Manton: Hi, this is Manton. Today I have a special episode featuring Vladimir Prelovac, founder of Kagi.
I’m excited to talk to you, Vlad, to chat today about Kagi and the open web. And I’m excited to talk to you. How are things going?
Vlad: Going great. Thanks for having me. Looking forward to our conversation.
Manton: Immediately Kagi resonated with me just because it’s — I’m so sick of ads everywhere on the web. It seems very ambitious to create a search engine, but just the idea of like, “Pay us, we’ll give you the best search results, clean results, no ads, no gimmicks.” It’s very compelling to me that someone is trying to do this.
Vlad: Yeah, I once read that ads are a tax on the web for the most vulnerable people. Tech-savvy people use ad blockers and things like that. But when you think about it, that means basically that the most vulnerable and least tech-savvy people are the ones paying this giant tax when using the web.
Manton: Mm hmm.
Vlad: And yeah, for two decades almost, we had this being the main business model of the web, which led to all sorts of problems and among them, the web becoming almost unusable. For something as important as consuming information, I thought that a better business model would be something where the transaction is clean and there are no intermediaries between you and the information provider. And hence, Kagi, a paid search engine.
Manton: No, it makes sense. Did you feel when you started that — I mean, I have thought from time to time, “Oh, I should work on a search engine.” Like there’s so much opportunity there. But of course I’ve never attempted that really because it just feels like a monumental task. Did you have any concerns when you started or did you just say like, “Someone’s gotta do this? So we’ll give it a shot.”
Vlad: Yeah, luckily, I think I was born without the gene for monumental task aversion. So I gave it a shot anyway. And there are multiple things that make it monumental, not just the act of building the product that people will actually appreciate, but also wanting to pay for in the sea of free search engines, free in air quotes.
So, yes, but fundamentally, I felt that the society absolutely needs something like this. I needed it. My family needed it. I knew my kids will need that. And because nobody else was really willing to give it a shot, I did, for better or worse. You know, fast forward five years and here we are, almost 50,000 paying customers and growing strong. Obviously there was demand and I think people are just starting to realize the necessity for something like Kagi.
Manton: Yeah. And it feels like the timing is pretty good because search is in, I don’t know, it’s an interesting place right now. Like you have AI, obviously disrupting search. You have people just kind of burned out on big tech companies. So I don’t know if that makes it easier or harder for you, but it feels like a good time to sort of reinvent and think about search. I don’t know if you agree with that.
Vlad: Yeah, I think you’re right. I cannot say the timing is bad. Although when we started in 2019, LLMs did not exist. So I was not prepared for the impact that will have on the way people consume information. Obviously there’s a pretty big change in paradigm happened and a lot of people consume information through LLMs or AI chat. So that has profoundly impacted the way one should think about product in the search space. And interestingly, now, instead of competing with free ad-based, we are now competing with free VC-subsidized AI chatbots.
Manton: Right.
Vlad: So anyway your turn and we are somehow competing with free, which is also a pretty hard thing to do.
Manton: Yeah, for sure. Although it’s kind of interesting because in some ways, I mean, I assume most like ChatGPT users are free, by far, you know, because they have hundreds of millions.
Vlad: Yeah, the vast majority.
Manton: Yeah, but they do have people paying $5 a month or $10, you know, whatever. I guess it starts at $10 a month. So I wonder if some people are being, I don’t know, with a new type of product, there’s an opportunity to reset things.
Vlad: Yean.
The hopeful side of me says people that are paying for AI, it will feel natural or more natural to pay for search if they’ve already been paying for ChatGPT or something like that.
Vlad: Right. Yeah, I agree with that. This notion of paying to consume information has been introduced, you know, most recently, thanks to chatbots, basically. ChatGPTs of the world. So that is definitely helping educate people that, you know, when something is as valuable as information, then perhaps it is okay to pay directly for it instead of having intermediaries pay for your queries and searches.
So I think for sure the question remains will people ever think about going back to search exclusively and probably the answer is no. Search is going to somehow get embedded into this new way of consuming information, which is very natural language oriented, even voice oriented, right? And definitely more powerful in terms of kind of queries or questions you can ask that, you know, the query space has exploded probably by the factor of two orders of magnitude. And people are able to ask all sorts of questions that they haven’t been able to in classic search engines.
So from Kage’s standpoint, we started as a search engine, but we are adopting to this changing climate of the way people consume information and that presents challenges on a product side of things and so forth. So yeah.
Manton: Yeah, for sure. And it’s like, how do you, I mean, it makes a lot of sense to me for you to do both and figure out how they sort of mesh together. I use AI a lot now to just ask random questions, because there is something about just having kind of all the world’s knowledge there to just ask in different ways. But there’s also times where I really want a web page. I’m looking for something that I know is on the web that I want to read and I don’t want the chatbot summary, right?
And so, yeah, I’m glad you’re experimenting with it and thinking about it because everything feels a little bit in flux, like what people want. But I don’t know, I love the web. I hope people always want to look at web pages and have some of that traditional web feel.
Vlad: Yeah. And let’s just remind ourselves that AI or the LLMs would not exist without the web. They’ve been trained on the web.
Manton: Right.
Vlad: So preserving the web I think is of utmost importance for the AI companies.
Manton: That’s true. We all want the same thing.
Vlad: They may not be acting as if their business depends on the web. They almost see the web as this resource that you once use and don’t need anymore. But yeah, that could pose a pretty big problem. So yes, one of the things that Kagi does is as you probably know, we have projects like Kagi Small Web that surface small websites, personal blogs, human discussions directly into Kagi search results. And this is one of the reasons people really love using Kagi because there is this element of discovery of sites that they would never find on mainstream ad-based search engines.
Vlad: And so we try to do a lot of things to preserve that the way the web felt 20 years ago, whenever you went to, you know, it was a site made by a human sharing their thoughts and opinions and self-expressing. And today, unfortunately, that’s a very small part of the web, probably less than 5% is made in such a way, and 95% is just a giant machinery for content producing with the sole purpose of being monetized by ads.
Manton: Ugh.
Vlad: So yeah, it is a hard thing to do, but it is the right thing to do. And we’ll continue to surface and help promote the human writing in Kagi search results.
Manton: I absolutely love what you’re doing with Kagi Small Web. I was looking at it earlier and I guess I had never looked behind the scenes of, and it really is just a list of, you know, tens of thousands of small websites and blogs.
Vlad: 15,401 as of today.
Manton: Yeah. I love it. And it’s, it’s curated, you know, I mean, someone could say something doesn’t belong there or if a website gets taken over by a spammer or something.
But yeah, that’s the web that I love. And like when I’m reading people’s blogs and subscribing, I mean, that’s the web I see most of the time. And I feel like, I only get sucked into the web of SEO and content farms, usually when there’s some problem somewhere, or like someone registered on Micro.blog wanting to just create like a spam account that we then have to go delete. it’s frustrating to see that.
And I’m sure with AI, it’s going to be even worse. There’s going to be so many auto-generated slop of content that’s just out there. Small Web also has a really nice kind of old school web design and feel to it that I really love. It’s really good. It’s a nice throwback.
Vlad: Yeah, exactly. Mind you, that’s how I started with the web. I think I wrote my first email in ‘94 or ‘95. And my first websites look a lot like the… What was the name of that popular website builder back in the day? GeoCities, something like that.
Manton: GeoCities maybe, yeah.
Vlad: Right, yes. And all the websites were these, you know, funky looking people just throwing out themselves on the web, right?
Manton: Yeah.
Vlad: So yeah, I do appreciate that, that age of the web. And every day you would explore and find new sites and new humans writing about interesting things. But it did become much more hard to do that over the decades as the ad business basically model just proliferated millions and millions of ad-driven pages with low quality or no quality content at all.
Manton: Yeah, it is frustrating. I mean, I love that. I think the way I think about it is sort of like when, I don’t know, it felt like there were a lot of blogs, you know, 10 or 20 years ago, but I know there are more, but they just feel a little bit sometimes like they’re being overwhelmed by all the junk on the web. But I know there’s lots of people that it just they’re writing because they love to, they’re sharing on the web whether they’re discovered or not, whether they’re you know people are finding their blog and they have one reader or a thousand you know it’s it’s hard to tell because there’s usually not any monetization or you know it’s just people are publishing to the web because they love it.
Vlad: Yeah, and many of these people are experts in their areas too. Like there are many people, many blogs in Kagi’s Small Web that are written by, you know, Nobel Prize winners. And, you know, this content is usually undiscoverable on mainstream search engines. And it’s just such a shame. So yeah, we are really trying our best there to preserve that and encourage people to write more. Obviously, as Kagi gets more popular, it will send more visitors to these websites. But even today, people just randomly find about Kagi by going through their analytics logs and seeing a Kagi referrer. And then that’s how they find about us. So I guess the traffic is already substantial in some way.
Manton: Hmm, that’s interesting. I remember when you announced Small Web, but I don’t, I wasn’t paying attention enough to remember how you kind of bootstrapped it. Did you start, did y’all just come up with a list to start with of random sites or how did it grow?
Vlad: Yeah, we actually have the list of sites where we bootstrapped it from in the README from the open source repo on GitHub. But it was mainly going through Hacker News threads or Hacker News submissions and then finding blogs there. So that was the main method. And that gave us the first few thousand or something.
Manton: Okay.
Vlad: And then there’s just people kept adding. And every day we add, I would say 10 to 15 new ones.
Manton: Oh, wow. Nice. I love it. And it’s like, it feels like a lot of people or maybe not a lot, but some people are trying to get back to this idea of sort of a smaller curated list of the web that you should check out. So like another one that’s kind of new is blogroll.org. I like what he’s doing with that, but it’s, it almost feels like, I mean, you rewind way back to like Yahoo! directories and you know, just like a completely different way of browsing websites. And I don’t know, it’s fascinating.
Vlad: Yeah, I would say the main use of this from somebody using a search engine is exploration. So when you are exploring a topic, the content on the small webs tends to be much higher quality than on the mainstream websites. And the other one is when you want opinion on something. Again, then having that personal, in many cases, expert opinion on a topic is much better than an article on Wikihow or another ad-driven website. And we promote these websites pretty high up in our results, as high up as number two or number three would probably, if it’s relevant to the query, it would surface in Kagi search results.
Manton: Nice. I was wondering about that. I didn’t know how it was integrated. That’s really great. So if someone has a blog and they’re listening, they should add themselves to it, right? I mean, that seems like a no-brainer, to get some more visibility and to, you know, someone will randomly click the random button or next site or whatever it is, and maybe stumble upon your site.
Vlad: Yeah.
Manton: I guess that’s part of the strength to Kagi is that you can kind of blend a few different things together. Because I’ve never really been clear on how you have your own search index, but you’re also like hit other services, right? And you sort of combine the results or I don’t know, is that just like a secret sauce kind of thing, or can you talk about how that works?
Vlad: It’s not a secret. So we use all the external indexes we can pay for. And then we also have our own index, focusing on the non-commercial web, because that’s the one that’s usually missing from everybody else’s. And then combining the two with a couple of things, one of them being downranking sites that have a lot of ads and trackers and raising the ranking of sites that don’t is probably describes the 90% of Kagi algorithm. The other strength of Kagi is that users can personalize their search results, meaning they can block websites they don’t like or raise the ones that they like.
And I think when you combine these two, so our internal algorithms, which are pretty transparent in that way, and then users personalizing the search results for themselves, you often get results that far surpass any search engine out there. And by the nature of things, because we use all the search indexes we can get our hands on, the added benefit of that is basically if you cannot find it on Kagi, it doesn’t exist on the web.
Manton: Hmm, that’s really interesting.
Manton: This is a little bit of a tangent, but have you been following some of the Google antitrust trial and the idea put forward that Google should, as a remedy, Google should just open up their search index somehow? I mean, that seems very hard to do, but I’m curious if you think about that.
Vlad: Well, actually, we posted this blog post at the end of last year when the remedy phase was coming out. We actually were the ones that I don’t want to say originally, but we did propose that to be as the main remedy. And I’m very glad that it got in.
And it’s now an official remedy that DOJ is pushing for. And I really think that that one would solve a lot of problems as well as proliferate a very healthy search landscape. Because if that happened, then anyone could basically bootstrap a search engine. And you would have a search engine for a small town that somebody would build just optimizing for that small town or a small niche like cats. So you could imagine tens of thousands of new search engines being basically possible.
So I think that’s a very interesting remedy. And otherwise, if that doesn’t happen, basically the expectation for any new search startup would be to rebuild the entire index of the web, which is a monumental task and super hard to do. So hard that, you know, Microsoft tried doing that for the last 20 years and spent $100 billion and still got nowhere. Like the quality is still pretty bad. It’s it’s almost impossible for a startup to do that. It would be akin to asking someone wanting to put a train on the tracks to rebuild the track infrastructure, the train infrastructure, the railroad infrastructure of the US before putting the first train on the tracks. So it just doesn’t make any sense.
And Google has accumulated that index over the years, many years of using a monopolistic position and it’s just virtually impossible for that to be replicated. So that’s why in my mind, this is the remedy that makes the most sense. And also would be completely in the spirit of the Sherman Act, which is all about creating a healthy marketplace. And as I said, this would proliferate many, many search startups, including competition to Kagi, which would be great. And I, you know, I’m completely welcoming that.
Manton: That would be amazing. Just the idea of having more than a couple search engines would be, just as like, I think we’re going to see more browsers also. And I want to ask you about y’all’s browser. But I also like the railroad analogy because obviously with antitrust, there’s lots of parallels there with yeah, having a monopoly on the tracks.
It would be interesting because like I kind of joked that I would love to make a search engine too. And it’s just so daunting. And if it was more accessible and the data was more like public data where people could build on it, that would be fascinating.
How is the browser going? To be honest, I haven’t tried it, I’m sorry. I got sucked into using Arc last year and I’ve kind of stuck with it. But it’s Orion, right?
Vlad: Yes. Our browser is built to be user centric with a very clear business model from day one. Same as search, meaning it is user funded, meaning there people paying for it, which is probably even stranger than paying for search.
There are hundreds of browsers out there and all of them are free. And yet we have thousands of people paying for our browser. And this is because there’s a fundamental promise of trust and being user centric, which when you think about is even more important for a browser than search because browser is literally the most intimate piece of software you have on your computer. You’re in it probably eight hours a day, depending on what kind of work you do.
And this thought that for the last two decades, the most intimate piece of software that we use has been paid by somebody else. And that somebody else mostly being advertisers is just mind blowing to me. So, Kagi is really about creating an ecosystem of tools that are all user centric, user funded. There’s a very clean, clear transaction, you pay for the product and you own the product. And there are no intermediaries that I think is really, really much needed as some sort of counterweight to the only other ecosystem that we know of, is, you know, somebody else is paying for your searching and browsing. And that just sounds like such an outdated idea to me.
I like to use the analogy with organic food. And how 30 years ago, nobody cared and we all ate junk food, or at least I did. I know many people did as well. Nobody cared. And there were a couple of people saying we need to eat better and watch what we put in our bodies. And, and that will be more expensive. But you know, and then 30 years after, you know, I’m buying avocado in Whole Foods for $3 a pop or something. And everybody’s doing that in a way.
So, sometimes these kinds of ideas need time to propagate. And I think we are just at the beginning of people understanding the importance of information they put in their brains and how valuable that is and how this information shapes us as humans, shapes our thoughts, behaviors, shapes our understanding of science, of health. And for something as important to let, and this is just a broader thought, let the entire societies have their citizens consume information from, you know, world’s largest advertising companies is just mind blowing. But that’s where we are. Slowly transitioning away, I would hope.
Manton: I think so. People are becoming more aware and like just getting back to what you said at the beginning when I was kind of complaining about ads that when your users pay for the product, your business model is aligned with their interests. You know, you make the product better. They’re happy or there’s nothing in between.
I feel the same way with everything I do with Micro.blog. People pay me for blog hosting. It’s aligned with my business. It makes the web better. Hopefully lots more small websites sites out there. And then when you throw ads in, it has this kind of, I mean, we’ve even seen this, I think with Apple and, and Google paying Apple, has this like corrupting influence where Apple is… People respect and trust them because they’ve put the user first and they’ve cared about design. They cared about privacy, but then they, you put enough billions of dollars in front of them and I don’t know. It just kind of poisons the whole thing. And the model, yeah, it’s just not as clean as people paying.
Vlad: It does poison the well in a way and that will have long term consequences obviously on how people perceive Apple as a company in this case. And the entire goodwill brand equity that Apple built was to building products people love and pay for. That’s it. That was a very simple recipe.
Manton: Mm hmm. Yep.
Vlad: And yes, I agree there. They’re sort of playing with something that is not in the spirit of the original product vision.
Manton: The ad money and the services and it shifts the company a little bit.
Vlad: A lot, I would say. But still, it’s a fraction of their revenue. It’s not as… But if they continue to pursue that, I’m sure that the products of the future will have very different outcomes for the users and they will not be good. And I’m hoping that Apple will abandon that idea and find different ways to continue creating products people love and worth paying for it. Because when you put ads into something, you’re basically admitting it’s not worth paying for. And you’re admitting it’s, it’s basically built for entertainment purposes only at best.
Manton: That’s an amazing way to put it. I think definitely on the social media side, I feel like that’s certainly the case. Just all that matters is the attention, how many eyeballs you can get. We talk about the small web and the value of someone an expert in their field — or not, you know, just writing something on the web and sharing it and making the web better — and Facebook doesn’t care about the quality of the content. They only care whether the person is engaged so they can serve more ads. That’s the metric that matters and it ruins everything.
Vlad: Pretty much so.
Manton: It is great talking to you, Vlad. Thank you so much for joining me. Best of luck. I’m really excited to see where you’ll go. I mean, I have been slow to get into using Kagi more. I should be promoting it more probably. But I love what y’all are doing and it hopefully it’ll be, yeah, like you said, maybe we’ll have more search engines in the future. And like, there’ll be people five years from now that will be inspired by y’all’s success.
Vlad: I hope so too. Thanks for having me on and encourage you to try the browser. I’d be happy to hear the feedback and how you’re thinking about that. But yes, thanks so much for having me and giving opportunity to tell the story about Kagi.
Manton: Awesome, thanks. Have a good one.
