View From The Top with Aravind Srinivas, Cofounder and CEO of Perplexity

Aislin Roth 0:00 [on-hold music]

Aislin Roth 0:05 Aravind, welcome to Stanford.

Aravind Srinivas 0:08 Thank you. Thank you for having me. I'm from Berkeley, so hopefully you don't mind.

Aislin Roth 0:11 [laughs]

Speaker 0:12 [laughs]

Aravind Srinivas 0:13 At least wanted to represent with the blue.

Aislin Roth 0:15 [laughs]

Aravind Srinivas 0:16 But yeah, it's great to be here. Thank you.

Aislin Roth 0:18 We're happy to have you here today. Now, many of us in the audience are active Perplexity users.

Aravind Srinivas 0:24 Thank you.

Aislin Roth 0:25 Especially with free Perplexity Pro for all Stanford students.

Speaker 0:29 [cheers] [applause]

Aislin Roth 0:34 So we couldn't be more excited to have you here.

Aravind Srinivas 0:37 Thank you.

Aislin Roth 0:38 Now, to get started, I figured I would turn to Perplexity to help me craft my questions. So let's take a look at what it said.

Aravind Srinivas 0:47 Wow.

Aislin Roth 0:48 So to get started, I figured I would put this prompt into Perplexity, and I asked, "I'm interviewing Aravind for a one-hour interview at Stanford with an audience of business school students. What questions should I ask him?" [laughs] Now-

Aravind Srinivas 1:01 Hopefully they're not, you know, too difficult.

Aislin Roth 1:03 [laughs]

Speaker 1:04 [laughs]

Aravind Srinivas 1:04 Well, you probably missed that in the prompt.

Aislin Roth 1:06 Ah. You-

Speaker 1:07 [laughs]

Aislin Roth 1:07 You might have yourself to blame for that one. [laughs] So Perplexity did give me a very detailed response, but in summary, it suggested that we talk about first, your personal backstory, second, the early days at Perplexity, third, the company today, and fourth, various leadership lessons. How does that sound?

Aravind Srinivas 1:30 Sounds good.

Aislin Roth 1:30 [laughs]

Speaker 1:31 [laughs]

Aislin Roth 1:33 I thought that sounded like a pretty good outline, but I figured we could get just a little more personal. [laughs] So I thought I would test out my follow-up prompting skills and I asked Perplexity, "What is something about Aravind that the audience may not know? What is the funniest thing about him? What are some questions that I can ask him to inject humor into this conversation?" [laughs]

Speaker 1:59 [laughs]

Aislin Roth 2:00 Very direct. [laughs] "What are some rapid-fire questions I can ask you at the end of the interview?" So stay tuned for that. And finally, "Is Perplexity ever wrong?" [laughs]

Aravind Srinivas 2:13 More, more, more than you think. [laughs]

Aislin Roth 2:16 [laughs]

Speaker 2:17 [laughs]

Aislin Roth 2:18 So this generated some more interesting insights than my initial prompt, uh, including, here's the very detailed list, but I'll just hit on a few of them. First, I learned that you love cricket. Second, I discovered that you actually taught yourself to program after missing getting into computer science at IIT Madras by only 0.01 points. And lastly, I learned about your connection to Sundar Pichai, the CEO of Google, who grew up in the same hometown in Chennai,

Aislin Roth 2:49 India. Now, Perplexity actually suggested that I watch this video of you and, you and Sundar from Chennai, so I figured [laughs] that this might be a very good place for us to start. I'll spare you the video, but Aravind, is there something in the water in Chennai that's led to so many successful tech entrepreneurs?

Aravind Srinivas 3:10 Well, I, I think, like, it's interesting, you know, like, um, I... More than Ch- I'm sure, like, a lot of peop- I- I- In India, like, you know, there are so many cities which are producing great people and, um, uh, so one thing that I would say is very common is this sort of culture of, um, really trying to excel and do your best at what, what you're meant to do. Um, and this sort of like real emphasis on education i- is very much present in Chennai and many other cities, of course. But like, uh, i- in, at least in my circles, um, people valued being scholarly and well-read even more than being rich. Like, it w- you, you kind of got respected a lot for it, and I think that translates to like, you know, going above and beyond and, um, not just read what's meant for your exam to do well in your exam, but like really try to go deep into what you're trying to learn. Um, and I think that's sort of like common in many people who come from there and excel in Silicon Valley or other parts of the U.S. Uh, I feel that's like a very common trait. Uh, and obviously, like, this cricket thing is there and, like, one thing Chennai is known for, uh, is they're called the cricket nerds. Like, they're really obsessed about all the statistics. You know, like back before like, uh, Google or like Cricinfo or all these sites existed, uh, we could still like, you know, recite all the stats of every player, uh, and, and like obsess about things like what's the run rate, what's the average. And like you, you learn like basic statistics before it even it's taught formally to you that like it's not just important to score a lot of runs in one game, but like you gotta be pretty consistent. And so, um, I think that's kind of common, I would say, for people from Chennai.

Aislin Roth 5:00 So knowledge over wealth, and attend cricket games-

Aravind Srinivas 5:03 Yeah

Aislin Roth 5:03 ... is what I'll take away from that one. Now, you left India to pursue your PhD in computer science at UC Berkeley.

Aravind Srinivas 5:10 Mm-hmm.

Aislin Roth 5:10 Not Stanford, unfortunately. [laughs]

Speaker 5:12 [laughs]

Aravind Srinivas 5:12 Oh, I, I, I didn't get in here, that's the truth.

Aislin Roth 5:15 [laughs]

Speaker 5:16 [laughs]

Aislin Roth 5:17 Well, we have you here today. We couldn't be luckier. But how have your academic roots shaped your approach to building Perplexity?

Aravind Srinivas 5:24 Yeah. Uh, it's, it's actually like pretty core. Um, you know, Perplexity started off with citations right after every answer. Now, this is obviously like real. I'm not, not just making it up. When I first ca- went to Berkeley and I thought, like, oh, a PhD is this amazing thing. You watch movies of, well, like Steven Hawking, where you just knock on your advisor's door, "This is the idea for my thesis." Like, I, I really romanticized how that's how it's gonna be. But it's not quite as, uh, uh, like, you know, easy as that. Like, you have to like kind of earn your way to like coming up with your first core original idea. So, um, it's, uh, it... Institutions like Berkeley or Stanford, they've kind of like- Amazing because they, uh, give the new student, uh, a, like, sort of like a framework to get there instead of just leaving them, uh, completely unsupervised. Um, and so when I first went to the lab, um, I was asked to kind of help a senior student, uh, by like, you know, work on their idea and, like, try to actually make it really work and write a paper. You learn the process, and then you kinda like get from there. So, um, after I s- wrote, like, a paper or two, um, they told me this concept of citations. And so what matters is not necessarily you getting your paper accepted, it's that other people have to cite and build on it. That's how you build your academic currency in the community. And, um, I was like, "Okay, that's cool," but, like, my paper seems, like, too complicated. So there's always this trade-off between writing very complicated, uh, like very, uh, creative ideas that might get the reviewers to accept your paper, but for someone else to build on it, like, if it's too complicated, nobody cares. So I think, like, you want that sweet spot of, like, very simple ideas that will get cited, but also will get accepted in a conference. I had to learn it the hard way. And once I learned that, I got kind of obsessed about citations and, um, you know, I also learned that that was the core inspiration for the Google search engine of how academic citation graph to, like, marrying that idea to the web hyperlinks. So, um, that sort of also, uh, came into Perplexity because we asked the question, okay, what if, like, an AI always responded like an academic? When you write a paper, every sentence you write in an academic paper needs to have a, uh, corresponding citation to it, or else it's kinda coming across like an opinion. Uh, it needs to be, uh, something that, like a source of truth from some other peer-reviewed paper in the past or some experimental result in your own paper. And we thought, okay, what if we do that for, um, AI, where every answer, every sentence in the answer needs to come from some source in the web that has some amount of domain authority or trust score, and, and if we can, um, uh, bake that into the prompt, i- it's like, um, inbuilt, then that sort of creates a very unique product experience, and that's actually how my academic roots helped me in Perplexity.

Aislin Roth 8:26 Now, I know Perplexity's academic focus feature has been a key driver of adoption here at Stanford, and lets you cite only academic journals-

Aravind Srinivas 8:33 Yeah

Aislin Roth 8:33 ... in your research, so very helpful for all of us who are studying today. Now, at Perplexity, you're very focused on improving access to information, and you're building the world's first answer engine. Why is democratizing access to knowledge so important to you?

Aravind Srinivas 8:50 Because I, I love using it myself. Like, I think, uh, like I said, right, I c- I come from a culture where, uh, being know- knowledgeable was ve- very valued. Uh, there is even this quote from Charlie Munger where the best thing you can do for another human being is to help them learn and know more.

Aislin Roth 9:06 Mm.

Aravind Srinivas 9:06 Uh, and, and it's almost like a moral duty for all of us to seek wisdom and become perpetual learning machines because nothing else can help us, like, keep upgrading ourselves. Like, there's only... Like, you can, you can probably focus on wealth and your net worth. You can try to use that as a metric for, you know, your pro- your progress in life, but at some point, like, it taps out. The, it's the, it stops motivating you if you reach a certain threshold. On the other hand, uh, there is no end to knowledge. Um, that's why in Perplexity, the tagline is "Where knowledge begins," because there's actually truly no end to knowledge. Uh, and, um, you know, you can o- you can only keep getting better. So if there's one metric ultimately that we can converge to, uh, to view ourselves as, like, making progress on that, it's just, uh, understanding the world better. So if that, uh, is so core to the human nature, uh, it's essential that all of us have access to the tools that help us get there in the most accessible way. And ob- we're trying to do our best to do that. And obviously, like, you know, the more premium services are behind paywalls. But, um, as these AI models are getting cheaper and, like, smarter, more efficient, more, like, distilled into smaller versions, um, it's, it's gonna be possible to create a version that's just widely accessible for all and helps them basically a- ask any question they wanted and get an instant answer.

Aislin Roth 10:31 Where knowledge begins.

Aravind Srinivas 10:32 Yeah, exactly.

Aislin Roth 10:32 I love that. Now, I wanna go back to the genesis of Perplexity. To tackle such a big problem, you needed a great team. So what qualities did you look for, and how did you build your initial founding team?

Aravind Srinivas 10:46 Yeah. Uh, I was lucky enough to know, um, one of my cofounders, Denis, uh, uh, during my PhD days. Uh, this is also where a- like the academic background helps you a lot. Like, you, uh, learn people who are, like, very motivated and, uh, deep thinkers. And so we wrote the same paper, like, literally the same idea a day apart. And, uh, so that's how we got to know each other, and he spent some time, um, as a visiting student in my lab. And, um, we used to brainstorm ideas on, like, what we could do together, but nothing really came out of it. Uh, one quality I would say, like, you look for your founding team is, like, obviously people with complementary skills. Uh, you don't wanna be, uh, as good as them in what they excel at. I- ideally, they should be a lot better. And, uh, also, you don't wanna, like, step on their toes when they do that. Uh, and in, in our case, like, it's kinda like Denis and Johnny are, were my, like, core founding team. And, um, I would say Johnny was world number one at competitive programming. He represented United States at the IOI, and, um- Those who are aware of competitive programming, there used to be this guy called Tourist, um, uh, and, and like only one guy has beaten him in the IOI ever, and that was Johnny. Um, so that sort of like prodigy who can like not just write amazing code, but has the problem-solving skills to like, like instantly solve hard problems quickly. And the sort of like AI depth and background and software engineering background that Denis had, uh, combined together with like... Allowed me to like take bold risks and, uh, try to set up this sort of very overarching mission of trying to build a completely new search experience. Otherwise, it's impossible. Like, forget about like trying to do this right. Like, so, um, and then over time like you try to hire more and more people who can, um, bring in new skills. Obviously, none of the three of us have front-end programming skills, so we hired someone like really good at like the full stack experience, and we hired someone really good at writing CUDA kernels. Like, it, it keeps on like, like, uh, becoming an incremental additive as to this multiplicative force. Um, and I think that that sort of, um, effect is necessary when there's a lot of people who are... It's not... Like, I think there's a under- like, like a mental model of like a vector sum of all the people. But I actually think, uh, if you wanna create even like, like a truly great company, uh, there needs to be some kind of this Lollapalooza effect. Like there's, uh, factors that become, um, multiplicative in nature. Uh, and, and one example I can give is like our design team was very well known, too, but that's not-- none of the founding team had that design quality. So we went and h- uh, hired someone specifically for that, where that person actually wanted to build a product like this, but he did not have the AI background or depth. He used to work at Quora, where humans c- came and answered questions. So when we gave him the platform to use this AI to be able to answer questions, like his imagination skills came in and created something that's completely like multiplicative. And, uh, that's kind of how I've generally tried to scale up the team.

Aislin Roth 14:06 Find the people who multip- multiplicate you.

Aravind Srinivas 14:08 Yeah, exactly.

Aislin Roth 14:09 Yeah. So less than a year later, you're in the middle of raising your Series A financing round when you find out that one of your key competitors, OpenAI, has just launched their own search competitor. So when you heard the news, how did you respond, and what gave you the confidence that there was still room for Perplexity?

Aravind Srinivas 14:27 So this was-- You're asking me about my Series A?

Aislin Roth 14:29 The Series A and the-

Aravind Srinivas 14:30 Yeah

Aislin Roth 14:30 ... the news that OpenAI has a search competitor.

Aravind Srinivas 14:32 Yeah, yeah. But actually, like just, just a correction there. I- the-- At that time, OpenAI wasn't launching search, it was actually Microsoft was gonna launch Bing.

Aislin Roth 14:41 Mm-hmm.

Aravind Srinivas 14:41 And, um, I, I... This is ex- almost like a Silicon Valley story, the TV show kind of story where-

Aislin Roth 14:47 [laughs]

Aravind Srinivas 14:47 ... you know, we, we were at the, uh, office of NEA, and like we handshook on a set of terms there. And, um, and then like, you know, I went to Blue Bottle Palo Alto here in U Ave with my-- with Denis, and we're just chilling. "Okay, finally it's done." Uh, and then, um, The Verge publishes a story of Bing releasing on Monday, and screenshots were already leaked because of some A/B test. And, uh, you know, in, in, um, venture fun- funding, there is this period called due diligence, like thirty days. Um, in fact, some... Another VC who had also offered us a term sheet, after seeing that, they increased the diligence from thirty days to forty-five days. And I was like, "Okay, like this doesn't seem quite right." Like, uh, you know, maybe they was trying to back out here.

Aislin Roth 15:36 Mm.

Aravind Srinivas 15:36 And, um, and then the NEA also like calls me on Saturday morning, "Hey, do you have time for a phone call on Saturday morning?" I was like, "Okay, maybe they're just gonna say like, uh, uh, they, they wanna back out." But they actually said, "Look, no, we believe in you. We saw the Mi- Microsoft thing. Uh, don't worry about it. Um, you'll figure out a way. We're..." You know, like, "So we're not gonna back out of the deal, and, um, you keep going."

Aislin Roth 16:03 Mm.

Aravind Srinivas 16:03 So that gave us a lot of confidence, and, uh, I feel, I felt like that was very, very crucial. Because I've heard like lots of stories of how you get term sheets and actually don't get the funding, but they like were true believers.

Aislin Roth 16:16 Yeah. Luckily, your investors had your back. And as the underdog, you must have needed to get creative several times when fundraising, and you've been very successful at it. You've attracted Jeff Bezos from Amazon, Yann LeCun, the godfather of AI, and even NVIDIA. So how did you assemble such a great group of investors, and what war stories can you tell us?

Aravind Srinivas 16:37 Well, yeah, this is funny story. Like, um, Denis was at NYU, so he kind of already, uh, knew Yann. Um, but obviously Yann is a celebrity. It's so hard to reach him. So Yann was on a vacation in, in, in, in, in France for a long time, and we just heard he came back, uh, to New- the NYU campus. So, um, we were already in New York at the time, so we, we just basically camped in front of his office-

Aislin Roth 17:04 [laughs]

Aravind Srinivas 17:05 ... um, for like multiple hours. And he went for lunch, and like he was like, "Yeah, you guys are waiting. Okay, fine, like I'll come back." And then we finally got like half an hour with him, and we built this search over Twitter demo-

Aislin Roth 17:19 Oh

Aravind Srinivas 17:19 ... uh, where a- all we had to do is like let him search over his own tweets and who's replying to him and like, uh, how many followers does he have, the... All those kind of interesting questions everybody has about themselves. And he loved it, and he's like, "Okay, fine, like I wanna invest." Like he just made the decision in like, uh, ten minutes of using the product. Same thing happened with, um, you know, like other, other investors like Karpathy. You know, he's a celebrity here. Uh, he, he, he asked for a deck, and I just sent him the, uh, the link to directly try the product and, uh, and, and same thing with Jeff Dean. All these people were just, um, impressed by just using the product. So- The l- l- the, the main takeaway here is, uh, so you're, you're, you know, there's this thing called a circle of competence. If you're not good at making decks, don't try to, like, do it.

Aislin Roth 18:09 [laughs]

Speaker 18:10 [laughs]

Aravind Srinivas 18:10 Right. But and I w- I wasn't good at it.

Aislin Roth 18:12 [laughs]

Aravind Srinivas 18:13 So instead, just make sure there is a link, like, they can actually use or try instantly, and make sure it works. Because there are some people who do do that, who do that, and, like, the moment you just click on it, it just crashes or it doesn't work. That's not a good experience. But if it works, I think it communicates a lot more than, uh, having a deck. Because number one, most people don't have the time. They're on their phones. They're not, like, on their computers, like, reading every small part of the deck you optimize for. So, um, the other thing is if you're not good at it, like, like me, um, then don't try to do it. And, and by the way, I haven't really done decks much, even for our series A, it was very minimal. Um, series B, no decks. S- C, D, like, I just write memos or d- n- Notion documents. Um, I actually try not to do it because I'm just not good at it. So I, I, I... There was, there's even, like, um, a lot of successful decks from the past, uh, like Airbnb, LinkedIn, Facebook, and you, you see all that, and you're just even more confused how to make one.

Aislin Roth 19:22 [laughs]

Speaker 19:22 [laughs]

Aravind Srinivas 19:22 'Cause they're all so different-

Aislin Roth 19:24 [laughs]

Aravind Srinivas 19:24 ... and you don't know which one to copy or, like-

Speaker 19:26 [laughs]

Aravind Srinivas 19:26 ... uh, how to be original. You know, like, what is... And it, it, it's very confusing, so I just never try to do it.

Aislin Roth 19:33 [laughs]

Speaker 19:34 [laughs]

Aislin Roth 19:35 So play to your strengths instead of being a copycat.

Aravind Srinivas 19:38 Yeah.

Aislin Roth 19:39 It's a, it's a good lesson. You have a lot of prospective entrepreneurs in the audience, and so I think this is a good reminder that it takes grit, determination, and, and hard work, and that that can really pay off.

Aravind Srinivas 19:49 Or, like, you don't have to be really good at many things, like, that people may... Like, if you can be founder, CEO, and not know how to make decks, it's fine.

Aislin Roth 19:57 [laughs]

Speaker 19:57 [laughs]

Aravind Srinivas 19:58 Yeah.

Aislin Roth 19:58 You have a lot of consultants here who all we do is make slide decks.

Speaker 20:01 [laughs]

Aravind Srinivas 20:01 [laughs]

Aislin Roth 20:01 So maybe we have the opposite, the opposite skill set.

Aravind Srinivas 20:04 Yeah.

Aislin Roth 20:04 Be a, a good team if you put the two together. Now, at Perplexity, you're building an answer engine, but you don't own the content, and you don't own the models. So what is your technical moat, and why is the Perplexity approach better than direct vertical integration?

Aravind Srinivas 20:21 She's politely asking me why... You're just a wrapper, so tell us-

Aislin Roth 20:25 [laughs]

Speaker 20:25 [laughs]

Aravind Srinivas 20:25 ... how are you gonna build a-

Aislin Roth 20:27 Those are your words, not mine. [laughs]

Aravind Srinivas 20:29 Yeah. But yeah, like, like, this is kinda like, actually I would be, um... One year ago, like, the whole, um, community was pretty divided on which startups to invest in or, like, which kind of startups to build. Should, should these companies be training their own models, or should they be, like, using APIs? And, um, we had a conviction that number one, models are gonna get increasingly commoditized. Uh, that, uh, and if you do wanna be one of those players that, that build it, like, like, are, are a provider of the models, you need to have an insane amount of funding, and you need to be, uh, a company that is losing billions of dollars a year and it's still fine. And, uh, we didn't wanna be... We, we were not in a position to be, and we didn't wanna be either. So we decided to use other people's models and shape them to be really good for a end-to-end consumer experience of searching and, you know. We felt like there was a lot to do outside the model, uh, there. And, um, I think that bet ended up being right in the sense there are a lot of companies that were trying to build their models who no longer exist. Uh, and I think that, uh, was a clear, uh, proof point that, like, you know, you either raise $10 billion, or you don't do this thing at all. You, you do something else. And, and, and for us, we were, like, working on, uh, giving answers to people.

Aislin Roth 21:55 Mm-hmm.

Aravind Srinivas 21:55 And is, if the answer to this question, for, for giving accurate answers to everybody, do you need to build your own foundation models? Uh, if the answer to that question is an absolute yes, yes, we should, we shouldn't be doing this thing without raising 10 billion. But I felt like if open source makes progress and models keep getting cheaper, uh, the cost of, like, these APIs is going down 2X every four months. So assume that trend continues for another year or two, we're at least gonna ride the wave of a 10 to 100X reduction in the cost for the same intelligence, and the level of intelligence and reasoning is also going up, and open source is keeping a check on these closed source models and bringing the price down. It's a perfect time to be an application, their company using these models and, like, post-training them to be good at summarization, referencing, formatting, custom UIs for, like, so many different verticals, finance, sports, uh, you know, reasoning, all these kind of like charts. There's so many things to do outside the model there, as is we felt like it was just completely worth it to build a differentiated business. And at the end, like, most successful businesses are wrappers of some form, right? Like, uh, before they existed, something else was the more valuable thing, and then, uh, something comes on top. Uh, there's even a thing of, like, Coca-Cola wouldn't have really worked if the refrigeration technology did not exist, right? But Coca-Cola is an extremely valuable direct-to-consumer product. And so you can always c- create something that's some magic formula, the right packaging, uh, that works with the foundational technology, but in the hands of the consumer provides immense value to them that it's totally worth, uh, building. And so th- that's what we wanna be.

Aislin Roth 23:38 What should you build yourself, and when can you leverage things that already exist?

Aravind Srinivas 23:41 Yeah.

Aislin Roth 23:42 It's a great strategy. Now, you've been openly critical of Google's over-reliance on advertising.

Aravind Srinivas 23:47 Mm-hmm.

Aislin Roth 23:48 And yet just this past week, Perplexity announced that it was also introducing advertising-

Aravind Srinivas 23:52 [laughs]

Speaker 23:52 [laughs]

Aislin Roth 23:52 ... for the very first time. So what is your monetization strategy?

Aravind Srinivas 23:57 Yeah.

Aislin Roth 23:57 And how big of a role will ads play in the future?

Aravind Srinivas 24:00 Yeah, so, um, Perplexity's ads are different from Google's ads. Uh, where Google's ad, the problem is the same ad unit- Is, is als- whatever is the answer unit is also the ad unit there. In the sense Google gives you a li- a bunch of links as like, you know, for most queries, and that is also the unit an advertiser can influence by paying. So that way, when you're looking for relevant answers, uh, information, and if the ordering of the links was manipulated with ads, it frustrates you. Uh, if we can avoid the trap and pick an ad unit that's lower margins, lower profits, yet, like, allows us to be true to our users and still make money, it's a reasonably better... Or like we say, I would say a much better sweet spot than what Google went for. And, um, we said, okay, like there's an answer that should be unbiased and truthful to whatever you ask for. But after the answer, there are a bunch of questions, uh, that we suggest you to ask next. You don't have to literally ask that, but at least it influences you on what you wanna ask next. And, um, let's say a sho- like, like having a shopping related query of like, you know, I'm looking for running shoes and this is exactly what I want, and like, uh, these are the brands I like, and it gives you an answer. Uh, the follow-up question that we could suggest there is some, uh, shoe brand that tries to get your attention there. Uh, which could be like, what makes Adidas better than Nike for tennis? Or something like that, right? Uh, that's a que- question they might have picked because your first question was probably, "I'm looking for shoes for playing tennis." And that's a very high intent question compared to just a ad word of a shoe. But we are not like making a particular brand appear one in front of the other on the original answer, but we could still get your attention on the brand as a follow-up question. You can choose to ignore it too. So that is an ad unit that we are experimenting with and, um, we are working with a few brands who are willing to try it out. First of all, uh, the major concern is right now for brands, they're afraid like how the answer can come out to be, because they don't really control the answer. No brand is influencing the answer, they're only able to pick the question. So, uh, it first takes courage for some brands to come and experiment this fra- uh, you know, style. The ROI is not exactly like clear because it's not, uh, necessarily driving a lot of traffic to you. So, um, it's still very early days for us, but what we are very, very clear on is, um, not trying to influence the accuracy and truthfulness of the answer. Because once we do that, then we're gonna end up in the same path as, uh, Google where people are like frustrated with the answer.

Aislin Roth 26:44 So when I type my initial prompt, for example, for this interview, I will never see an ad in response.

Aravind Srinivas 26:50 Exactly. Yeah.

Aislin Roth 26:50 Okay. It's re- reassuring to hear. Now, Perplexity is obviously innovating very quickly, and yet this pace of innovation has attracted some controversy.

Aravind Srinivas 27:00 Mm-hmm.

Aislin Roth 27:00 So News Corp, the parent company of The Wall Street Journal, has sued you for copyright infringement.

Aravind Srinivas 27:05 Mm-hmm.

Aislin Roth 27:06 The New York Times has also issued a cease and desist order-

Aravind Srinivas 27:09 Mm

Aislin Roth 27:09 ... for inappropriate content use. How are you handling these recent challenges and what is your vision for ethical AI development?

Aravind Srinivas 27:17 Yeah. So here, here's what I, uh, we believe and we've, uh, said this on our blog post too. Uh, no one has a copyright or ownership over truth or facts. Uh, this is true in the world of journalism too. If there is an article, for example, right now in our interview, you referenced that The New York Times has sued, uh, Perplexity. Now, that was reported by somebody else.

Aislin Roth 27:45 Mm-hmm.

Aravind Srinivas 27:45 But you're using that in our interview. Now, can someone-

Aislin Roth 27:49 Right

Aravind Srinivas 27:49 ... uh, claim ownership over that and disallow you from saying that particular thing? No. Right?

Aislin Roth 27:54 Mm.

Aravind Srinivas 27:54 So truth is supposed to be distributed widely, right? So the specific expression of truth, the specific way in which something is written, there, that, that has some copyright angle there, and that's actually the core OpenAI New York Times scenario. But what we are doing is we are referencing truth that already exists in these outlets and summarizing and synthesizing it for the user in the context of a search experience. So, uh, people need to differentiate the use of AI that trains on proprietary content versus AIs that just use them as sources and give answers and there's no actual training happening. So, uh, we made that very clear in our response and, um, we've also made it clear that we wanna... Like, we, we can only survive and keep getting better as a product if there is an open and thriving ecosystem of journalism. Because we do need real-time information to be created every single day, and if there's not the right financial incentive for them to do that, then it's not good. So what we did is like, okay, we are gonna make revenue through ads, and we're gonna share that ad revenue with publishers, and that way you, you enter into this publisher program that we came up with, which is not exactly paying you money just to license your data for a certain period of time, and then once we've absorbed it, like we don't want to. Like we don't wanna create that sort of a short-term model. A long-term model is as we scale in usage, as we scale as a business in revenue, we wanna share that revenue with you on a query level basis. So it's very clear. This is more inspired by how Spotify shares revenue. And, um, Fortune, Time, Der Spiegel, have all like signed up to be part of it. WordPress signed up to be part of it, and we are also gonna announce more partners in the coming weeks. So we are very, uh, confident that that program will soon resonate with everybody in the journalism community. And we also made grants to Northwestern University to kind of like do more research on how, uh, tools like ours can help journalists like, you know, write better. Because all journalists do fact checks, and we are an amazing tool for like doing fact checks. So I, I, I'm very confident that this current period of turbulence will go away and, uh, a year or two from now, like we'll have a system that, uh, helps both, both these different set of people to like flourish together.

Aislin Roth 30:18 Mm. And just to follow up on that, so earlier today we talked about Perplexity's academic roots and the importance of citations.

Aravind Srinivas 30:25 Yeah.

Aislin Roth 30:25 So given that, how do you handle these journalist allegations of plagiarism in particular?

Aravind Srinivas 30:31 Yeah. Yeah, so exactly. So that- s- when, when, when you wanna like go deeper into the definition of plagiarism, it's like s- if you don't attribute the source.

Aislin Roth 30:39 Mm-hmm.

Aravind Srinivas 30:39 That's like a very core part of it. Uh, and when you're always attributing the source, it's very hard to say like you're plagiarizing content. Uh, and also like, uh, you're not exactly reproducing things as... Sure, AIs are unreliable at times, and like there are times when they're like o- o- word overlap of like more than three or four words, and you can argue like to what extent that is like, you know, uh, exact reproduction versus trying to synthesize. But what we are trying to say is like, uh, we are trying our level best to summarize, synthesize from diverse set of sources, and make sure to give credit to all the original sources. And that way, like we, um, in our-- to, to our, uh, best possible extent, we can control these AIs. Uh, we are doing our best to make sure that the credit attribution part is clear.

Aislin Roth 31:25 Mm-hmm. I, I like the Spotify analogy. It's just... it's such a rapidly changing landscape-

Aravind Srinivas 31:29 Yeah, exactly.

Aislin Roth 31:30 And-

Aravind Srinivas 31:30 So if you-- if we make an ad revenue where you're a source, uh, we are gonna share that revenue with you.

Aislin Roth 31:35 Hmm.

Aravind Srinivas 31:35 Uh, who never shared ad revenue? Google, because they gave you the traffic, but they made the ad revenue on their platforms. And, uh, the only way for you to monetize the traffic that you get is to put pop-ups and ads on your site-

Aislin Roth 31:51 Mm-hmm

Aravind Srinivas 31:52 ... through another product of theirs called AdSense. And so that is what frustrates many users to come directly read on, uh, many journalist sites, because there's a lot of ads on the site and they have to like, you know, close a lot of pop-ups. And so it's... the this system of like just referring traffic and m- making you monetize with more ads is not sustainable.

Aislin Roth 32:13 Mm-hmm.

Aravind Srinivas 32:13 Right? Like we... you need to create something that the user really wants. And, and we are also offering our APIs so that they can build AI native products and chatbots on their, their websites. So if people wanna just come there and ask questions about only articles that they have written, we are offering our APIs for free to, to these people. And, uh, we're also offering our tools for free to all the people who work at a particular out- journalist outlet. So that way, like we, we, we can create a system that, uh, is economically, uh, pretty lucrative for them.

Aislin Roth 32:44 Mm-hmm. That's... it's an interesting future ahead. So if we take a step back now and we think about the biggest technology companies of all time, in almost every case, these are category-creating companies, Uber, Facebook-

Aravind Srinivas 32:57 Mm-hmm

Aislin Roth 32:57 ... Airbnb, Salesforce. So a decade from now, if we look back on this moment in time, what is the history-defining company that you are building?

Aravind Srinivas 33:09 I would say like if we can help people, uh, get answers to all their questions and get help for all their tasks, uh, we will be like a... we, we, we'll be in that league for sure. And, uh, we are pr- getting pretty close to being a reliable answer machine. I know you asked the question, is Perplexity ever wrong, and I, I'm telling you, like there are a lot of mistakes we still make-

Aislin Roth 33:32 [chuckles]

Aravind Srinivas 33:32 ... on a daily basis. But, uh, zoom out and think like if models keep getting better and our coverage of the web is getting better, uh, the mistakes are gonna be like whatever is like one in a 10 or one in 100 is gonna, you know, be reduced to like one in 1,000, one in 10,000. That order of magnitude improvement's gonna come. So if we are a reliable answer machine to everybody, and widely accessible, and not just give you answers, but help you accomplish tasks too, make transactions, buy things, book things, uh, book flights, get the best deals, um, and like, you know, make your life more productive, uh, give you, give, give you back all more time, I think we are gonna be a pretty industry-defining product and a company.

Aislin Roth 34:15 It's exciting. We hope to see you there a few years from now. Now, there's so much more we could talk about when it comes to Perplexity, but I wanted to save a few minutes to talk about you and your leadership style. So in just two years, you've progressed from a scrappy founder to CEO to the leader of a $9 billion AI company.

Aravind Srinivas 34:33 Mm-hmm.

Aislin Roth 34:34 What has your leadership journey looked like across all of these different stages?

Aravind Srinivas 34:39 Yeah. I, I've tried my best to keep, keep upgrading, um, and I'm still not, you know, the most seasoned, um, polished CEO, but I would say there's an extreme bias for action that, uh, I try to like bring in and try to like encourage everybody else in the company to adopt, and I think that's what's helping us continue to be fast, even when you've gotten it to about 100 people. A founder that I really admire told me, "Once you get to 100 people, you're gonna mo- you're guaranteed to move slow." And I was like very determined to prove him wrong.

Aislin Roth 35:15 Mm-hmm.

Aravind Srinivas 35:15 So, so far so good, you know. But at, at some point, like definitely we're gonna hit the, uh, problems of scale and like how to move fast. So I'm determined to solve that problem, and if whatever final solution I come up with, I hope it's helpful for other people, too. And the other thing I would say is giving people who haven't necessarily, uh, become experts at one thing the opportunity to like go do something they're not yet proven for, uh, is something I've done a lot. Like, uh, you don't have to hire the former head of growth at Instagram to be like the head of growth or head of product at Perplexity. That's a trap that a lot of people fall into, is like, oh, if I want the best, uh, person for, you know, doing A/B tests-

Aislin Roth 36:06 Mm-hmm

Aravind Srinivas 36:07 ... I'm gonna get the person who did it at the previous best consumer company and hire them here. I have not fallen into this trap. I've actually tried to hire people with some chips on their shoulders who are like very talented, but they have not had their first major hit yet. Uh, chips on shoulders put chips in your pockets.

Aislin Roth 36:25 [laughs]

Aravind Srinivas 36:25 So not my original quote, so I, I don't remember who said this, but-

Aislin Roth 36:28 [laughs]

Aravind Srinivas 36:29 ... it's, uh, it's, it's pretty cool. So I... That, that's something that I wish more people did, the sort of like, um, experimentation, like putting someone in the waters and letting them s- figure out how to swim rather than hiring the most, uh, you know, well-known expert at that topic. Main reason is that, like, uh, most people are unable to push themselves for the second success in general. Uh, that I've l- I've seen that, and, um, I think, I think, like, it's very hard to be extremely motivated to do like, uh, grueling hours when you've already had like a big success in your life. So, um, that's one way I've tried to be different, and bias for action, trying to do things on my own, uh, to understand like what, what it is. And I use the product quite a lot myself. Pretty much like at least 10 queries a day is my average, but there are some users who do it more than me, so I'm very happy. Um, and, um, I think like that helps me to s- make the right decisions. If at some point you stop using your own product, uh, that the company's building, uh, you... it's very easy to lose touch with reality, and you're just like making decisions based on what other people tell you. And, um, it's very essential that you're as close as possible to the source of truth.

Aislin Roth 37:50 Hmm.

Aravind Srinivas 37:51 So when people complain on social platforms like Twitter, "Oh, this thing's not working, that thing's not working," I love doing customer support. Um, I think if... We have people who do customer support too, I'm not trying to say they're not needed, but it really helps you to like, um, um, understand what the customer frustration is, the user frustration is, and be, be that sort of a user yourself. Like complain about your own product to your engineers, your product managers. Say like, "This should be fast, this should..." You know? You don't have to like, uh, s- uh, sort of like just be doing whiteboarding and strategy all the time. You can actually just sit for hours and hours using your own product, and you can make better decisions.

Aislin Roth 38:34 How we stay scrappy at scale.

Aravind Srinivas 38:36 Yeah.

Aislin Roth 38:36 It's, it's a great one to, to stick with. So before we open it up to audience Q&A, I have one final question that we're asking all of our speakers this year at View From The Top. Our theme is leaving your mark. So Aravind, how would you like to be remembered?

Aravind Srinivas 38:53 I would love for Perplex... I mean myself and Perplexity to be known as helping the world, like, makes, h- helping make the world smarter. If people who use Perplexity feel smarter after using it because they learned something new, uh, they slept wiser than when they go to bed and they woke up, I, I would feel really, really glad if we accomplished that. Uh, because I think, I think that's not easy. Most products, most consumer products end up like wasting people's times. Um, like our, you know, there are obviously not gonna... I'm not gonna mention which, but there are some products I, I am addicted to. I use it a lot, but I don't feel good at the end, like I've wasted so many hours.

Aislin Roth 39:34 [laughs]

Aravind Srinivas 39:36 Um, and, um-

Aislin Roth 39:36 [laughs]

Aravind Srinivas 39:37 ... Perplexity is not that, at least like I, I don't think it is that sort of a product that ma- you know. Even the discover feed that we have, people tell me that they learn something when they scroll through it.

Aislin Roth 39:49 Hmm.

Aravind Srinivas 39:49 Um, and I want that to be, continue to be the case. And, um, I also want us to help people do things. Like, not a lot of people can afford to have assistants, executive assistants, personal assistants. And, um, I remember in 2018 when I was an intern at OpenAI, uh, Sam Altman did this fireside chat with Bill Gates, and he asked Bill Gates like, "What do you think will the world would look like when there is AGI?" And the answer Gates gave was very interesting. He said, "It would, it would basically, it would basically be like living my life." You know?

Aislin Roth 40:27 [laughs]

Aravind Srinivas 40:29 [laughs] L- live like a billionaire, where if I wanna know about a topic, if I wanna learn about a topic, I don't have to read any book. I can have people read it for me and g- prepare a report for me, and even make a presentation for me. I- if I have to get somewhere, I have a jet. People take care of all the travel planning, meals. You know, if I wanna like work out, everything is like done. I want the best nutritionist. I know what to eat. I don't have to think. Life is so easy. It's like running life on cheat code.

Aislin Roth 41:00 [laughs]

Aravind Srinivas 41:02 Now, I think that sort of a life can be made more and more accessible to most people if AIs can do stuff for you, truly understand you, you know, help you plan stuff, help you book stuff. Like all the mundane work that you have to do with on the web, if a tool can sort of increasingly get better and better at doing it, I feel like your life will be like a, a billionaire, right? And then the meaning of the word billionaire also sort of like loses significance over time. So, um, I think w- if we can be one such tool, I'm not saying we wanna be the only tool, but if we can be the w- one such tool that helps people do that, I feel like I would have made a good mark in the world.

Aislin Roth 41:42 If Perplexity can give me Bill Gates' life, I will be very happy one day. [laughs]

Aravind Srinivas 41:46 [laughs]

Aislin Roth 41:47 Now we'd like to open it up to audience Q&A. So if you have a question, please raise your hand and one of our View From The Top mic runners will come find you. If you're selected, stand up, state your question and year, and ask your question.

Speaker 42:03 Aravind, um, firstly, thank you. I think you're one of the young leaders we all look up to, given where we are today. Um, I'm a fan of the book Atomic Habits. So in your journey from PhD student to Perplexity CEO, what is the one habit you had to, or daily habit you had to let go of, and maybe a new one you learned that helped you become the leader you are today?

Aravind Srinivas 42:26 So the one habit I let go of, uh, I'm not sure if I had to, but I definitely let go of this, and I feel like it helped me, is, uh, waking up late was just something I stopped doing. So [laughing] uh, yeah, I think it f- makes me feel like I get more hours in the day. Um, and so that also means going to bed early. Like early to bed, early to rise, you know? The, the... I, I feel like it helped me. It's a very, um... So it's, it's been like probably at least two to maybe three to four years since I've woken up later than 8:00 AM in the morning. And it doesn't matter which city or where, like I've always done this. What was the other, other question?

Speaker 43:08 A new one you learned that, that you had to pick up.

Aravind Srinivas 43:11 Uh, I think like I... Yeah, I've at least trying to get better at like getting, you know, three days a week work- workouts.

Speaker 43:21 [laughs]

Aravind Srinivas 43:22 And I, I never used to work out much before, so...

Speaker 43:25 [laughing] Thank you.

Speaker 43:29 Uh, thank you for being here, Aravind.

Aravind Srinivas 43:31 Thank you.

Speaker 43:31 Uh, my name is Shravan. I'm a second year MBA student. Went to IIT Bombay for my dual degree in electrical engineering. Uh, I think you answered the technical mode question really well, but would like to hear from you what do you think is the biggest risk or challenge that you feel the company is facing?

Aravind Srinivas 43:45 Sorry, sorry, could you repeat that?

Speaker 43:47 Uh, what do you think is the biggest risk or challenge that you feel the company is facing today?

Aravind Srinivas 43:51 Yeah. Um, I, I think it's the same thing I said earlier, which is, you know, um, I think every startup that's tried to scale, uh, somehow ended up moving slower once they got like several hundreds of people, a thousand people. It's very difficult to do... Somehow it gets difficult to do things. Like you expect progress to be at least linear, and the number of people you can do more projects. But it v- it gets very difficult to simultaneously execute well without some drop in quality. And when there is a drop in quality, like users notice that and they think like, you know, you're, you're regressed, your, your product has gotten worse. There's this whole phrase called enshittification, where when you're trying to scale and like scale your business, scale your users, uh, product gets worse in quality for the initial loyalists who loved it for the quality. So I, I s- um, that's in my opinion, the biggest challenge for us.

Speaker 44:51 Hey. I've got a follow-up to Aislin's question earlier about the ethical issues that Perpec- Perplexity has been facing. I know that you personally have experienced a lot of criticism surrounding some of those issues, and I imagine that as a leader in those spaces, you spend a lot of time thinking about that. I'm curious how you've approached ethical issues, whose opinions you seek out to inform yourself as you step into bigger and bigger roles, and if there's an example where you've seen yourself change your position on something really important.

Speaker 45:19 What's your name again?

Speaker 45:20 I'm sorry. I'm Lauren. I'm an MBA 1.

Aravind Srinivas 45:23 Oh, thank you. So I think like, uh, we have a good set of people in our company. Uh, i- actually to, for the publisher program itself, uh, it's, it's the brainchild of like not, uh, our chief business officer, Dmitry. Um, and um, so clearly like, you know, one thing that I've learned is, you know, I, just because I'm the CEO doesn't mean I have to be the solu- the one who solves every problem. Uh, if someone's better than me at doing something, you sh- you should trust their instincts there. Uh, one thing I did kn- believe in is like, it's... There's a lot you can do by just engaging, uh, and, and like trying to educate people on the other side of what you're trying to do. Uh, I'll give you an instance, like Forbes obviously was, um, pretty unhappy with some of our attempts to like, you know, do pages and things like that. Um, but when I actually met the person who criticized me on Twitter and explained to him what we are doing, he, uh, he at least like shook hands and said, "Oh, I never understood. This is exactly what you're trying to do." And, um, so I think like there's more work to be done there. I haven't met everybody in the community, uh, who's talking about us. And I think just trying to like... There's even something, right, when there is even a sign of war or like it's foggy around there, it feels like something might bad might happen, the first thing you gotta do is, uh, make a phone call and, and, and talk. So that's what I intend to do.

Speaker 47:06 Hey, Aravind.

Aravind Srinivas 47:07 Hey.

Speaker 47:07 Thank you so much for coming here. Um, I'm Varun, M- MBA 1 student, my computer science from IIT Kanpur. You're a big inspiration.

Aravind Srinivas 47:13 Thank you.

Speaker 47:13 Um, I really resonated with the point where you said that you're trying to help people find the answers they're looking for, but then also when you talked about introducing ads and suggestions where you, you suggest like, uh, why do you think Adi- Adidas is a better shoe for sports than Nike or something. Don't you think that affects the kind of information or like the view that you're presenting the users? Like, that might not be the answer they're looking for, so don't you think that it's-

Aravind Srinivas 47:40 Yes. It's, it's a question though, it's not, it's not an answer. Um, it's just a suggested question. But if you do decide to engage with that question, the answer to that question is still unbiased. It's, it's not something, um, Adidas is influencing us to write relative to the other brand. And, and, and the same way. So, uh, at some point we would really understand you quite deeply enough that, uh, even those questions are quite personalized to you and not a generic, uh, sponsored question that the brand picks. And there's been evidence that like, you know, for example, uh, a lot of people, uh, feel like the Instagram ads are pretty relevant to them. Uh, and, and a lot of purchases happen as a result of that. So I feel like relevance is the true answer to making ads work, and making sure that it's not interfering with the core value of the product, which is any question you ask, if the answer to that is uninfluenced by ads, then it's always gonna be truthful. The product serves its value to you. But if we want to bring such a quality product which is almost never wrong, so you can trust what it says at scale, the company does need to figure out, like, uh, some intelligent sweet spots of monetization too.

Speaker 48:58 All right. Thank you so much.

Aravind Srinivas 49:00 Yeah.

Aislin Roth 49:00 I think we have time for just one more question.

Speaker 49:02 Hi, Aravind. I think this might be a good one to end with. By the way, I'm, I'm Abrar. I'm a MBA one. So as the Cofounder of Perplexity, what is the question that you find the most perplexing?

Aislin Roth 49:14 [laughs]

Speaker 49:16 [laughs]

Aravind Srinivas 49:16 This one.

Speaker 49:17 [laughs]

Aislin Roth 49:22 [laughs]

Speaker 49:31 [laughs]

Aislin Roth 49:32 [clapping] It's, it's a great answer.

Speaker 49:34 [laughs]

Aislin Roth 49:35 I'm glad, I'm glad you stopped there. We do have one final tradition here at View From The Top, where I ask you a set of rapid-fire questions, and you respond with the first thing that comes to mind.

Aravind Srinivas 49:45 Okay.

Aislin Roth 49:45 For this one, I used Perplexity to generate all of these questions. Should be easy, right?

Aravind Srinivas 49:51 Hopefully.

Aislin Roth 49:51 [laughs]

Speaker 49:52 [laughs]

Aislin Roth 49:54 If you weren't the CEO of Perplexity, what would you be doing right now?

Aravind Srinivas 49:59 I'd probably be doing research. That's what I was doing before, so just AI research.

Aislin Roth 50:04 You're a cricket enthusiast. What is your all-time favorite cricket moment?

Aravind Srinivas 50:10 When India won the World Cup in 2011.

Aislin Roth 50:12 [clapping]

Speaker 50:13 [clapping] [laughs]

Aislin Roth 50:17 If you could have dinner with any tech visionary, dead or alive, who would it be?

Aravind Srinivas 50:22 Larry Page.

Aislin Roth 50:24 Hmm.

Aravind Srinivas 50:25 I mean, I'm not, I'm not saying this. Actually, Steve Jobs or Larry Page would be my, my picks.

Aislin Roth 50:30 Bring them together, dinner for two [laughs]

Speaker 50:33 [laughs]

Aislin Roth 50:34 And finally, what is the strangest search query you've seen on Perplexity?

Speaker 50:38 [laughs]

Aravind Srinivas 50:40 So, uh, we released shopping on Monday, and, uh, we're just looking at some of the orders, and someone basically bought this, uh, face mask that only has an opening for the eyes and nothing else.

Speaker 50:52 [laughs]

Aravind Srinivas 50:53 And so either they're looking at it in the context of skiing somewhere or, like, biking somewhere that's really cold, uh, or they're conducting a heist.

Speaker 51:02 [laughs]

Aislin Roth 51:03 [laughs]

Aravind Srinivas 51:04 And, uh, we weren't sure which, so we actually went and looked at the query, and the query was like, "Okay, I'm, I'm really trying to bike in this weather in the coming months, and I need something that can, like, cover my face, keep me warm, let me breathe, but only has an opening for the eyes." That, that was the intent level of the query, and then we got, we got them some pretty good answer that they just bought a product right from there.

Speaker 51:26 [laughs]

Aislin Roth 51:27 [laughs] Well, let's hope that the use case was the former and not, not the latter.

Speaker 51:32 [laughs]

Aislin Roth 51:33 That, that's a great place for us to wrap up. So thank you so much, Aravind-

Aravind Srinivas 51:36 Thank you

Aislin Roth 51:36 ... for being here. It's been a pleasure.

Speaker 51:37 [clapping]

Aravind Srinivas 51:39 Thank you.

Aislin Roth 51:41 Great stuff. Thank you. [outro music]