#216 Perplexity & the Future of AI with Denis Yarats, Co-Founder and CTO at Perplexity AI

Hi, I'm Maggie, VP of Curriculum at datacamp, and I know firsthand how overwhelming it can be to keep up with the rapid changes in data and AI. But staying ahead doesn't have to be hard. That's why our curriculum team has curated the best learning content to help you and your team stay ahead, covering everything from data and AI literacy for busy executives to cutting edge generative AI for developers, and everything in between. Whether you're upskilling yourself or your team, we make learning practical, engaging and flexible. It just works. Power your data and AI transformation journey on datacamp.com.

Google search engine is like the most sophisticated system humanity ever built is just like insane. But on the other hand, I think there is like certain things that can be done better. Specifically you can save a lot of time when you don't have to like sift through 10 links and kind of like do a lot of manual work yourselves. Like if you just have a question, you just want to get an answer and then you can imagine in the future this is going to evolve into much more sophisticated workflows and pipelines where you can like going to give those things tasks rather than just like simple questions and they're going to do like work for you.

Hello everyone, I'm Adal, Data evangelist and educator at datacamp. And if you're new here, Data Framed is a weekly podcast in which we explore how individuals and organizations can succeed with data and AI. Arguably, one of the verticals that is both at the same time most ripe for disruption by AI and the hardest to disrupt is search. We've seen many attempts at reimagining search using AI, and many are trying to usurp Google from its throne as the top search engine on the planet. But I think no one is laying the case better for AI assisted search than Perplexity. AI Perplexity doesn't need an introduction. It is an AI powered search engine that lets you get the information you need as fast as possible. This is why I'm so excited to be speaking with Denis Jarretz. Denis is the co founder and CTO at Perplexity. In his previous roles he was a research scientist at Facebook AI Research and a machine learning engineer at Quora. Throughout the episode we spoke about the process of building Perplexity from the ground up, how Perplexity ensures veracity in citations for all its search results, how he sees the future of search evolving with AI, his views on how LLMs will evolve in the years to come, and a lot more. If you enjoyed this episode and the Data Frame podcast. Make sure to rate it wherever you get your podcasts. And now onto today's episode. Dennis Yarads, it's great to have you on the show.

Hey, thanks for having me. I'm excited to be here.

Awesome. So you are the Chief Technology Officer@Perplexity AI, one of the fastest growing AI startups today. So something I want to know about here. I think Perplexity was first launched in August 2022, if I'm not mistaken. How has it been like being on this rocket ship over the past couple of years?

It's been definitely very exciting. I wouldn't say it's work, it's more like lifestyle, but it takes a lot of time. That's why I say it's like a lifestyle, but it's very exciting. So it's like every day you wake up and there's something new. Super cool. I haven't had this experience before, so I'm very excited about it.

Yeah, it must be very interesting being at the center of a really big moment in the technology space. Right. And I think we can jump right into that. I think I want to first talk about what makes Perplexity different from many of the other platforms we see today. And what I really appreciate about using Perplexity, I'm a Perplexity user myself, is that you can see the thoughtfulness put in into making sure that any answer that is provided is truthful. There are citations, it's not biased. So maybe as someone who's overseeing the technology aspect of things and Perplexity, what goes into making sure every answer I get when I type in a query on Perplexity is as truthful and as relevant as possible.

Yeah. So we kind of, Arvind and I, we come from academic backgrounds, so I think sort of like citing when we write papers, like citing stuff is very natural things for us. And when we started, we very quickly realized in this like LLM world, this is going to be very essential component because, and especially in the time we live in right now, there's like a lot of misinformation and very hard to trust what you read. So that's why we. This is probably one of the most fundamental aspects of our product and something we focus a lot on. In terms of like technologies, it's definitely very hard problem to solve. I think it requires like a lot of coordination across like multiple teams, multiple technologies and such. But more than anything, I think it's. It has to be like a mindset of the entire company. It's just basically if there's only like few things we can focus on. This is going to be the, the very first thing that we're going to be focused on. Like speed and accuracy I think is essential. And from the get go we organized ourselves like we decided that we're not going to be too spread around so we're just going to focus on the few things but we'll try to do as them as best as we can. Turns out as we like learn more about this problem. This problem is like very challenging and there's like a lot of things to do. In terms of like technology I think it's basically combination of having very good models, having very good ranking system like search engine and service and then like trying to make sure that the information that you retrieve from the web pages is if there's like multiple sources, maybe like tell them like different things. So you have to figure out like who contradicts what. So maybe if it's hard to resolve it then you can give like several opinions about this. So you don't want to be biased, you definitely want to make sure that you cover all the ground and then yeah like essentially have the models that can like not only generate the answer but also like self verify and see they like made a mistake or not and if they made a mistake then kind of like see how things can be improved. And the most important part is just like to establish datafly will and learn from those mistakes and then keep getting better.

And you alluded to this earlier when you were talking about kind of the principles that you guys wanted to focus on when building complexity. I want to keep dive into that a bit more since the advent of ChatGPT. I think kind of ChatGPT really mainstreamed chatbots or AI assisted search engines in a lot of ways. There's been quite a few players in the space. I think not a lot of people have done as good as you guys have. And one building that flywheel, getting so many users, being able to put yourself on the map as a main player in the space. When you think about kind of the perplexity culture, kind of the things that you focused on, what are the main principles that you've been focusing on to make sure that you're building a differentiated product that works versus the rest of the space?

I think there's like several key factors in terms of product itself. I think we wanted to make sure from the beginning that it's very simple, it's very easy to use, it's very intuitive. That's why we also realized like from the beginning Just like chat interface is not what we want to have and that we can spend a bunch of time thinking what it needs to be from this point of view, even from that time we decided this is going to be one of the differentiating factor. The other one was simplicity and quality. It's like the problem is just that like in general like problem surgent, like you know, it's AI complete, right? And like Google is obviously by far like the best search engine so far, but still, you know, there's certain things that it cannot do. And because this problem is so monumental, we just decided okay, so let's just try to solve this like small piece but try to do it as, as best as possible and, and see how long it's going to take and like where it's going to take us. And yeah, it turns out it's actually a much harder problem than, than we even initially anticipated. And preserving still that mindset that let's just do the simplest things but do them right and high quality and just keep doing that. And I think instead of trying to maybe go after like many things at the same time and then like doing them maybe like not as good. And because of that we decided what kind of like team we need to assemble. We definitely need to have people who care deeply about those things. So like care a lot about a very good infrastructure moving fast, building things with the high quality and like things that work fast. Because essentially I think Google told everybody that you have to get instant answers, you cannot wait. And especially we launched enterprise if you want to save people time, we not only have to provide them very high quality answers, but also do this very fast. And so doing those things together it's, it's very challenging. So essentially I think doing small things but go very deep there and very high quality is just what we trying to do. And I think we're going to continue to do that.

And you mentioned here doing small things but doing them quickly, doing them high quality. And I think kind of the culture of fast iteration and operational excellence and perplexity even bleeds out when looking at perplexity as an outsider myself and looking at the velocity of features being added, the velocity of new releases, new launches, you mentioned enterprise. Maybe walk us through that culture of excellence a bit more and depth and maybe what can you attribute as the key success factor behind this culture of excellence when you're building out the team you're hiring? Walk me through that. As a leader of the organization?

I think it was very important from the beginning is just get like very Strong foundational core team. Arian and I started that we were so happy to have Johnny, whom I used to work before, and he's this legendary coder. He was like a world champion. So this guy is just does everything very fast and very high quality. And so we kind of like decided, okay, so let's just try hire slowly but try to focus on very good people. And from the beginning, I think like until maybe when we hired 10 or 15 people, we were actually had this trial period. So it wasn't like a normal interview. We would like invite somebody, they would work with us for a week or like sometimes even longer. One thing is to do interviews. Like you can miss certain things, but like when you work with a person for like several days, it's like very clear and that even though it was like a time consuming, I think it turns out to be a great idea because we were definitely were able to get those first 10 people who are very trustable, they're aligned on the mission very fast, very strong. And the reasoning behind that was just okay, so those 10 people, that each of them is going to bring like 10 more people people. And so like you need to have a solid foundation. And that's like long term basically paid off because those people now like the new people coming, this is like already integrated in Alcara. We have to move fast. If you can do something today, you have to do it today rather than doing it tomorrow, like next week and sort of have this insane pace and momentum is just something we basically. This is like one of the things that I like right now spend a lot of time maintaining is just how do we, as we get bigger, how do we don't lose this momentum? Because it's a nature like if you stop moving, things slowing down, you have to start, you have to continue pushing. In my mind, the most successful companies is just those who were able to prolong this lose of velocity as long as possible because eventually it's going to happen. Once you become like a super big company, it's just like organizational things become very hard. But yeah, we're trying to sort of like push it as far as possible.

Yeah, it's really interesting and I want to kind of expand here on like, because I definitely agree with you wholeheartedly, like having a relentless space is absolutely essential when it comes to building a company successful as perplexity. And you, you mentioned here bringing in people on the first week and having a trial period. Right. Well, what is an evaluation of success after a week? Is it like time to value and Driving iteration on the product within the first week. Is this something you hope for on the first week?

The few things we paid attention to is first of all like alignment and mission dreaming. So can these people, you know, I figure out like ambiguity. So you just give them like high level goal or task and you see if they can execute on it. But honestly it's like surprisingly very easy to. When you work with somebody, you can like literally sit next to them and see how they write code. Or like when you, when you talk to them like you go for lunch and you just talk to them. It's like it actually does not take a lot of time to understand if you want to work with this person for a long time or not, if they're like capable or not. Like the best people. You literally best people, I can tell like in 30 minutes. So this person is very good. So we should get. And it's kind of like the same, same story like true on the other end. It's also like relatively quickly you can say that maybe it's not the right person for whatever reason, but usually the way it worked because we were like, at the time we were like, okay, so a very small company didn't have like anything. So so maybe first day you spend kind of like understanding if you want to hire them, but like the rest couple of days you're trying to sell them if you found somebody good that you want to hire. But obviously those people have a lot of options. And so you're like trying to figure out like how do we convince them to join us. So that was also like a big part of that process. So not only sort of like assess people, but also try to sell them in join us.

Yeah. And on that selling aspect, there's not a lot of people on this planet who can build large language models or build systems that operate at the scale of the perplexity. So how have you seen our effective models for competing with talent when you're competing with the metas and with the OpenAI's and with the Googles of the world, what is the differentiating factor for joining Perplexity when you have this conversation?

Yeah, I mean it's definitely very tough. I think like for whatever reason those companies like more established, they have bigger clusters, they have compute Compute. Right. So it's. And to be honest, they have way more resources that we have. So they can literally pay as much as they want, which is something we cannot afford. So the strategy is trying to discover people who are maybe not already like established, like research scientists at Fear Right, okay. So yeah, probably if you try to get somebody like that, they're like going to be very expensive. They already have some ideas of what they want to do. Like maybe they're not going to be very aligned. Try to discover these people who have a lot of potential. Yes, maybe they'll require a little bit of teaching and learning, but who can very quickly, within a couple of months ramp up and actually be very productive. And so that super well for us. And so maybe like getting people from adjacent fields who like maybe like strong engineer or have very strong like math, math background or like competitive coding experience and then just teach them like I mean I and like a few other people on the team who used to be like research scientists as well. So we have experience these models and so we can teach them, but they have like other skills and the most importantly they have like desire to learn and ramp up in that field. So which is honestly very important. And that's how we've been doing things. Obviously at some point you also need to get like experienced people. I think now we can actually, because we got like a little bit bigger, like have more resources, a little bit more recognizable brand. So then we can try to get a little bit more senior people, more experienced. But yeah, I think it's still very tough to compete with, you know, OpenAI and places like that because they just attract to all the best people.

Speaking of talent here, when we were talking about technical talent, something you alluded to is taking in strong engineering folks and trying to give them AI chops. Do you find that that has been a successful model in being able to immediately upskill people on AI and get them up to speed on the space and maybe I'll reframe. The question is you value more research skills or technical engineering skills at the beginning when you're looking for an early hire?

Yeah, I think it's basically from my experience and from working with many, you know, famous researchers in AI, I think like usually best people, they're both good researchers and engineers. It's very rarely where you have a strong research scientist who doesn't know like how to code. And so that's why I feel like the best people, they can do both very successfully and then they can do engineering very good. But then they can also do like some researches. It's more about just in general, like intellectual, especially with Transformers. Yeah, there's lots of secrets, but you can kind of teach people how to do this. And in the end a lot of it is still like very hardcore engineering, especially once you start Training large models. It's a lot about distributed system and stuff like that. Obviously you need to know some fundamentals of machine learning as well. But yeah, those things is not rocket science.

Definitely a lot of people are tuning in right now who are aspiring AI engineers or want to work in AI. What are the technical skills that you recommend people to develop at the beginning and what are the cultural natural traits that you want them to have if they want to break into the field?

Yeah, yeah, I think like basically you definitely have to be very comfortable with coding. I think you have to, you know, like Python, at least know in Python, you know, ideally like eventually like C and stuff like that. But I think it's more important is to have this like curiosity and like desire to learn things and just be very, very proactive. Because even right now there's a lot of information and very useful information Internet you can find. There's like many people you can talk to and learn. It's just like trying things quickly iterating. I think it's very important to very quickly iterate and get results. That's how you learn the fastest. Like trying things yourself rather than, you know, okay, so you can read like a paper and then just like oh, this makes sense. But if you haven't implemented yourself, you like honestly probably not going to have very deep understanding of things and try and those ideas quickly I think is very important. So that's why it's very important to be like a good engineer. So you can like implement things very quickly and then like test hypothesis and see if things are going well. And I think the other very important quality is not be afraid of like things that seems like impossible. You know, like before you say oh yeah, like Google is impossible to do Google like you don't, you don't actually need to build Google, you have to build something else. And so a lot of people who maybe have a lot of experience, I feel like they're like maybe less flexible in thinking. It's just some things are possible and some things are not because they're like, okay, so let's say they work at Google, they're like know like how complicated it is, like how much infrastructure Google hasn't. Like you talk to them and they're like say oh this is like impossible to build, right? But then, but then it's clearly possible, right? Like OpenAI for example, show that you can do something very cool without like having Google infrastructure be flexible or like have this trade off where you have like necessary skills but you're also not too rigid to try new ideas and let the past go.

Pardon my word here, but like it's a healthy dose of being delusional a bit and wanting to break through that barrier. That's very important. You see this also with world class athletes, for example. Like you need to be a bit delusional to think that you can be the best in the world at something, right?

I think it's very important. Yeah. Like I remember clearly early on in perplexity, like thinking myself. Oh dude, like when we're especially thinking about like building search engine. Yeah, this seems like kind of like an impossible task. But if you have this mindset from the beginning then you just like probably you shouldn't even start. But yeah, luckily we didn't have that mindset.

And I want to talk to you about search, but before that I also want to ask you a question about how your career trajectory had changed. You mentioned you were from research. You know, you worked at Meta and Research and you also worked at Quora. What was the biggest challenge for you transitioning from research leading a large scale technical organization at one of the biggest AI startups on the planet? Like yeah, walk me through kind of the leadership experience that you gained, what you had to learn.

Yeah, I did like two transitions initially I started as like engineer, so I actually work at BIM and then Quora. So I was like engineer. But then like I did like a lot of like machine learning and ranking. But then in 2015 when I first went to Neurips, the biggest machine learning conference, I kind of like got very convinced right away that deep learning is the way to go and it's going to be like the biggest thing. And so I joined Facebook AI research as an engineer initially and then go get my PhD and like convert it to scientist. So that's why I had like very strong like engineering background and I was like able to learn like we were able to like if there's like any algorithm or training system needs to be implemented. So I was able to do this relatively quickly and then learn from that. And then I through like PhD and a research scientist again like more high level problem setting, problem thinking skills. And so now I basically had a very deep understanding like how it needs to be done, but also what needs to be done. That basically helped tremendously, creating perplexity and running it. Because early on I was able to, you know, basically me and John build everything from the beginning. Right. But then also like as we start bringing in people, I was also were able to provide them guidance and like trying to organize them Together like trying to set like directions and goals and sort of like if they struck with some technical problems then I can also like unblock them. So at least for me it was very important to not only be like high level and know how things in theory should work, but also if you needed to just go implement them myself or help people to do that.

Yeah, definitely, that's super helpful. But maybe walk me through as well the people management side of things as well. Like I'm sure like past years have stretched your capabilities on that front as well. Like I can imagine being a researcher to a CTO as well.

Yeah. So this actually surprisingly we very flat organization right now. So there is literally no engineer managers yet. Actually we just hired one person like a week ago before that was basically half of the company reports and then Johnny has the other company reports and it's kind of like more not a manager but like a tech lead type of thing. And that was like very important I think like what happened as we hired those three processes that we had where we had like very aligned people who basically don't need a lot of like babysitting or anything like that. It's just they're like excited to work and it's kind of like wild now that I think about it, that for like almost two years now we haven't have to deal with this. So like there is no time wasted on like people management talking about things or like career growth. You know, what should I do? Like we just work, have like excited project exciting projects and we just like trying to do our best to get there. I think it's very unusual. For example, when I, when I was at Quora, I joined there when there was like 30 or 40 people and at that time there was already like at least seven, eight managers or like at least we like now like I feel like you know how many people? Like 65 people I think. And we don't have managers yet, we only got one. So that was very unusual I would say, which helped us to move very fast.

That's pretty great. And so it's pretty interesting also seeing it's a new technology paradigm, but it's also you can see kind of like the DNA of a lot of old school tech companies bringing in here. So maybe shifting gears. You mentioned here competing with Google, right? Being able to have that mindset of we can compete with Google. So the main competitor of Perplexity is not necessarily ChatGPT is Google. Maybe my question first, setting the stage pages, why do you think search needs disrupting?

Yeah, so There are like few things happen, right? I think like from the beginning Google had like very brilliant idea like page rank and stuff like that. Something that really values high quality information. Something that is very helpful. I mean it still is but like you've seen over last 10 years, maybe more, the system got gamified a little bit and mostly through like ads and SEO. And so like people kind of like figure out how to break original page ranking. And that's why Google is still like tremendously useful. I think it's still going to be very useful and throughout the years I think it's just, it's an amazing system that they build. Like to me like probably like Google search engine is like the most sophisticated system humanity ever built. It's just like insane. But also on the other hand, I think there is like certain things that can be done better. Specifically you can save a lot of time when you don't have to sift through like 10 links and do a lot of manual work yourselves. If you just have a question, you just want to get an answer and then you can imagine in the future this is going to evolve into much more sophisticated workflows and pipelines where you can like going to give those things tasks rather than just like simple questions and they're going to do like work for you. And I think this is a, I wouldn't say it's necessarily directly competing with Google. I think it's just like a different market. And I think like people still even right now like ChatGPT happened, got a lot of traffic, right? But then Google not necessarily lost any traffic itself, they just created a new traffic, right? And so people still, for example we see people use perplexity, people use ChatGPT, people use Google. It's just like all different tasks but because people can do like things much faster now, they still have time to do all of this stuff. And we trying to be basically like complementary. We're kind of like trying to create a segment where we can do probably like best and class experience and then hopefully this segment is going to be large enough where you can create a successful company.

It's really interesting because you're mentioning here kind of the emergence of a new category of products, right? And right now I would say a lot of AI search engines, you can correct me if I'm wrong, Perplexity included is mostly synthesis. We provide you 10 links, let's say they're synthesized, here are the citations, et cetera. But you hinted at this, is that there is a future where there's a lot More agentic use cases to explore. Right. I can easily see a future like I have a question about Python or C and then you have an AI that shows you a lesson in real life and able to let you execute a code. This is kind of where education. Right, so I'm giving you education use case but then there's also you know, book me a flight something along those lines. Right? Yeah. How do you see the future of agentic use cases in search and when do you think that will happen? I assume you guys are working on it already.

So this is something that's definitely going to happen. I think it's already undergo. I feel like technology is not quite there yet. But it doesn't mean you should not be preparing for it because it's definitely happening. That's why we don't have to rebuild Google search engine because this is not exactly what we need. We need something different, something that actually going to support those new use cases, those like agents. And it's not necessarily like having ranking system that like runs like top 10 results because that's not going to be useful. And that's why it kind of like actual like different infrastructure, different system and something that we definitely working on. And I mean I feel like it's definitely going to happen I think and the most important thing is just like there's so much to do as you mentioned there is you can book flights, you can do like lessons, you can do like tasks. So that's why I don't think there's going to be like one winner take all is just a lot of this is going to be product differentiation like something that's why we have we very confident we have a very good shot. OpenAI is going to be there like Google is going to be there. Hopefully we're also going to be there as well.

And you know another question that's always, at least in my mind that I always think about when it comes to search and AI search is that search represents a large amount of the Internet's gdp. If you disrupt the business model of search, this has really large scale cascading effects on publishers. We talked about flight bookings, you know, Expedia's business model radically change if search is no longer the way we're used to it. Maybe for publishers, creators, services. How do you imagine the business model will look like if AI assisted search becomes the norm? And how do you see that changing over time?

Yeah, definitely things will have to change there. I think like there has to be like figure out like some monetization strategies where those can, you know, like right now, like usually how it's happened, like Google for example sends traffic to you and that's fine, but we like perplexity right now also sends traffic. It's not as much as Google, but I think like at least like 10% of our requests result into people clicking links. But I think like the key observation here is just like each click is not equal in terms of the intent, right? So like in Google you can just find something, you click and then you bounce back because it's just like not relevant. So here, if you click, if you click, intent is like so much higher. And so if you wait by intent, so maybe it's not as much of a drop off as currently. So these things, I think that's how you can reward publishers and stuff like that and other actors. But also I think there is, there could, there could be like other models where you can work together. Content generation is still going to be very important, I think. So I guess we'll have to make sure that we can figure out the ways to work with those companies, with those providers. So everybody's happy.

Think about for example, the state of media and the state of journalism. The business model there is already shifting for the past 10 years and it will continuously shift, especially with this technology. And maybe as we're talking about AI, to close out our conversation, I want to shift a bit gears and talk about how you see the large language model space and the generative AI space in general evolving. Perplexity lets you choose which model to use and there's an LLM interface, whether it's an open source model, closed source model. This week we saw releases of two big, actually three big, I think open source model, Phi 3, Llama 3 and now Arctic from Snowflake. Walk me through how you look at the open source versus closed source AI kind of debate, right? And how you think about it and what do you think kind of how the future of AI will play out here?

I basically think like yeah, I'm a big proponent of open source. I think it has a lot of benefits and I literally see why Meta is so eager to do this because many people don't realize it. I think like setting up the standard and just like having the one architecture that everybody's using it, it's very massive. Yes. Some of the, the other arguments is just like the community is just so much larger than any company, right. And like people are so excited about. So if they can figure out like a trick or two where you can save inference or discover like some bugs in the model, I think it's just like super valuable to the original creator of those models. I think in terms of like how all of this is going to play out, I feel like right now we're still like in very, very early stages of this where people kind of like oh, they know what is the difference between GPT4 and llama. So ultimately once you go to very big market, most of the people just not going to know like what it is, what are you talking about? So that's why the only thing is and which something we focus a lot is designing a product design where people don't even need to know what kind of model you're using as long as it gets the job done right. And with this, that's why there is so much opportunity to leverage different models like those open source and closed source and just create a system that can symbiote those things together and they can work very efficiently together and quickly. And in terms of how it's going to go, I think. Right, so different models are going to require different compute considerations for inference and I think different tasks require different compute considerations as well. And so you will have to figure out how to utilize resources efficiently. And then if somebody needs more compute for their task, then I think you can also figure out a way to offer that as a service. If somebody needs very quick answers like summarizations, yeah, you can use a very small model. But if somebody needs super expensive agenting behavior, do some complicated stuff, then I think maybe it's going to be cost a little bit more.

More interesting. So you're alluding to this and I think that connects to my next question because different tasks require you can say like different raw intelligence capabilities within LLMs. Right. And when you're looking at kind of the next generation of LLMs, I often think about, you know, there's different levers at the community's disposal here, like better data quality, larger amount of compute, more data scaling, parameter size. What do you think will need to be true to get significant stepwise improvements in the performance of today's state of the art LLMs?

I think it's basically figuring out like synthetic data. I think that's probably going to be the biggest one. I feel like already like maybe pre training is passing the point where pre training cost is actually not the biggest cost but like data synthetic data generation is the biggest cost. And I think it's definitely going to be true in the next generation of the models because you have like data generation which is like super expensive. That's why like inference is going to be super expensive than you do like pre training. It's like amortizing things together, trying to figure out like synthetic data and self pay. I think that's if you can enable this thing. This is kind of like ultimately basically going to be like unlimited amount of intelligence. But it's clearly very difficult to do and I think. But a lot of people trying to. I'm very excited about that.

Yeah. And we're already seeing the precursors of that to a certain extent. Like 5.3 released this week. A really small model but trained on synthetic data performance.

Like 5.53 is a very good example where pre training compute is much lower than data generation compute.

And we've been talking about intelligence or intelligence. But I think also an equally important, if not but highly underrated, I would say aspect of succeeding in building a generative AI product like Perplexity, because it's not just about calling the API and getting an answer right, but a great user interface and user experience as well. Maybe walk us through some of the lessons that you've learned in building a great generative AI product that has excellent UI and ux. Like what are kind of the UI UX principles that you've built around building generative AI products?

Yeah, yeah. I think it's very maybe underappreciative area where like people, whatever, like UI is not important. But that's clearly not true. I think. Like then a lot of credit goes to our designer Henry, who from the beginning was kind of like I said, we shouldn't do chat interface, we should do something else. I think that was very important. And the more important thing is just as this technology merge, there is going to be a lot of room for something creative in the UI UX space. I'm super excited about it, but it is very important, I would say it's super important to have a UI that is very intuitive, that is like pleasure to use that fast responsive. Yeah. I just cannot stress this enough unfortunately myself I'm kind of like bad generator of ui, but I'm very good discriminator. So I can like tell if I like the UI or not or UX or not. But we have a very good people who can generate it as well.

Yeah, that's really great. Again you talk about this as well, like creating a pleasant experience to use, but also to get the best responses out of an LLM. Right. Like ui. I'm a firm proponent of the concept that good design will like eliminate needing the skill of prompt engineering. Because if you have a great ui, you will always be able to get a good response. Maybe. How do you see that as well the intersection of getting the best out of an LLM with great UX and ui?

Yeah, yeah. I feel like it's silly, you know, like LLM's getting better and better, but it's still. They're not ideal and I think it's going to take some time and that's why kind of like really thinking through how people gonna use LLM and how to minimize the amount of time where you unhappy with the result. So either like providing like some tools or UI elements that can eliminate those pain points. I feel like that's why it's very important to sort of like have your product team and AI team work as closely as possible and brainstorm things together. So then this process should not be disconnected. Otherwise you will not be inhabitable to Lightfoot product.

Dennis, as we close out our chat, do you have any final call to action to share with the audience predictions about the data space, AI space?

Love to.

Can I hear your final thoughts here?

I guess I cannot stress enough like how early we are in this journey. I think it feels like even though it's been like almost two years since I guess one and a half years since ChatGPT happened, but I feel like it was ages. But on the other hand there is so much more to come and so yeah, I'm just like excited to, to continue this journey. I'm excited for different generations of the models see how can we solve the synthetic data problem. I was thinking to myself, what else right now can you be doing that is going to be more fun than AI? And there is none.

That is awesome. 100%. Thank you so much Dennis for coming on Data Framed. Cool.

Thank you so much for having me and that was a great chat to have.

Likewise.