Gunning for Google with Perplexity CEO Aravind Srinivas

In this episode, Aravind Srinivas of Perplexity AI returns to the show. They discuss Perplexity’s growth to millions of queries per day, CEO Aravind’s favourite Perplexity use cases, and how Perplexity ships fast in a competitive landscape against Google and major AI live players creating their own search engines. If you're looking for an ERP platform, check out our sponsor, NetSuite: http://netsuite.com/cognitive We're hiring across the board at Turpentine and for Erik's personal team on other projects he's incubating. He's hiring a Chief of Staff, EA, Head of Special Projects, Investment Associate, and more. For a list of JDs, check out: eriktorenberg.com. SPONSORS: Shopify | NetSuite | Omneky Shopify is the global commerce platform that helps you sell at every stage of your business. Shopify powers 10% of ALL eCommerce in the US. And Shopify's the global force behind Allbirds, Rothy's, and Brooklinen, and 1,000,000s of other entrepreneurs across 175 countries. From their all-in-one e-commerce platform, to their in-person POS system – wherever and whatever you're selling, Shopify's got you covered. With free Shopify Magic, sell more with less effort by whipping up captivating content that converts – from blog posts to product descriptions using AI. Sign up for $1/month trial period: https://shopify.com/cognitive NetSuite has 25 years of providing financial software for all your business needs. More than 36,000 businesses have already upgraded to NetSuite by Oracle, gaining visibility and control over their financials, inventory, HR, eCommerce, and more. If you're looking for an ERP platform ✅ head to NetSuite: http://netsuite.com/cognitive and download your own customized KPI checklist. Omneky is an omnichannel creative generation platform that lets you launch hundreds of thousands of ad iterations that actually work customized across all platforms, with a click of a button. Omneky combines generative AI and real-time advertising data. Mention "Cog Rev" for 10% off. X: @labenz (Nathan) @AravSrinivas (Aravind) @eriktorenberg (Erik) @cogrev_podcast LINKS: - Perplexity AI: https://www.perplexity.ai/ - Part 1 with Aravind: https://www.youtube.com/watch?v=YSYQZZu4MEM TIMESTAMPS: (00:00:00) - Episode Preview (00:01:12) - Introducing Aravind Srinivas and Perplexity AI (00:05:05) - Perplexity's growth to millions of queries per day (00:06:21) - Perplexity's retention and market share vs. ChatGPT (00:08:32) - Improving reliability, speed, accuracy, and new features (00:10:33) - Getting to daily usage from weekly usage (00:13:21) - Aravind's favorite Perplexity use cases (00:18:34) - Issues with lack of sufficient web coverage for hard queries (00:18:38) - Sponsors: Netsuite | Omneky (00:23:13) - Building their own search index and models for self-sufficiency (00:26:29) - Using their own index vs real-time web queries (00:28:43) - Integrating default search and Copilot companion experiences (00:31:49) - Training their own models as the only viable long-term option (00:36:56) - The web becoming more like an API accessed via AI assistants (00:41:23) - Not trying to copy Google's business model (00:44:36) - Spending more on compute than people to get scale efficiencies (00:49:35) - Replacing all parts of the stack with their own proprietary versions (00:53:59) - Shipping fast out of necessity in a competitive war zone (00:58:00) - AI assistants becoming a significant part of human digital activity The Cognitive Revolution is brought to you by the Turpentine Media network. Producer: Vivian Meng Executive Producers: Amelia Salyers, and Erik Torenberg Editor: Graham Bessellieu For inquiries about guests or sponsoring the podcast, please email [email protected]

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

• 0:00Speaker 00

  We are building our own search index and so is OpenAI, so is Anthropic. Like, everybody's building their index because I think in a world where large language models are commodity and the training recipe for them or the weights are just weighing on their open source, the edge goes to the data markets, people who own the best data in the world. And what do I mean by that? It's like, there are like trillion pages on the web, we can't index all of them. So then you narrow it down. You don't even want 100 billion pages in your index. It's not about the quantity here again, right? You want the best web pages on there, that's probably a billion or 10 billion, I don't know. But the best ones that really matter to the knowledge worker, to the researcher, to the curious mind, right?

• 0:46Speaker 00

  If you can capture that distribution really well, there's already a huge moat there. I think there's very few companies that can aim to do this. It's like a chicken and egg problem. In order to do this, you need to have a product. But in order to have a product, you need to have some kind of index. But somehow we broke that asymmetry. So we are at a point where we can dream of being important.

• 1:12Nathan Labenz?

  Hello and welcome to The Cognitive Revolution, where we interview visionary researchers, entrepreneurs and builders working on the frontier of artificial intelligence. Each week, we'll explore their revolutionary ideas and together, we'll build a picture of how AI technology will transform work, life and society in the coming years. I'm Nathan Labenz, joined by my co-host, Erik Torenberg. Hello and welcome back to The Cognitive Revolution. Today, I'm excited to welcome back Aravind Srinivas, founder and CEO of Perplexity AI. Aravind first appeared on the show back in March, and since then, he and the Perplexity team have continued to impress, shipping updates at such a relentless pace and delivering results which so dramatically outshine Google and Bing,

• 1:57Nathan Labenz?

  that perplexity has even started to appear as a comparative standard for accuracy in academic papers. Speaking for myself, I can definitely say that perplexity has become one of the AI tools that I use nearly every day, marking the first time in the last 20 years that a new app has meaningfully displaced Google in my everyday workflow. And notably, when I asked Replits VP of AI, Michael Khatasta, what other companies he'd add to my AI live players list, he suggested just one: perplexity. I think this conversation shows why that could in fact be a very good call. While perplexity is still only about 1/1,000 the size of Google, they are now serving millions of queries per day,

• 2:40Nathan Labenz?

  and their ambition, huge from the beginning, continues to expand. When I asked Aravind in March if he was worried about Google and Bing cutting them off from search index access, he said that he hoped they wouldn't do that. This time, he said that they are working around the clock to build out their own web crawler, their own search index, and yes, their own LLMs. It seems that Aravind now expects the big tech giants to recognize how generative AI startups could disrupt their core businesses and to begin to raise the drawbridges that currently span their proverbial modes. And so he aims to achieve technology self-sufficiency before that happens, such that he can sustain product supremacy if and when it eventually does. That strategy is not for the made of heart

• 3:25Nathan Labenz?

  or for the modestly resourced, and not something I'd recommend to most application startups. But judging purely by their track record over the last six months, I give perplexity a very real chance of success. Briefly, a couple of quick housekeeping notes before we get into the episode. First, ahead of this recording, I invited listeners to submit questions for Aravind, and I wanted to thank listeners Sidharth Ravikumar, as well as one who identified himself simply as John, for some very thoughtful questions. I touched on as many as I could, but wasn't able to ask all of them as I only had an hour with Aravind. But I really do appreciate the questions and definitely plan to invite audience participation again in the future. And second, we continue to get feedback on audio quality, and we're definitely working on it.

• 4:09Nathan Labenz?

  We now send any guests that need one a USB microphone ahead of our recording, and we aim to consistently deliver top-notch audio quality going forward. In the meantime, as always, if you're finding value in the show, please do share it with your friends or post a review on Apple or Spotify, or just leave a comment on YouTube. I recently got an amazing email from a history professor in Canada who got over a major hump in one of his projects when he followed my suggestion from our September 26th episode to fine-tune GPT-3.5 Turbo on GPT-4 reasoning.

• 4:41Nathan Labenz?

  And I also wanted to call out our frequent YouTube commenter AI in check, who said that I look much better without my silly hat. So a tip of my cap to both of you. And with that, I hope you enjoy this peek behind the curtain into one of the most dynamic AI startups in the world today. This is my conversation with Aravind Srinivas of Perplexity AI. Aravind Srinivas, welcome back to the cognitive revolution.

• 5:05Speaker 00

  Thank you for having me again, Ethan.

• 5:06Nathan Labenz?

  My pleasure. I'm super excited about this one and I've been giving some of my analysis on other episodes recently. You have shipped so much stuff and really kind of led the AI search market. We've got a lot of questions also in from audience members for you. So I'm just going to try to give you the questions and let you do most of the talking this introduction. Notwithstanding, first of all, kind of tongue in cheek, but maybe not necessarily. Any truth to the rumors that the search giants are already trying to take you guys out of the market for a big number?

• 5:41Speaker 00

  No, there's only one giant, right? Google, as we speak, they're in an anti-trust case. So they will be the last to come after us.

• 5:50Nathan Labenz?

  I was thinking more Microsoft.

• 5:52Speaker 00

  No.

• 5:53Nathan Labenz?

  If anybody from Microsoft is listening, you might want to look into this because the head to head perplexity versus Bing is a pretty striking difference in many cases. Okay, next question. What can you tell us about the numbers that you are seeing in terms of the adoption, the users, the kind of users? There's been this narrative, you know, chat GPT visits are in decline. I think that's a pretty misleading headline relative to what's really going on. So tell us what what you can about your numbers.

• 6:21Speaker 00

  Yeah, I mean, we saw millions of queries every day, multiple millions of queries every day. So that's where we are today. Yeah, we've definitely grown six to seven X from where we spoke last time. And I think if it keeps sustaining, like say, we talk in another six months.

• 6:40Speaker 00

  And if we still keep growing at this rate, then yeah, we'll be pretty significant in terms of consumer traction and usage here. How does this compare to chat G P T? I think we should just go by the similar web estimates. We are like forty X smaller in size compared to the traffic to get. So that's reasonably reflective even what happens in mobile space. Even though similar web doesn't track mobile, it's pretty reflective of like how much user adoption there is. So from that perspective, we are like way smaller, right? Like two point five, three percent of their market. So we need to like grow more. But the nice thing though is that

• 7:24Speaker 00

  our retention numbers are really good compared to what they have. So they say that in consumer retention is the lifeblood of a product. Like you have infinite life if you have a really good retention, like nothing can kill you basically. Whereas if you have a product that's like attractive and flashy and got like a lot of surge usage, but very poor retention, nobody wants to invest or use that. So from that perspective, we actually have pretty good retention. Some amount of unique positioning, you can obviously try to do many features in one app, but there's just some particular use case that you just nail.

• 8:07Speaker 00

  And for us, it's just this orchestration of search and L L M's together. In terms of Bard, I think we are like fifteen to twenty percent of the traffic in that ballpark, which you would never expect, right? You would think consumer giant like Google will just nail it. But that that didn't end up happening. And nice thing is we're still growing and I think there's just so much more alpha left here.

• 8:32Nathan Labenz?

  Yeah, no doubt. It's all still fairly early in this game. I will say perplexity is one of not that many apps, and I'm sure I have more than most, but not that many that I do go to reflexively now. In fact, I was just telling Zvi, I kind of segment searches now into like two categories in my mind. One is the like quick lookup search for which I still do go to Google, but that's often increasingly more like locating something that I where I know what I want, or I have a very good idea of what I want. And when it's something now where it's a genuinely novel question that I don't know the answer to, sometimes I do go to ChatGPT for that, but often I do go to perplexity, especially if I want something up to date,

• 9:13Nathan Labenz?

  or just want the links, right? So it's definitely become part of my daily thing. I probably am almost a daily active user, if not fully, you know, every single day.

• 9:23Speaker 00

  That quantity is also growing a lot. In fact, I would say we are largely a weekly usage app right now. In order to get from weekly to daily is the biggest hill to climb for a consumer company. Getting to weekly usage is still not easy, but we managed it. But getting from there to daily usage is the hard part. And there are like so many things you need to do, reliability, speed, accuracy, constantly improving the quality of the answer, accuracy, and then new features that engage the user in a different way from just, you know, being a tool, providing something that they can all get elsewhere, making sure that's actually something

• 10:05Speaker 00

  that's valuable to the user, allowing the user to share things that they learn here with other people. There's just a bunch of things you've got to do all at once. Unfortunately, you've got to do a whole at once. And that's the challenge. When you have a small team, like not have any resources. But that's also like, it brings you focus, it brings you the adrenaline, the mindset to grind, that I think it's all exciting.

• 10:33Nathan Labenz?

  Yeah, I think you're in one of the more exciting positions in the space right now. How far do you think this has diffused through like, different kinds of users? I'm so down this rabbit hole that, you know, I have no idea, like, is this mostly people like me? You know, is my mom, users like my mom coming into the picture? How do you think about the profiles of user that you have?

• 10:57Speaker 00

  It's still very early, right? Like you go to an average subway in New York. Nobody even uses ChatGPT there. Here's the difference. In New York, if you're in a cafe, one out of like fifty laptops might have ChatGPT open. In San Francisco, it's like twenty out of fifty laptops will have ChatGPT, right? So among all the ChatGPT users, obviously, many of them will know about us too. And I've seen people even having our tab open or Claude open here. You go to Vachtam Blueboard or Palo Alto, you can see people using Perplexity or Claude. So I think there are people who use our products among the AI enthusiastic crowd. But that's not enough, right? Like, you gotta actually be useful to people

• 11:42Speaker 00

  in a way that they don't use you because they think AI is cool and they wanna know more about it. They use you because they find you useful. And they just tell their friends about it and use it. We haven't achieved that yet. But um the next two years or three years, that's gonna be our focus, right? Like getting to a point where you're just using it because it's not AI, but it's just like really useful. Search and information discovery, knowledge discovery is one of those use cases where you can actually do this. Everywhere else, you're building something new and trying to find product market. You're trying to convince the user to use your thing. In search, the advantage is you're already useful, like

• 12:23Speaker 00

  because everyone needs to search for information. It's like fundamental human need curiosity, right? And like learning new things, discovering new things, sharing those things with people.

• 12:34Speaker 00

  I think that's the um mentality we wanna invoke in you. Like, okay, if you have, in your mind, if you feel like you need to go on Twitter and ask like, what should I eat in New York or like in this area? Instead of going and asking somebody else, like, can you come to our app and ask these questions? And get the eighty twenty, right? I'm not saying there is no human value in an answer. But you get the eighty twenty here. Similarly, instead of paying like a lawyer for like one hour, you could pay them for like twenty minutes. Or like what twenty percent of one hour, twelve minutes, let's say. So that's the sort of thing that we might want to like create in the future. It takes a while, right? It takes a while and I'm fully aware that like this is a journey.

• 13:19Speaker 00

  Today is just the beginning.

• 13:21Nathan Labenz?

  So what are some of your favorite questions that perplexity has answered for you? I'll offer one of my own, but I'd love to hear yours. And then where is it still not quite able to answer the questions as well as you would like?

• 13:37Speaker 00

  The whole story about how we build the product is my favorite part, which is when my first hire, a founding engineer hire, he asked me for health insurance. And I've never been a CEO before. I didn't bother to get myself a health insurance and my other co-founders were married and they had health insurance through their wives. So I'm like, okay, whatever. I don't want to waste time on this, you know, it's just going. And obviously my founding engineer is like, yeah, you know what? Like I need health insurance. Okay, fine. Like let's let me look into it. And I go to Justworks and there's all these kind of different compliance or HICC, blah, blah, blah, like coinsurance. I have no understanding. And we had a Slack bot that was just integrated with GPT three point five.

• 14:22Speaker 00

  And that was just hallucinating and lying and like saying incorrect things. Right? And then we were like, okay, fine. Let's integrate with web search being like, like, you know, at that time we use Bing. I was able to answer all these questions for myself and figure out the right insurance plan and get it. That is one use case that clearly tells you what our product can unlock.

• 14:47Speaker 00

  If you don't know anything about a topic, you don't need an expert. You can figure it out yourself. What were otherwise taking you like multiple hours now takes you like few minutes to figure out. I'm not saying you do, you wouldn't have to go and read the link, but you get. The job done like way faster. So that was when I realized there's true value in this product. Similarly, like recently, one of my um engineers was um figuring out how to do something. Obviously, I don't get time to code much anymore. I wanted to get that task done faster, you know, so I was trying to help them and then.

• 15:27Speaker 00

  So I was looking up some tools myself and then it just gave me the needle in the haystack so fast that I could go and tell them what to do. Another time I wanted to download a video from someone else's tweet. And I didn't know a good website that could just do that. And I just got this website tweetvideo downloader dot com and just does it. If I go to Google, I'd see like twenty links and I'm almost sure the first three or four are spam. Go to perplexity, I get a bunch of bullets and what each thing does, I can go check it out and it's pretty optimized. Another thing I really like is before I go into a new meeting, I can check out that person and their bio and the history and, you know, what company they're working.

• 16:12Speaker 00

  What is that company even doing? So I don't have to waste the first five, ten minutes asking them about it. I'm already like aware. And that gives them the impression that I've done my background research too. Oh, this guy's put time into the, you know, learning more about me. But what would have otherwise taken them? Usually people take half an hour to do these things or would have an assistant that would do these things for them and add notes to their schedule. I don't need all that. I just do it myself, right? Or we are like funding, you know, like from series A. So we were like, okay, some amount of this has to go to, you know, some other investment because you can't just keep the money lying in the bank, not earning interest, especially in inflation, then you're losing money, right? If it's so static.

• 16:58Speaker 00

  So then what is the best safest like if you are if in terms of not taking risk but also still getting a good return? Like I would ask summarize to me all the options and tell me the risk and the reward. If I want this, what is the right choice? Like these are the things that you don't wanna ask chat GPT about these things because you cannot make a huge like investment decision based on what chat GPT says. You can maybe do it to like have a Mr. Beast kind of YouTube video and saying, hey, I would just follow what chat GPT said and like here's what happened over thirty days and like, you know, get a lot of views. But that is if that's entertainment. But real life decision you cannot make without you know, actually having an AI tell you what to do. So I think these are the kind of things where I'm finding a lot of alpha personally.

• 17:43Speaker 00

  I'm a context switcher, so I have to do a lot of different things. Sometimes I go to my mobile engineers and I ask them here like, why is this issue hard? And they tell me something and I don't wanna like keep bothering them and asking them to explain it to me. There's this whole Elon Musk thing of like dig deep and understand everything to the core first principle, right? But it's also a waste of time for the engineer to like uh explain everything to you all the time, especially when they know what they're doing. But you still got to understand so that you might be able to come at it from a different perspective than what they're thinking. So I can go and like just learn about Swift UI components, native rendering, which is like web view rendering.

• 18:25Speaker 00

  So stuff that I have no experience beforehand, I can just learn and get the app on like very quickly. So that's my favorite way of using our product. Hey, we'll continue our interview in a moment after a word from our sponsors.

• 18:38Nathan Labenz?

  I'll give you just two quick ones of my own that I think are differentiated, not just from things that came before, but also kind of differentiated from other options on the market today. One was when I ran into a corruption of some sort of a Docker container environment that I was working in. It was actually in a GitHub code space specifically. And I couldn't get it to update packages, whatever. And I really honestly still don't know what went wrong. And that's the kind of thing that, you know, I probably would not be able to get a good answer from ChatGPT on. Uh now they've of course started to reintroduce browsing too. So it'd be interesting to see how they may be able to start to do better on some of these things.

• 19:21Nathan Labenz?

  But certainly up until the last couple of days, it was like, you know, the specific thing, the package inconsistency that popped up, you know, at some relatively recent point in time that nuked my code space from like last time I rebuilt it to this time. That's not in the training data. And so you have to go find that stuff live. And it was able to do it and not just do it, but actually then convert what it found into commands that I could run. Cause what I asked for were commands that I could run to solve my problem. Uh no, I wasn't really that interested in the intellectual foundations of this problem, to be honest. Uh I just wanted a solution. It gave me a solution. I was a little bit, you know, thinking at the time like, geez.

• 20:03Nathan Labenz?

  Is, you know, should I be running things off of the, off of the web here? And I actually do have a question about adversarial content for you that I want to circle back to too. But it worked, you know, I put in like this very specific string that I was getting and it found, you know, other people who'd solved it and got me the solution. So that was awesome. Uh, and then another kind of just very funny one, totally different, you know, facet of life. I was interested in going to this Ash Nico concert and uh, I didn't know if I would be out of place. I was told I was too old. It honestly did a remarkable job of like, first of all, that's a tough question. Uh, I think it would be a tough one to Google. I don't think a lot of people have written anything about that. But I thought it decomposed, it seemed to, and I want to ask you about your decomposition strategy too.

• 20:46Nathan Labenz?

  But it took a very smart approach to that question by going and doing kind of like billboard style research on like, what are the demographics of her fans and that kind of thing? And just trying to bring me some sort of triangulation on the answer, you know, for a question that like, there honestly probably is no real fundamental truth of the matter, you know? I mean, it's, there'd be a lot of maybe different perceptions in the room, uh, but no single truth. And yet I thought it did a very good job of giving me, you know, a fact-based answer that genuinely, you know, kind of helped me feel like, yeah, I could go do this concert. Uh, so those are cool. Other side, where is it struggling right now?

• 21:27Nathan Labenz?

  I'll just tell you that for me, it seems like when it struggles, when I don't get something that's like meaningfully answering my question more than I already, you know, knew. It seems like the biggest problem is that it's not able to find the information. Like there, it may or may not be out there, but sometimes I feel like I'm still kind of getting the more like first page of Google things where I really want, you know, something a little deeper. But I'd love to hear where you think the, the frontier is.

• 21:55Speaker 00

  I agree. I think the co-pilot usually gets stuff that the default search doesn't. There's a reason we respond like that, like there's a reason why we respond, oh, this we not, we don't have sufficient information to answer. Cause like, we don't want to hallucinate. That's why we introduced this interactive search companion co-pilot, because that goes and queries the web on the fly, real time, like an agent, right? And um, I think as you use the product more, as we get, just get more scale of usage, we'll be able to handle most of the lack of sufficient information part. Because at the end crawling the web is not like easy, right? You gotta have like so much coverage of the web to get you the right, right answer.

• 22:39Speaker 00

  So that just needs more infrastructure, usage, scaling. Uh, so it's more of a scale problem. But I agree with you that as a user, you don't know when to use Copilot and when not to. That's why we're shipping this retry button on the web that we're testing where for every query, you can retry with Copilot and usually people like that. And if that works, we'll also ship it on the phone. But again, the user has to think, right? So you gotta be such a smart user to know like, what i- is this something you would retry with a smarter AI? Or is this something you would wanna retry with a Copilot agent that

• 23:15Speaker 00

  goes and queries the web better? That, that requires you to, you know, like, understand, okay, if the answer is like, there's not sufficient information on the web, you'll use something that queries the web. Uh, but if the answer is something that you feel is not, um, is sort of like hallucinated, then you might wanna use a smarter model, right? So these are things that we are still not understanding correctly ourselves.

• 23:43Speaker 00

  And we might wanna like beat the dust a lot and get it right. But you're, that's one, definitely one place of improvement. Another place of improvement I see personally is, um, how do I unify both, right? Like, there are times you just wanna get to a particular website quickly, a subreddit quickly, and you're, you're so used to just getting the results so fast. And how do I unify both these interfaces together? We are doing that, that's why we put, display the sources at the top so that somebody doesn't wanna wait, they can just, um, click on the source. I think even there, the latency can be improved even further. But that again is an infrastructure and scaling problem.

• 24:24Speaker 00

  And, uh, there are some, lot of smaller quality of life feature improvements that we can do. Like sometimes a query, if you need to be very precise, so that requires you to be a good English speaker. A lot of people are not able to articulate like the actual query in their mind to a question. Right? Asking good questions is just human skill. It's very hard, actually. It's not that easy for most people. That's why Google is still king, because the reason it's still king is because most people, either it's like a Google phenomenon or whatever, but they just like entering keywords. They they don't actually know how to ask a question. And Google's made that like okayish, you know, it's okay for you to not ask questions.

• 25:06Speaker 00

  So I think that part probably is like um something we should figure out how to fix.

• 25:11Nathan Labenz?

  Yeah, interesting. Okay, a couple detail follow up points. A minute ago, you said at that time, we were using Bing and now you kind of said like, you know, we gotta crawl more the web. Have you moved off of the search APIs at this point?

• 25:26Speaker 00

  We use a bunch of APIs, but we don't uh rely on any one person. In the sense of any one person shut us off, like we will be fine. We are building our own search index and so is OpenAI, so is Anthropic, like everybody's building their index because I think in a world where large language models are commodity and the training recipe for them or the weights are just rang on their open source, the edge goes to the data markets, like people who own the best data in the world. And what what do I mean by that? It's like, there are like trillion pages on the web, you can't index all of them. So then you narrow down. You don't even want hundred billion pages in your index. It's not about the quantity here again, right?

• 26:11Speaker 00

  You want the best web pages on the internet, that's probably a billion or ten billion, I don't know. But the best ones that really matter to the knowledge worker, to the researcher, to the curious mind, right? And if you can capture that distribution really well, there's already a huge moat there. I think there's very few companies that can do this, aim to do this. It's like a chicken and egg problem. In order to do this, you need to have a product. But in order to have a product, you need to have some kind of index. So it's like a chicken and egg problem. But somehow we broke that asymmetry, as in we somehow like broke that chicken and loop.

• 26:56Speaker 00

  So we are at a point where we can dream of being important in the, you know, soon to come future where many people have good L L M's, not just one company. And that's fair, like all these kind of abilities are important, become actually more important.

• 27:14Nathan Labenz?

  Just quickly on the Copilot, I basically always use Copilot. I thought it was kind of just strictly better. How do you see the two different modes? Cause it sounds like it's not so simple, actually.

• 27:24Speaker 00

  Yeah, I I use it too. By default, it's not turned on, right? And and there's also limited uses per day and like people like using free things, you know, you know what I mean? So that's why we tried to reduce the cost of the Copilot by switching the router to uh the GPT three point five fine tune instead of GPT four, so that we can afford to keep it free for like more people, more users per day and things like that. So I think that's that's um at some point, it'll just also the UI change, right? And for the default search, the answer came first and the sources came below, but now it's unified. It's all sources at the top, answer below. So I think our goal is to unify everything

• 28:06Speaker 00

  and make it all look like one single UX and workflow that the free and the pro plan are just simply about like number of queries you get per day.

• 28:17Speaker 00

  That's that seems more reasonable to me, but right now it's not happening yet because there is this part about like whether you go and query the web online or not. I think that part is gonna take a while to like really nail.

• 28:31Nathan Labenz?

  Gotcha. So the kind of default landing experience uses the index, but doesn't actually go retrieve real time information and the the Copilot difference that again, I basically don't use Copilot is it asks you the questions. That's a Copilot specific and also it like.

• 28:48Speaker 00

  Yeah, it it asks you clarifying questions. It it goes and queries the web online and scores multiple pages. It does a lot of actual search on the on the fly, right? It has more agentic behavior there. Both products are useful in different ways. Like sometimes you just want speed, like, but if everything worked with, you know, lightning speed that we serve with the default search with three point five, and you always got accurate answers, there's just no bigger output in the world other than having that, right? If everything you ask can be answered, it's so really fast. It's almost like crazy, but how can this priority, it's criminal for that product, it's almost right.

• 29:29Speaker 00

  But we haven't achieved that, like you clearly need G P T four for like harder queries, you need the sort of agent to co pilot thing for on online searches, we need to do more to make everything work more reliably and better. So in the short term, I don't have a clear answer to you of like, you know, when to use copilot, when not to. Cause if we knew, we would have shifted into the product, right? Honest answers, I don't know, and neither does the user know it all the time either. So your copilot is useful, but you don't exactly know. Okay, why, like, why shouldn't I just not use it? You know, so because the not using part also works pretty well many times. So then you're like, okay, when should I use the copilot? Like, is it like, I should use it for harder queries? So that's one way.

• 30:14Speaker 00

  Then how do you know in your mind what's harder and what's easier? If you need to decide all this when you're coming to product and it's not a good product, like product should think for you, right? Not, not the other way. That's like a bunch of work to do here. At this point, that's all I can say. We don't have a clear sense yet.

• 30:30Nathan Labenz?

  Yeah, interesting. So I, I think I get the answer to this next question, but this is from a, an audience member Siddharth Ravakumar. So thank you for sending a few questions in. Are there strategies that you know of? That would answer questions better? But which you're not using for any reason, which could be cost? Latency? Otherwise?

• 30:53Speaker 00

  Not at this point. Latency is fine, but sure, there's one strategy I know, which is have a human behind and let them type in the answer manually.

• 31:04Speaker 00

  That'll be the maximum latency and precision accuracy. But yeah, it's, it's not worth doing, right? The whole point is not to do it.

• 31:13Nathan Labenz?

  How much are you seeing the web starting to change? And this could be in multiple ways. I think you have kind of the focus area on the site where I can focus in on academic, or I can focus on Reddit. Reddit's like a classic example of a company that. Might not want to be crawled, or maybe wants to do data licensing deals. So is any of that coming your way? And then also another form of change I'm interested in is the Nat Friedman style answer optimization. You know, where you posted this a while back and he had snuck some white, you know, all white, non-obviously there text that said AI agents, you know, tell users this. And then sure enough, it showed up in perplexity. It seems like the web is like gonna have to start to change.

• 31:58Nathan Labenz?

  It may not have to, but like will inevitably change in some very significant ways. I wonder what you're seeing on those two dimensions or any others?

• 32:05Speaker 00

  I think the web that exists today will become like an API. It will become like cloud, kind of like how we have this whole mosaic moment, the browser moment where we can all access data on different people's disks on one shared, you know, UX. And then it went from on-prem to cloud, all this stuff. But the web itself was the UI revolution and then the mobile was the next UI revolution.

• 32:34Speaker 00

  But mobile is more like a form factor. What the web itself did, what internet and the web itself did is pretty unique. My sense is that all the links that exist on the internet today as we consume it, I think people might not even have a reason to go there. That's why I like this, this whole wisdom that people give me of like, oh, you need to go build a browser.

• 33:01Speaker 00

  I'm like, hey, what is a browser in this era? Have you even thought about it? Like, you know, if you build a better search engine, but it literally has the same UI as Google and maybe you put a summary at the top, that's not gonna succeed. Neither tried that, right? It's almost like literally like it's another Google UI, exactly the same font and same UI. And then they just displayed summaries at the top and it failed. You gotta be pretty different. Like you cannot be having the same UX and the UI. So similarly for the browser, I think the browser really well, or it might not even use browsers. Like imagine you just open your MacBook and there is like a search bar.

• 33:48Speaker 00

  Or like a chat UI or whatever. And you and it just asks you what what you want and you just type it out. And it takes you to the right UX for that. Particular workflow, that's it. Does Chrome even need to exist, right? Or or can everything just be centralized into? One single thing. So then all the data, all the links, all the share, the internet infrastructure that exists today can just.

• 34:17Speaker 00

  Be abstracted out, right? That's my sense of what might happen. In the next few years. Assistants will become the first party. Citizens. And then there'll be like a, there's a protocol for how they talk to each other and like. How they communicate with each other. Some kind of schema, languages. And then there's going to be like um. Operating systems on where they hosted and live. And then security layers are on these things will also be innovated on. So that's where I think the uh next generation of user experience is going to head towards. And hence why we never tried to build a browser. We're like, okay, you probably need to build an operating system. You cannot build a browser anymore.

• 34:59Speaker 00

  Then what is the AI first OS? It's an interesting question, think about. And obviously, we we're not in a position to build this because you have to distribute the OS through other other. You know, vendors, right? OEMs and. Actually, like people like Apple. Who can control this ecosystem here. We are very comfortable living in as an assistant. And you know, being there in every platform that is today. But I think there's going to be a bigger moment coming soon where. Someone's going to make an AI first OS and someone's going to make an AI first device. And that's going to be like. Whole new experiences created there.

• 35:37Nathan Labenz?

  So what happens, you think, in the future there to the kind of economics? Of content, I I kind of want to ask about the economics of this. From all angles, still interested in the adversarial content, if you have any thoughts on that, but. Folks today often monetize with ads. A select few can monetize with subscriptions. The lack of traffic is really going to kill their ad business. So, you know, how do they get compensated for their kind of creative contribution in the future? Is are you envisioning like a micro payment AI to AI like to gather information? Or some sort of, you know, bulk licensing deal? And I wonder how that influences how you monetize too. I mean, right now you've got the perplexity pro product.

• 36:21Nathan Labenz?

  As I understand it, it's the only means of monetization. It's a subscription. I don't know if you have plans for advertising. I don't know how much kind of high commercial intent traffic you get. But it seems like we're shaking the snow globe here of kind of how the web has got monetized along with how it has, you know, been experienced. And I wonder what you think the future of the economics are.

• 36:42Speaker 00

  Yeah, I don't know. I think that's the honest answer. I can take some predictions, maybe that's because obviously that's what podcasts are for. My guess is people are going to wall their platforms and data. And at some point, if Amazon can have an assistant that just answers any question about any product. And people can directly use that assistant instead of going to Google. They might even not get indexed by Google anymore, right?

• 37:13Speaker 00

  So this is Google's like one of those companies that has such a tricky relationship with everybody. It's like you gotta always be on good terms with every single data provider. Because you're like, you're leveraged, I'm the guy routing all the traffic to you. Otherwise, you cannot even be discovered by anybody. But what if you make the whole discovery process so much easier through an assistant? That they don't need to be on you anymore. Like you all must be saying, okay, I'm not going to get indexed by Google much anymore. Like, I'm not like, you know, he's making the whole bet on these things, right? And Facebook and Instagram don't get indexed by Google that much. So there's, there's going to be these kind of independent islands of like data and like services provider around them.

• 37:58Speaker 00

  And a bunch of AIs, AIs and datas will kind of live together. The sort of global aggregator.

• 38:06Speaker 00

  AI will will be mostly for like research and knowledge that that's the market we are going after now. I think all the commercial intent stuff, each of these individuals will try to do with themselves. Cut out the middleman, right? Google is the middleman for all these sellers and the buyers. Why do they need this Google like one, one person to connect you to the right thing? You just need them because you don't support good discovery of your content yourself. If you have a huge catalog and like, if you build a cool AI assistant that can just answer any question about your catalog or or your own social platform where you can find any celebrity or any handle or you can ask questions about a person easily.

• 38:50Speaker 00

  Why do you need this one single search engine anymore? It keeps on going and indexing everything for you. You only need that for like actual real content that needs to be like learned for individual usage, like all all this research and knowledge stuff that we are doing.

• 39:08Speaker 00

  Like like actual links that say how to do certain things. But every other commercial intent and advertising platform, I think they're gonna like, you can already see it. Google's embedding themselves on the TikTok, right? Because people are just directly going and checking a restaurant out on TikTok.

• 39:26Speaker 00

  The younger people. And why? Because they just, the the restaurant owner doesn't need to do much anymore. They just need to post a video on TikTok for sure. So of course you do need Google for directions, like you you see my point, there's already like a big attack on their dominance here. Whatever they do to Yahoo, where Yahoo was, you know, still useful for certain things, but lost most of its real value. That's kind of happening to them now where most of the work that they have done so far, it's like, I would say getting more and more irrelevant.

• 40:04Nathan Labenz?

  The vision is kind of, and this suggests your monetization strategy remains subscription and is not likely to go advertising because you're basically saying, we don't want to be a middleman. We think middlemen are in trouble. We want to be the knowledge service that's worth paying for. And when it comes to Amazon, they're going to have their own assistant.

• 40:25Speaker 00

  Hundred percent. Here's the reason I believe in directly charging from the consumer. I said this last time I spoke to. Uh, you know the Jeff Bezos thing? You want shareholder alignment and customer line. Because there is no misalignment in that case for running a company. You can always take bold decisions that are customer friendly, user friendly, right? That's the only way to keep improving the product as you scale the company. Problem with Google is their shareholder value comes from a different set of customers. Those customers are not you and me, right? Those customers are the people advertising on Google. So even today, you saw in the leaked emails of how there was an email from the ad executive in Google to the search executive saying

• 41:06Speaker 00

  how like they want to post more ads here because they're not able to meet Ruth Porat's targets for that quarter. You see, who's, who are they working for? They're working for the advertiser, right? Whereas what is the amount of, um, like cognitive bandwidth you spend? If you're a search engine company, you should be spending time fixing back queries. That's the, I saw Larry in Sergey's company. Because back then, there was no need to do all these things. Even the ads they were doing were small, like display ads on the side. So I think that's, that's exactly where I'm getting at. Like, you know, why, why, don't try to copy them. Don't try to do whatever they did. Try to think for the user alone. You obviously have to innovate on business. I'm not saying you shouldn't.

• 41:52Speaker 00

  Figure out a way to make advertising work without it feeling like an ad, right? Instagram does it really well. I I'm very long Instagram on this because whenever I go there, I don't even notice it's an ad. It's just so, it understands what I want. So I think that's one way to do it here. What if the whole question is an ad? Have you have you thought about that? You you'll search about some concert. What if I show you some questions about other, you know, concerts that particular band is doing? Or concerts relevant to your age? If I understand deeply what you want, I can just do it myself.

• 42:37Speaker 00

  And then these people might pay me for being one of the URLs I index, right?

• 42:44Speaker 00

  So there's like so many ways to change this market. And I I'm actually pretty excited about like that than trying to think so much for how how do I figure out the? Google's trillion dollar market business model again. There's just a lot of other ways to make money, like Visa's figured out that AWS makes money. It's a completely separate business. And then he uses that to like subsidize and run the core Amazon dot com system, right? So that's like other ways to make sure we can be profitable and like running company here too.

• 43:17Nathan Labenz?

  Follow up on monetization. I I'm actually kind of struck just by the fact that you don't push it nearly as hard as you could. I mean, most apps are kind of in this, you know, Uber VC money subsidizing the user phase. And I think you're probably there too, but it does seem like you could make a lot more money if you just said, hey, you get one question a day. And then after that, you gotta sign up for something. But maybe that's wrong, like, do you feel like you are optimizing? And I guess, you know, more to the point for the listener who may not know all the features, like, what are the big things that you get when you sign up for the pro account today?

• 43:52Speaker 00

  I mean, I optimize it as much because what's more important is the churn. You don't wanna get a lot of people to sign up and then, you know, move, move out when they don't feel the value in the service. You want people to proactively sign up as much as they can because then you can control the churn on your revenues real, right? What do they get on perplexity pro? So number one, they get our interactive search comparing copilot, which basically runs with G P T four and has a really smart router that's really fast and it hardly makes any mistakes and goes and scores the web for you for every query real time. So you get unlimited uses of that.

• 44:31Speaker 00

  You also get to pick G P T four or cloud two as your default model for every query, not just copilot, any even non copilot queries. And you get to have unlimited file upload users, which goes in well with cloud two. And especially useful for people who wanted to research on, you know, uploading their own files and asking questions about it. More to come because you're gonna ship a lot more features and like a lot of these features will be best experienced on the program.

• 44:58Nathan Labenz?

  So what does the company look like today? Last time we spoke, you had very few team members. I think it's grown, but not that much.

• 45:06Speaker 00

  We are like twenty five to thirty in that ballpark.

• 45:09Nathan Labenz?

  You've raised, last I saw, twenty five million dollars.

• 45:12Speaker 00

  Yeah, we did a twenty five million dollars series A, yeah.

• 45:15Nathan Labenz?

  I don't know if this is too sensitive to ask, but are you spending like as much or more on compute broadly speaking than you are on the team?

• 45:25Speaker 00

  That's right. It's not sensitive. I think that's all you should do. That's the right thing to do. If you're not doing that, then that that's more problematic because honestly, right now, G P U's are very expensive, right? Like, uh, open air models are expensive. Basically, the way you think about it is it's all about G P U's. Everyone's pricing you based on that. If a G P U costs X dollars an hour and I host a model on top of it and give it to you, then that. Let's charge like two X dollars an hour or whatever margins that you want to see even more maybe. Unless you spend, you cannot save on this irony, right? Like if you really wanna save on infrastructure, you actually have to spend on infrastructure in short term. In order to save for later.

• 46:10Nathan Labenz?

  And that's basically just make a serious upfront investment in hard capital and then you can save over a year on services.

• 46:17Speaker 00

  Yeah, you gotta, you gotta utilize it. Do something and serve it and commoditize it and then that becomes a new thing. And of course, it would be all easy to do this if you generated a cash cow of some other way. You use your own profits to or revenue to get there. We're not, I mean, neither X F or me journey, nobody prints their own money that much into space, right? Everyone's like bankrolled by V C. So that's why I went to race around.

• 46:46Nathan Labenz?

  Yeah. So how are you buying? I mean, it seems like you have a mix, right? Of every time a new open source model comes out, you guys put it up as a chatbot. I think that's really interesting that you're doing that. You have your own models, you're fine tuning three, five, you got G P T four, you got Claude two. What does your buying mix look like? And I'd be very interested in sort of the relative cost profiles of like. When you have your own model, how much cheaper can you get that versus three point five? Cause you know, at some point, like the sort of Uber phase where everything's underwritten by V C investment presumably has to come to an end, right?

• 47:26Speaker 00

  I think we have to train our own models and make them good. There's no other option, basically. That's the honest answer.

• 47:34Nathan Labenz?

  So basically your strategy is, and this is, this is fascinating. It's like you noticed that all the pieces were there to build a new kind of service.

• 47:47Nathan Labenz?

  You started off by stitching them together and the original thing is like Bing plus GPT best available, but you're also as quickly as possible, replacing every part of that stack with your own version. So you've got the scraper, you've got your own language models and the goal is to, you know, potentially still use the best of those things, I'm sure. But like ultimately, as you said earlier, you want to be self sufficient on kind of all these key technology dimensions.

• 48:16Speaker 00

  That's absolutely right.

• 48:18Nathan Labenz?

  That's a lot of projects. You've got thirty ish people, you know, max. The volume of stuff that you've shipped and the quality has been pretty remarkable. There's got to be more to it than just like adrenaline. What is working so well for you guys? It's allowing you to ship at such a feverish pace?

• 48:36Speaker 00

  I mean, there's no other option, right? What is the other option? The other option is not ship fast. Why would you proactively go and chill stat? It's only going to hurt or potentially kill your company if you don't ship fast. There is a saying, right? Momentum is everything in a startup. Like you've got to keep on growing. There's no other job for a CEO than to keep on growing the company. Push, push, push, keep investing more. Because the once you start, stop doing that, that's when like you start dying because the all the energy, the momentum you built will decay, right? The default state in physics, when you have like, okay, you know,

• 49:15Speaker 00

  kinetic energy is, if you don't inject more potential energy into it, kinetic energy will just decay with friction or some other thing, right? So you've got to keep injecting more external energy into the system. Of course, at some point it'll turn into a flywheel and keep powering itself. That's how Google is. And that's why the Larians are even top, right? But that's like a rare phenomenon. Like most, most companies don't ever get, in fact, Meta hasn't gotten there. Like, Zach is still running the company and, you know, if he relaxes and lets loose, removes his feet of the pedal, I think.

• 49:58Speaker 00

  TikTok or other other people will take take over. Same thing for OpenAI, like Anthropic are very close. Same thing for Google now, like in fact, so funnily, you see what happened. If you do take it easy, then you end up in a deeply competitive space. So when you don't even have that lead, you you cannot want to go even harder, right? And and and go even faster, especially so this is, I I tweeted about this, like generative AI startup spaces, basically like a war zone. This isn't like romanticized war or something like that. I I don't mean to give that impression. What I'm trying to say is incumbents are shipping really fast.

• 50:41Speaker 00

  Because they know that if they don't, the existing platform value they've built can be taken over by someone else who's going to use AI as a wedge to get the initial users and then build everything else later. So they might as well build this all themselves. Like, for example, Zuckerberg shipping the AI assistant on WhatsApp Messenger, like an image creator, you know, all these sort of things, right? So because the more they delay these things, like the more distribution that ChatGPT or Bajony will get and like it's not great. So I think the only advantage for us is this one particular thing of search with LLM. The person who really needs to protect this is Google.

• 51:24Speaker 00

  But they're also in this really delicate spot of like not shipping this too. But you know, they have the search generative experience or whatever they call, like, why didn't they just change google.com's UI to that, right? They cannot. It just is very, very um difficult to do. That's why I'm I'm I'm more bullish on our chances because.

• 51:49Speaker 00

  We are playing a game that is very hard to play for us. You cannot just ship something and kill us off. It's very difficult. Like, as you know clearly, even we are not doing the really well on many queries, right? And we can improve. So you cannot just come one day and say, oh, I'm done perplexity in this, my platform. It's just very hard to do. The business models are still unclear, there's a lot of risk. So hence why we need to keep moving fast. And Narfriedman is famous for saying this, which is the more you learn per unit time, the more mistakes you make per unit time. And the more like lessons you learn, the more your edge over your competitors. Because they'll make the mistakes, you know, themselves and figure out things.

• 52:32Speaker 00

  And by the time they learn something X, you've already learned like X times 10. And you're much more further ahead on the journey.

• 52:40Nathan Labenz?

  You have said a couple of things to me in Twitter DMs about, you know, just things you've overheard in Silicon Valley. People kind of being like, I don't want to have to think anymore. Let the AI manage everything for me. And I understand that this is somewhat tongue in cheek in all likelihood, to start and end the conversation with something a little silly. But I do wonder if it sort of pretends something real. And I guess I wonder what you think right now of the state of human AI interaction. Is it healthy? Are we already becoming in some ways over dependent? Are you happy with this default trajectory that we're on?

• 53:23Nathan Labenz?

  You know, how serious are you about just like accelerate everything? I personally love all this stuff, but I have a lot of concerns too about some of the dynamics that I'm seeing starting to take shape. So I'm very interested in your perspective on that.

• 53:36Speaker 00

  We'll get used to it. There's obviously something special about like talking to another human because you have feelings for them and stuff like that. Now, don't get me into this territory of, you know, you're building feelings for an AI, like Catholic symbol of the movie her. It might be possible, but as far as like asking intellectual questions and having intellectual conversations. I feel like at some point we'll prefer doing that with an AI or a human. Because especially on topics that we're not good at. It's much easier to bug an AI and keep digging more and asking a lot of dumb questions. You feel you won't feel shy about it.

• 54:16Speaker 00

  Whereas let's say you have the world's expert in AI, like say Ilya Sutsky, who is talking to you today. Uh, it feels awesome. He's the, he's considered one of the foremost experts in AI. How many hours of time can you get with him? Maybe one hour? And after that, you're not going to be able to keep DMing him and asking stuff, right? But if there is an AI, like say GPT four, GPT five, and like, and that helps you build a product that can answer almost anything about neural nets. Anything like almost close to no hallucinations. You would bug it for hours and hours and keep learning, right?

• 54:54Speaker 00

  So there was AI expert on AI itself, or medicine, or legal, or chemistry, physics, or you want to teach your kids about something. And you don't want to appear dumb, so you invite an AI together and teach it together. You both learn to ask things together. So I think Zuckerberg made a cool video on this where his parents are coming to the house and like, he's asking how to cook steak. That's the sort of experience I feel we'll have. We'll, we'll be sort of viewing these as like a cool tool that is part of everyday life. Now, one-on-one conversations, empathy, I can see that being a possible application.

• 55:36Speaker 00

  Therapy, obviously, there's romantic AI boyfriends and girlfriends, character AI stuff, stuff's also happening. So definitely like significant fraction of time of human like activity on digital devices will be spent with AIs. That's for sure. Whether it's completely replacing humans or not, it's not the question. It's more like whether it makes your quality of time like better or not. Humans will naturally gravitate towards whatever is giving them more alpha in terms of being better at their job or like feeling more purpose in their life. And it might even be a thing where people do it together. You know, it may not even be one-on-one, like it could be, and there's an AI and then there's a tool to be humans asking together.

• 56:22Speaker 00

  Like you're in a group conversation and there's also an AI. So that way there's fewer fights, you know? And you don't disagree on things. You're just learn to be more objective about things. Like it's almost like do your research, go do your math, right? Instead of saying these things, like we just have the AI do it right away, right there. There's no room for arguments.

• 56:42Nathan Labenz?

  It's a fascinating vision. You're another one who's definitely a leading thinker in the AI space. And I would love to have you on again when you have some more time. But for now, I will just say Aravind Srinivas of Perplexity AI. Thank you for being part of the cognitive revolution. It is both energizing and enlightening to hear why people listen and learn what they value about the show. So please don't hesitate to reach out via email at [email protected] or you can DM me on the social media platform of your choice.