algorithm in the sky?

• 26:19Speaker 01

  This is a great question. We can get very deep into this. There are research papers written about, about these explorers. And I come from a cybersecurity background as well. And I remember the days of SQL injection and this analogy of SQL injection to prompt injection, right? And we don't do any of this black hat, you know, SEO or GEO strategies. The reason is that it's just not. I don't consider ethical. If you read the papers, like there are ways, like if you have an unguarded model with no guard rails, like yes, you can influence it through prompt injection. The joke is like adding white text to the bottom of your website, this sort of thing. But that's not how we operate as a company, so we don't engage in that.

• 27:04Speaker 00

  So to be clear, the two guiding principles of Athena HQ's I can find is make sure you're delivering far more value than you're charging for and be ethical. All right, that's got to be the best takeaway I've ever had from a chat with somebody. I really dig that. I'm joking. Y Combinator, you guys are part of Winter 25. I think you raised the full half million from YC. Is that correct?

• 27:25Speaker 01

  We've raised more than that.

• 27:26Speaker 00

  Oh, okay. What can you tell me?

• 27:28Speaker 01

  We've raised over 2 million from investors and angels from companies such as OpenAI, Anthropic, DeepMind, also at the consumer companies that you expect, like Instacart, Shopify.

• 27:44Speaker 00

  Brands that are going to want to do well on AI search.

• 27:48Speaker 01

  Exactly.

• 27:49Speaker 00

  Now, on the Anthropic and OpenAI side, I don't know if you can tell me this, but are those investments made by individuals at those companies? Or were they done through OpenAI's venture arm, for example?

• 28:02Speaker 01

  These were made by individuals. I want to be very clear about that.

• 28:05Speaker 00

  It's very interesting that people at those companies see the potential in what you're doing. I know you guys were a part of the last demo day. I think that's when I first heard of you. How has growth been as a business in the last couple of months?

• 28:20Speaker 01

  Yeah, growth has been explosive. Our customers are seeing rises in traffic from ChatGPT, Perplexity, Gemini, AI overviews. And that's just a tip of the iceberg in terms of the users who are impacted by AI search. So we've been, you know, working night and day to meet up with the need demand.

• 28:44Speaker 01

  It's a space that's also rapidly evolving. So we're also staying up to date ourselves on the latest developments. For example, looking at how different integrations with these model companies are possible, specifically in the e-commerce sense of shopping or different verticals. A lot of surface area here, a lot of exciting things because this is the future of how product discovery is going to be in the next foreseeable future.

• 29:10Speaker 00

  What would you do if Anthropic just showed up and was like, we'll give you $100 million in stock, just come work for us?

• 29:18Speaker 01

  That's something I can't make that unilateral decision without our team.

• 29:23Speaker 00

  I'm curious about what's going to happen to what I consider to be very promising, early-stage AI-facing or AI-predicated startups, because there's a lot of just very wealthy AI model companies. And right now, we've seen OpenAI try to buy Cursor, and now they're looking at Windsurf. And so I'm kind of curious if there's going to be a suction effect in the market of the model companies, the foundation model firms, just buying people that are doing awesome stuff. And frankly, because I'm a fan of your company, I would put you in that bucket, if that makes sense.

• 29:54Jason Calacanis?

  Founders are all about efficiency and performance, and that's why they automate and optimize every system within their startup. But when's the last time you optimized yourself? Well, startups move fast. Sleep and workouts get skipped, we know that. And health becomes an afterthought until it's a problem, but now there is Superpower, the ultimate founder health membership, of which I am a member. Full body testing across 100-plus biomarkers, and you get guidance for organ health, hormones, inflammation, and more, giving you actionable data and clear insights. You track your KPIs at your company, why not track your health the same way? Well, with Superpower, you can. I'm using it. Lon Harris is using it, a couple of other people on the team, and they have 150,000 person waitlist because it's so cool.

• 30:39Jason Calacanis?

  But because you're a Twist listener, and we're going to get you ahead of that line, J Cal is going to take you through the VIP entrance at superpower.com/twist, and you're going to claim your spot in the VIP line. It's not even a line, it's just the back door, folks. And you're going to unlock peak health today because better health equals a better founder, which equals a better business. Again, superpower.com/twist to go through the VIP entrance with your boy J Cal.

• 31:03Speaker 01

  It all depends on how fast to get to AGI.

• 31:06Speaker 00

  Well, that's either going to be six months to 60 years depending on who I'm reading. So you've raised more capital, growth is explosive. How big is the team these days? I know you have a couple of folks on your YC page, but I wasn't able to figure out more.

• 31:18Speaker 01

  We're a team of four. Our team is lean and mean. If you think about, I model our team after the teams of tomorrow. If you think about the teams that are winning right now, they're the teams like Cursor, like Loveable, like these

• 31:30Speaker 00

  Yeah.

• 31:31Speaker 01

  small, tight knit teams because context is everything. Right? And context is what's most expensive in this startup world that we live in. So our team members have full context on every part of the business.

• 31:45Speaker 00

  I think I know what you mean by context. And so if you're working inside of like, I don't know, let's say you're working at LinkedIn, which is superior to Microsoft. You're going to be put onto a team, say a product team, then you're going to be filtered down to like the sub-team. And then you're going to do machine learning on how people click on a button. At some point, you're so niched down into the pixels that you can't at all see the forest for the trees. But on a smaller team where there's a lot of shared responsibility, there's a natural co-mingling of operational information. So that way everyone actually knows kind of what's going on with everything, i.e. greater context. Fair?

• 32:18Speaker 01

  It's exactly that, seeing the forest for the trees while also being able to scope down into the tree level of getting clarity.

• 32:24Speaker 00

  So here's my question. Four people? Hell yes, I totally get it. Eight? Sure. 12? Getting hard. 15? Ew. So does this desire for greater in-truck team context imply that there's a upper level on how many people can work at a thin HQ?

• 32:42Speaker 01

  Definitely.

• 32:43Speaker 00

  That's interesting.

• 32:44Speaker 01

  I think we can scale a company with under 20 employees.

• 32:47Speaker 00

  Including like customer success?

• 32:49Speaker 01

  In the near term, in the near term. So not, not in the long term.

• 32:53Speaker 00

  How long is the near term?

• 32:55Speaker 01

  The next year. If you think about the amount of leverage that each individual employee has, where using, if you're using the right tools in the right way and you're training them together, there is this I can very much see the road down the line of what Sam Altman describes that one person billion dollar company. I accept one person billion dollar, maybe a bit further along, but.

• 33:17Speaker 00

  Yeah, but I mean, directionally, he's drawing an X on the map and telling us we're walking in that direction, which I, which I kind of agree with. Would you consider it to be like a, I'm going to draw a hypothetical scenario here, but like, let's say five years from now, you guys are preparing to go public and you have 50 people. Would that be a win or would that be an under investment in human talent?

• 33:38Speaker 01

  I wouldn't say that's a under investment in human talent if the talent density is very high. That's a key metric is talent density of.

• 33:47Speaker 00

  Yeah.

• 33:49Speaker 01

  Of your team.

• 33:49Speaker 00

  For folks who want to take a look, what is the URL and what is, well, normally I ask, what's one role you're having a hard time hiring for? I'm not quite sure that's relevant, but answer it if you like, Andrew. But first, what's the URL?

• 34:00Speaker 01

  The URL is athenahq.ai. Yeah, the role that we're open to is we're hiring for selectively across roles in engineering and technical growth. So everyone on our team codes and that's part of our culture. We are highly technical. We come from highly technical backgrounds. Our investors are highly technical. Our team members are highly, highly technical. We, we are highly, this is our culture. You know, this is how we serve our customers. We are that technical advisor in some sorts to help them adapt to an increasingly technical world of AI search.

• 34:36Speaker 00

  I really dig it. And when you hit 10 million ARR, come back on the show. I want to hear all about it and how fast you're going at that point. But in the meantime, Andrew, thank you so much and good luck.

• 34:44Speaker 01

  All right. Thank you, Alex.

• 34:45Speaker 00

  Fun. Yeah. I for one, am incredibly excited to see what comes next in search. And I think it's going to be a little bit more than just the AI answers we see today. But if the SEO era taught us anything, it's that no matter what form search does eventually take, companies and brands are going to want to dominate it. So frankly, I have very high hopes for what Athena HQ and its competitors like Profound, another twist 500 company have coming in the next few quarters. Now, next up, we're talking to browser use. We've all heard the hype about AI agents, you know, AI tools that can do a lot more than just answer questions or write you a limerick. No, these can go out and actually do tasks for you.

• 35:25Speaker 00

  I mean, hell, today, we're actually talking about agents working with other agents, perhaps even led by a chief agent, and things are only accelerating. But how will these agents actually interact with an internet and all the internet applications that are designed for us humans? Well, Browser Use has an idea of just how to bridge that gap. Let's learn more. Today, we have yet another Twist 500 interview, one that I've been excited about from the very first moment that I discovered this company back during a YC demo day a couple of months back. Essentially, here's my thought. Right now, when I use AI, I tend to go to an interface, I type some stuff in, I click enter, stuff comes back. Very easy, very simple. But I'm a human. What if I was, for example, a different AI system, perhaps an agent? Well, then going out onto the web might be a little bit harder.

• 36:10Speaker 00

  Getting the information that I want could be difficult. What if there was a tool that can make the whole web a bit more approachable to your local friendly AI model? Well, I think that's what Browser Use is doing, so please welcome to the show, it's Magnus Mueller. Magnus, how are you?

• 36:24Speaker 05

  Hey, great to meet you. Thanks for having me.

• 36:27Speaker 00

  Everyone should know that Magnus is calling in late at night from Switzerland, so he made this happen and we are thankful for that. Magnus, first of all, when was the company founded and how many people currently work at Browser Use?

• 36:39Speaker 05

  We founded this whole thing five months ago, in Tenerife this year. When we, you know, we started with a side project during our master here in Switzerland, me and Craig, in a builder's house here in Switzerland, where it's forbidden to study, it's from the university, but it's forbidden to study, only allowed to work on his side projects. And we started to work on this random side project and we got into YC, spent there the last three months. And now, yeah, five months later, we are, it's me, Craig, and we hired two full-time engineers, built all together.

• 37:11Speaker 00

  And if you want to know why we're talking to a company that's so brand new, it's because in five months, they've already accreted, as of current count, 61,005 GitHub stars, which is a simply insane run. But Magnus, here's the problem. I'm a little bit awash in technology to help AI do stuff. There's Anthropic's Model Context Protocol, helping AI kind of bring data in. There's Google's A2A or Agent to Agent Framework, which lets agents talk to one another. Microsoft has a new thing called NL Web. So first of all, for folks out there who don't know, what does Browser Use do and why is it critically important?

• 37:45Speaker 05

  Yes, so we enable your AI to control your browser to navigate there. Imagine it like you can tell your browser your goal and your computer does your clicks for you. You don't have to click anymore yourself. Yeah, like you described in the intro, all those websites like have those search bars, which are highly optimized for humans, those visual contacts. But agents don't directly understand that. And to get the full power of agents, to have many agents work for you, actually doing things for you. We convert websites into something which agents can understand so that they can take action and do things for you on the browser.

• 38:23Speaker 00

  So it's a translation technology that takes, is it any website that's out there in the world today? And using kind of, I presume, a shared toolset, it can convert that into something that an AI can talk to. But it's broad, so everything works with browser use.

• 38:38Speaker 05

  Yes, yes, every website works, can convert everything. Some we can convert better than others, but all work by default. And you can imagine it as, you know, many websites, they have APIs with which you can automate things. But then there are also many legacy systems or things which don't have an API for your specific things. And we enable all those websites without API that your agent can just go there and do things for you. Like filling out a form or getting back your API key. You know, all those AI tools like GPT or perplexity, they're very, very good in extracting information from the web. But they all stop where information gets interesting. They all stop at login walls. All your interesting information is behind login walls.

• 39:23Speaker 05

  And if you have agents who can take action, they can just log in into a bank account and extract your current sales numbers or they're going into your company CRM to create new contacts.

• 39:33Speaker 00

  Or they could log into my bank account and take out all of my currency, you know, just, I'm kidding. So I was going through the GitHub page and I was reading some of the code and I'm not gonna lie, I got a little bit lost. So I'm gonna ask you a very simple question. Is browser use the goal to help agents, AI agents, do things on the web? Or is it almost an agent in and of itself that can also take written commands and then do stuff for you?

• 39:58Speaker 05

  We. Build an agent with memory system. Okay, so it can take hundreds of steps, go down and can do things for you on the web. Prompt this, you type in your prompt, say, okay, go, go to my CRM and create a new content.

• 40:14Speaker 00

  Got it.

• 40:15Speaker 05

  Okay, but all our things are based on actions. And then click something, do a Google search. And you can simply extend this, so you can write your own action, like calling an API. Imagine it like before MCP was out, it's, it's, it's basically like you can include more and more MCP tools, which can do just things and the agent can choose which tool call to do. So then decide, okay, what's the best action?

• 40:40Speaker 00

  Okay, so essentially it's a framework that lets me control the web. But I could bring, for example, some of my own homegrown technology and still use browser use on top of it as almost like a hook onto the web as it exists today.

• 40:54Speaker 05

  Yes, but mainly browser use is used to actually take action in the web. Do things where you don't have APIs for and automate that.

• 41:03Speaker 00

  So really it kind of takes the entire human internet and makes it machine understandable and interaction friendly. Is that fair?

• 41:12Speaker 05

  Yes, and the main component is really to take action, to actually click buttons to fill out a form, to select from a dropdown.

• 41:19Speaker 00

  So people have been talking about agents like nonstop for the last, I want to say 12 months. But before technology like browser use existed, were those agents just incredibly limited in what they could go out and do? Or were people kind of building a similar product to browser use internally? And then using that to go out and talk to the web with their AI agents?

• 41:41Speaker 05

  That of course depends on the definition of an agent. If you just define an agent as an LLM with tool calls, okay, then all those chatbots which can output JSON structure are usually agents. But I was always wondering myself if, when people said, oh, we built those agents. And I thought, this is just a chatbot, right? If you can interact in it, maybe can query the current web weather API for you. So I always imagine agent like some, something which actually can take action for you, just like a player who can take action.

• 42:11Speaker 00

  Okay, that, that makes good sense to me. Now, I want to talk about the ramp because you guys, actually, I have a chart here showing the browser use GitHub star growth. And it's one of the craziest things I've ever seen. This is from basically right before the new year through to May, and you guys just grew very, very, very quickly. So how did you get the word out about browser use and the kind of first iteration of your technology? And were you surprised by just how quickly the market was like, yes, we want that?

• 42:43Speaker 05

  Yes, we were very surprised. We built the first prototype in five days, pushed it to Hacker News, and it took off. And I mean, I never built big open source projects before. It got a couple of hundred stars, people created issues. And the, the main thing is we made it extremely simple for people to run it. We see the setup is run two commands and, and it runs and you can actually prompt it and it can control your computer.

• 43:06Speaker 00

  And that's, that's where I got confused between, does it help other people's agents do stuff or as an agent itself? I think that's where I got a little bit mixed up is the fact that you can prompt it.

• 43:16Speaker 05

  Yeah, exactly. It's prompt based. You can, you can do both. And the cool thing is we could create very cool demos with it. Like it's, it's a really, really cool tool to create demos to, to go and see, wow, this AI actually does something on the web. And with this, we could also create many controversial demos, like applying to jobs online. And by just creating very, very small, not very beautiful demos and sharing them on X, that started our growth engine a couple of weeks later, where then millions of people reposted those examples, created their own, shared them. And I think the key factor for us was there were many things happening in the world, like Operator came out, DeepSeek was, came out, and we jumped on those hype trends.

• 43:56Speaker 05

  We said, okay, this is the free Operator. And then DeepSeek came out and we said, okay, you can use now DeepSeek together with Browser Use for, against Operator. And then MCP came out and we always use those external events in the world, which gave us many, many more users.

• 44:12Speaker 00

  And also, I think when Manus, the Chinese agentic project came out and, and really had one of the biggest weeks of press I've ever seen, they were using Browser Use's technology inside of their agent, right?

• 44:24Speaker 05

  Yes, they were using part of our open source code to convert their website into something which which their agents can understand. Of course, they built many things around.

• 44:34Speaker 00

  Oh, of course.

• 44:35Speaker 05

  But yeah, they used the core part of our library and developed it further and that it was really friendly from them to give us those credits and definitely boosted us a lot.

• 44:46Speaker 00

  All right. So I'm gonna just show a quick clip that you guys have up on the GitHub page. And I was hoping you could just kind of maybe tell people what we're looking at as we watch this kind of do a sportscast, if you will, to help people just kind of get a bit of a visual element. So this is the AI did my groceries example. So Magnus, just tell us what we're seeing here.

• 45:03Speaker 05

  Yeah, this was a friend of mine and he is right now building a startup where he does grocery shopping for other people. And his problem was he always needed to do shopping on his on his own. So now he created a big prompt for browsers. It tells browsers, okay, go to this website and do the shopping for me with those elements. Now you can see, okay, browsers goes to the store and types in, for example, your oranges, new products, and then can click on the actual products and add them to the chart. And with this, he can automate for his new startup, his shopping process, something he would have done manually before.

• 45:41Speaker 00

  And

• 45:42Speaker 05

  And this is just a thing of one big prompt. You can see on the right, this is like a big prompt of me, 100 lines of code. This is just natural language without studying computer science or anything or learning how to automate things. Just a prompt and then it can get started.

• 45:59Speaker 00

  And one thing I noticed watching this video is that it often draws, well, browsers draws what appear to be like boxes around text. Is that how the system parses discrete bits of information to figure out where different options of like potato types would be listed? Or what am I seeing there with the squares, the colors and the numbers?

• 46:18Speaker 05

  Yeah. So we combine the actual HTML with the image of the current screen. And inside the HTML, like DOM, all the elements where we have all the access. Okay, this is a button. Those are the coordinates. This is the core part of our library. There we extract, okay, what are the the elements where humans can interact with, like buttons, input fields?

• 46:43Speaker 05

  And in there, you have also access to the coordinates. So we can just take those coordinates and highlight them on the screen as bounding boxes. And then we can take a screenshot and can combine both and send it to an LLM.

• 46:57Speaker 00

  Oh, okay, that's really interesting. So you're taking images of the website's elements and then sending that back to an LLM to essentially parse it. And then you can tell which thing to select and then it can tell the cursor equivalent to actually click on the right thing. That is a lot of compute back there. That feels like a relatively compute heavy. Am I wrong?

• 47:19Speaker 05

  It depends how you define compute having. It's definitely much, much more than normal chatbots. And the main reason is that we have multiple steps. For example, let's say to fill out a form, you need 20 steps. And at every single step, one step, we take the current website and we send it to an LLM. And the LLM sends us back, click on button five. Then the LLM sends us back, okay, go to this page. Then the LLM sends us back, okay, go and fill out this input field. And every single step, we send out the current website. It's around 8,000 tokens. Let's say around 8,000 tokens together with the image. And the LLM gives us back the output. So it's around, I don't know, one, two, three cents per step, which you pay in token cost for GPT-4O.

• 48:04Speaker 05

  Now for bigger, for newer models like Lama 4 or DeepSeek, we can drastically reduce those prices.

• 48:10Speaker 00

  Well, Lama 4, I hear, is not the world's most exciting model, but it does share something very important with you guys, which is the open source component to it. And I really feel like after watching Databricks become one of the most valuable private companies in the world, people have really decided that open source projects can really build super strong companies. But I'm just curious, from the early days of the company, your code's on GitHub and such. Why did you choose to go the open source route versus a closed source approach?

• 48:38Speaker 05

  Yeah, in the beginning, when we built this, this first pilot in the first five days, there was just, okay, should we do it closed or open? And we said, oh, let's do it open. And push it to Hacker News. And I think for us, it was definitely core for the, for the Prof engine. Like, we for sure wouldn't have raised such a round and get all the traction with like a closed source tool, just because we get so much feedback from, from the users and can iterate based on that.

• 49:02Speaker 00

  That's, that's what I always thought was the most impactful element of an open source project. Not only can people look at it, terror to park, tinker with it and play with it, but they can just tell you what they needed and want. And people always talk about startups need to be, you know, listening to customers, iterating quickly. Well, what if you had just tons of feedback from smart people who are looking deeply at your product?

• 49:22Speaker 05

  Yeah.

• 49:22Speaker 00

  It just makes a lot of sense to me. But on the business model front, I'm curious about your guys' plans. There's been a couple of different ways people have been monetizing software projects like this. I presume you're going to offer something along the lines of a hosted version of browser use?

• 49:37Speaker 05

  So with open source models, you normally have like hosted solutions, you have support which you can sell, and you have enterprise features on top which only enterprise care about. And right now, we mainly look at the, like, in the end, it will be a combination of all three. But we have a cloud solution where we host the browsers and the L L Ms. So you only send us via an API, the task, we run it, and we send you back the result. It's very neat if you want to run like 10,000 of queries, which you don't want to do on your own hardware. And then, of course, we have so many features in our product. So some people, they're a little bit confused. What can it actually do? How can I make this reliable? How can I make this cheaper? And so we can help them to configure those agents to actually save them money, save them token costs.

• 50:20Speaker 00

  So if you guys offer both the browser technology and the L L Ms as a hosted service, does the, does the end customer get to choose which models they're using? Or do you guys select for them like, you're all getting 4-0?

• 50:36Speaker 05

  In the beginning, we did this, but because there's such, the biggest cost part is definitely token usage in all of this.

• 50:43Speaker 00

  Yeah.

• 50:43Speaker 05

  It's like, let's say for T B D four O, three cents per step. And if we offer them cheaper models, we can offer it for one cent per step or even cheaper. And I think this is where those agents get really, really powerful. If you can run hundreds of agents in parallel for you for a couple of dollars per hour. I think this is where we see a lot of productivity boost.

• 51:03Speaker 00

  That's interesting. I hadn't done the math. If it's one cent per step running a hundred, it could be dollars per hour, but who cares? Humans cost tens or hundreds of dollars per hour. I'm curious, what are the first customers you guys are seeing? What industries, what big companies, little companies, startups? I'm really curious, who's come knocking on your door?

• 51:23Speaker 05

  Some are individuals who want to optimize their outreach. Some are QA testing companies, okay, who write, have millions of scripts to test their websites. And the scripts break and they now want to use agents to test their websites. Or even white coders, you know, who white code their website, then they need to test it. Just now want to use browsers to test their website. Then a big industry is definitely around form filling. And form filling goes really, really wide. It can be for C I M, your contacts. It can be to fill out, I realize there are millions of people on the planet who do the entire day, nothing other than receiving a PDF, example, with patient data, reading the PDF and filling

• 52:07Speaker 05

  out a form based on a PDF. For companies, it's just about converting data from one format to the other. What's happening is some internal database. And there are millions of people.

• 52:18Speaker 00

  Sad.

• 52:19Speaker 05

  Who just get an invoice and need to fill out an internal database system. And with this, millions, millions of people.

• 52:27Speaker 00

  O C R isn't new. Why are they still doing it by hand?

• 52:33Speaker 05

  I don't know, it's this, it's this, many companies, they have never built this, this, you know, you have OCR, you get the data out.

• 52:39Speaker 00

  Yeah.

• 52:39Speaker 05

  How to fill out the form, you know?

• 52:41Speaker 00

  It's amazing that some of the stuff I, I guess I just assumed existed, maybe didn't. And so now we're filling in those workflow gaps with AI tools. Gosh, all right. Well, you, you, you guys also raised, um, I think it was a $17 million seed round, but I'm curious about how the round came together and what it was like to fundraise for your sector in 2025.

• 53:04Speaker 05

  Yeah, it was definitely one of the wildest weeks of my, of my life. But while we got all this open source traction, of course, many investors reached out and we scheduled them all to YC Demo Day. So like beginning of, of March, all in one week. Okay, so I had like 140 investor meetings in this one.

• 53:20Speaker 00

  You had 140 investor meetings in a week?

• 53:22Speaker 05

  Like I, I, I started, I scheduled them one by one. Okay. And then the interesting part was like the week was really crowded. So some investors, of course, they got four more and they thought, okay, they don't got a spot in the round. So they emailed me, hey, I really want to invest. Can we meet earlier? I said, no, we start here. And then he said, okay, just take my money, take uncapped saves without valuation, without meeting. And we got like uncapped saves without even meeting us. Okay. And this was before, before, and we can march. So this got us a lot of leverage. Then when we actually started with the meetings, we just said, oh, we could just keep going with, with the uncapped saves, right?

• 54:00Speaker 00

  So I guess what's coming next for browser use that people have been asking for, requesting or kind of pushing you towards?

• 54:08Speaker 05

  It's definitely for all those agent companies, also all who came before us, like multi on and some. The core problem is always the reliability of those agents. It's, you can create super fast, super cool demos of browser use. Okay, you just prompted and wow, you have a super cool demo, share on X and you win the hackathon. But to make them really reliable, to get like the last mile, then it's, it's really, really hard. There were many companies with millions on, on fine tuning those agents. And I think this is the, the core challenge in this. And this is what we are currently trying with workflow use, that we combine deterministic scripts, which you can record with agents who then step in to heal those things. And yeah, this is what the big feedback which we get from many companies that they.

• 54:51Speaker 05

  More reliable, faster and cheaper agents.

• 54:55Speaker 00

  And did you say that you can basically send an agent in to fix whatever breaks? So you can essentially use AI to kind of plug the gaps in existing AI tools?

• 55:04Speaker 05

  Yes, exactly. Previously, you would have like a playwright script, an automation script that runs. And then at one point, the website changes and your script just breaks and says, okay, I didn't find this identifier. If this arrow occurs, you can just now switch in browse views and say, okay, what, where should I click on? What's the next step? And it heals that script, basically.

• 55:26Speaker 00

  That's gonna...

• 55:28Speaker 00

  You know, I'm always really torn between celebrating technology and our advancements and how quickly we're building new things. And then I also worry about the economic impacts. So I was about to say, that's amazing. But then I'm thinking about all the people who probably get paid right now, like the PDF converters to fix those things. Uh, we're going to have to find new things for them to do, but this is going to be awesome when it's kind of pervasive and we're all just a lot more capable.

• 55:48Speaker 05

  The awesome thing about this, 99% of the tasks which people reach out to me, they hate those tasks. Like no one likes to fill out one million forms per day and just read PDF and fill out forms. And this is the amazing part, like I haven't seen any, any task where people would say, I love this. Because all of those tasks are super repetitive. And I think as soon as you do a task 1,000 times, you, you just hate it.

• 56:14Speaker 00

  Oh, absolutely. I, I think in 20 years, we're going to look back and be like, thank God we had AI automate all the boring stuff. In the meantime, we'll deal with the economics of it. Um, one last question before I let you go, which is San Francisco. You know, you guys are currently in Switzerland as, as we talked today. Again, thanks for staying up so late, but I'm curious about just how important you think it is to be living in the Bay. I've heard from a lot of founders that they think it's the place to be. But I'm just kind of curious from your perspective, uh, why it's the right place for browser views?

• 56:44Speaker 05

  And you know, when I, when I grew up, I, I often went to church and I, I could, I could pray from home, but I go to church because there are many other folks and many other people. And that's why we go to San Francisco, because all the agent builders are in San Francisco. And we don't want to pray for our, our self.

• 57:03Speaker 00

  So San Francisco is by analogy, the AI church.

• 57:07Speaker 05

  It's a church.

• 57:08Speaker 00

  And, and that makes Sam Altman the Pope?

• 57:14Speaker 05

  I wouldn't say so.

• 57:15Speaker 00

  Well, I mean, I think Sam, I think Sam thinks so.

• 57:17Speaker 05

  He's still young.

• 57:19Speaker 00

  Touche. Wait, how old is Sam? I actually have no idea, 45? That's old enough to be Pope. He can be President. Yeah. All right, Magnus, when is the next time we should have you back on? What's the next big release that we should have our eyes out for?

• 57:33Speaker 05

  Yeah, I mean, we now release this workflow, use this super beta breaks all the time. So now we will really iterate on this and get a lot of feedback and make this really, really awesome.

• 57:44Speaker 00

  All right. Well, what's the URL? And just before you go, what's a job you are looking to hire for?

• 57:50Speaker 05

  Yes, the URL is browser-use.com for main website. And there you can reach all our GitHub projects. And the main job is currently really correct, correct builders who are really obsessed, who want to join the hacker house and just love to, to build all day. Can be, can be young white quarters who, who just want to ship cool stuff and create cool demos. Can be really browser gods who have 10 years of experience around fingerprinting for browsers. Can be people who built the infrastructure for this. But I think there will be many problems which no one has ever solved before. And we just need obsessed, correct people who can solve problems.

• 58:31Speaker 00

  Well, you heard it from the man himself. Magnus, thank you so much. We'll have you back on the show. And in the meantime, good luck. Listen guys, I've been doing this job for a thousand years and one thing I can tell you very confidently is that talking to founders never gets old. You learn something, you get a little peek at the future, but also just their enthusiasm always leaves me pretty darn gassed up. I just love it. We have a lot more of these interviews coming up, so get excited. And if you want to get a peek at which companies might show up on TWIST, well, twist500.com for more. And as I said at the top, [email protected] if you want to suggest a company for us to look at. This is Alex. This is Twist. We'll talk to you on Monday. Bye.