Databricks Data + AI Summit 2024 | Tim Shokair | Condé Nast

theCUBE.net

Clips
More from Databricks Data + AI Summit 2024

Tim Shokair, Ph.D.

Senior Director of Data Science

Condé Nast

play_circle_outline Unlocking the Power of Data: An Exclusive Interview with Tim from Condé Nast on User Profiling

play_circle_outline Use cases for segmentation, subscription business, user engagement, and content recommendation

play_circle_outline Development of a tool for searching images within the Vogue Runway app

Info

Tim Shokair | Condé Nast

Tim Shokair, Ph.D.

Senior Director of Data Science Condé Nast

search

>> Good afternoon, nerd fam, and welcome back to sunny San Francisco. We're here at Databricks Data + and AI Summit. My name's Savannah Peterson for theCUBE, and I am particularly excited about our next guest because when I'm not reading about AI or sitting in front of this wonderful camera, thanks to the team at theCUBE, I am reading a Condé Nast publication, whether that be GQ, Vogue, Traveler, you name it, I'm a fan. And we've got Tim from Condé Nast with us right now on the show. Tim, thank you so much for coming to hang out.

Tim Shokair, Ph.D.

>> Thank you for having me.

>> They actually let me have this segment because I'm such a fan.

Tim Shokair, Ph.D.

>> Oh, that's great.

>> I know. I feel really lucky and I'm super curious. I'm sure everyone's familiar with Condé Nast, so we don't need you to talk about the company at all. But I really want to know what data is most important to the company? You've got so many different properties, so many different touch points, subscribers, online viewers, break it down. What's your job like?

Tim Shokair, Ph.D.

>> So, we think about really two kinds of data. There are two pieces that are essential to everything that we do. There's data on our users and data on our content. And from that, we can build everything else. So, if we have a good way to understand our content and we have a good way to understand how our users interact with that content, then we can start to build more models that draw more engagement, that can get as much value as we can out of our properties and our time with our customers.

>> How many touch points do you have, say, on your reader's side? How many different vectors are you looking at to make someone's profile, so to speak?

Tim Shokair, Ph.D.

>> Yeah, there's quite a bit. I mean, there's a lot of how frequently they interact with our content. We have what type of content they interact with. We, to some extent, know how many different brands people interact with. And then, there's the usual, what devices are they on? Where are they in the country?

>> Is there anything that's really surprised you since working for Condé Nast about some of the readers and the patterns they have?

Tim Shokair, Ph.D.

>> I'm sure there is. Let me think for a second.

>> Take your time, seriously.

Tim Shokair, Ph.D.

>> So, I think what I've maybe struck by is how loyal New Yorker readers are.

>> Oh, I like that. Okay, that's good.

Tim Shokair, Ph.D.

>> New Yorker content's pretty engaging, and it is not for everybody, but it is something that demands a lot of time.

>> It is, yeah.

Tim Shokair, Ph.D.

>> And I think what I was shocked by was how loyal those subscribers are, how once they commit to The New Yorker, how nothing changes that commitment to The New Yorker.

>> I am curious because I feel like loyalty is very much a New Yorker trait. I'm a former New Yorker myself. Do you think it's because New Yorkers are loyal, that they're loyal to The New Yorker?

Tim Shokair, Ph.D.

>> Oh, not all of our New Yorker subscribers are in New York.

>> Right, right. Okay. So, it's that demographic, it's that long-form, very well-written researched type writing?

Tim Shokair, Ph.D.

>> Yeah, and I think it's almost like the variety of the content, but just the depth of the style.

>> It is, yeah. Yeah. Ooh, that's super fascinating. I love that. I like that as a little nugget in my pocket. What's your favorite Condé Nast property?

Tim Shokair, Ph.D.

>> The one I probably go to most often is New Yorker, but I go to Bon Appétit quite a bit myself.

>> Oh, yeah, I can get down. God, I forget you own some... How many different magazines is it, 50?

Tim Shokair, Ph.D.

>> In all our brands and markets, we're about 60.

>> It's about 60? Geez, that's a lot. We're all reading it. Okay, so we're here at Databricks. Sorry, I just had to geek out on some Condé stuff there to start off with. How do you use Databricks at Condé Nast?

Tim Shokair, Ph.D.

>> We use it for a lot of things. So, I lead the data science and machine learning engineering teams here. So, we use Databricks. Our data engineering team is processing all our data and storing it within Databricks, making it accessible for us. My team uses it for model development. We do a lot of model exploration. We use a lot of the ML tools that they built in, and then my machine learning engineering team uses it now to take those models, turn them into services. We use feature stores, we use model-serving endpoints. We use feature-serving endpoints. We are big into ML flow. We use it quite a bit.

>> Wow. So, pretty robust across the board. Do you use it for a lot of different use cases? Yeah.

Tim Shokair, Ph.D.

>> Yeah. So, we build models really in a few use cases. The first is really around segmentation, so some of that is segmentation for advertisement, some of that is segmentation on how our users consume content. We also do work for our subscription business where we are understanding propensities and people's propensity to subscribe, people's propensity to churn, and then what their price tolerance is, essentially, for renewals. And then the last kind of more broad areas really around user engagement. So, we do things like content recommendation models. We do that personalized content recommendation models and more generic content recommendation models. We also think about content discovery. So, we've been getting into search both on articles and within images.

>> I bet that could get really interesting, especially across property too.

Tim Shokair, Ph.D.

>> Yeah. So, where we have a good image use case is within the Vogue Runway app. So, in the Vogue Runway app, what we have are about 1.2 million runway images that were from different runway shows, not necessarily labeled and not necessarily searchable for our end users. So, my team built from ideation to finished product, a tool that could take in text and search the images, and it's a really fun, cool tool.

>> That is a really fun, cool tool\, because I'm sure, I, as a user, when you see something in the magazine or on an app and you're like, "Ooh, I want that," or, "I want something like that or something more from that designer. What else is in that collection?" It's a really fun way to deepen that engagement. I bet it's really sticky.

Tim Shokair, Ph.D.

>> It is sticky. We've been seeing now most of our sessions on that Runway app include a search.

>> Oh, nice. Cool. That's got to be kind of validating when you help implement that.

Tim Shokair, Ph.D.

>> Yeah.

>> It makes you feel like you're really reading the data right. You must be a scientist or something.

Tim Shokair, Ph.D.

>> In my former life.

>> Yeah. Why did you choose Databricks to do all this for you?

Tim Shokair, Ph.D.

>> Yeah, so we've been using Databricks since before I came to Condé Nast. So, it was a nice structure that we had built into place, but I've used it in a number of places. It started for me as a data scientist, a tool that we could easily use to collaborate and building off of Spark, being able to process massive amounts of data or build models on top of massive amounts of data was really useful. And then, they've just continued to build features that make it easier for us as data scientists and machine learning engineers to put models into production.

>> It's nice to be able to carry over that toolkit. How long have you been at Condé?

Tim Shokair, Ph.D.

>> Almost three years.

>> Almost three? And where were you before? I'm curious, you mentioned your other life now.

Tim Shokair, Ph.D.

>> Well, I was a data scientist for many years before at CCC, at NBCUniversal, a few startups, but before that I was a particle physicist.

>> Casually.

Tim Shokair, Ph.D.

>> Casually.

>> You were a particle physicist and then you went into TV and now you're at Condé?

Tim Shokair, Ph.D.

>> Yeah.

>> If you don't mind me asking, because I think it's fun and I do think we're at... Well, I like telling stories, but I think that we're at an interesting time in technology where if you have a passion or love of something or a curiosity about something, you can actually career pivot and end up in an industry that's really cool. Can you tell me what guided a little bit of those career pivots?

Tim Shokair, Ph.D.

>> Yeah, so I think at the time data science was a new-ish field where there weren't programs that you could go and study data scientists. I think people were coming from different backgrounds. While working as a physicist, I heard people talk about how, "I was a former astronomer," or, "I was a former statistician or economist, and here's what I did in my work." And I found out how to apply that to tech and get to see really fast results. I was influenced, I got to say, by a talk by someone from Netflix talking about how they were able to use algorithms to predict what people would watch, and I thought that was a really cool use case of the same math that I was doing in physics to something that was kind of a fun application.

>> So, what's harder, particle physics or predicting what people will watch or read?

Tim Shokair, Ph.D.

>> Particle physics we have to be a lot more accurate, so-

>> Yeah, I would imagine there's a little more ramification if you're slightly off.

Tim Shokair, Ph.D.

>> Yeah, I mean they just don't accept your results if you can't prove them to high accuracy-

>> Right. You can't just serve them Tiger King anyway?

Tim Shokair, Ph.D.

>> No, no.

>> Oh, that's great.

Tim Shokair, Ph.D.

>> But in some ways predicting human behavior is pretty difficult.

>> I bet it is really difficult. If I'm someone who still reads pulp magazines in the flesh, does that make me old? Do the kids read magazines?

Tim Shokair, Ph.D.

>> We still have print magazines in my house and there are some kids that still read the print, but I think it's few and far between.

>> It makes us on your elder end of the subscriber base. What excites you most... I mean, we're in the middle of this kind of crazy technological revolution. You obviously have a robust scientific background. What excites you most about this intersection of tooling science and potential and velocity with AI revolution?

Tim Shokair, Ph.D.

>> For me, I think what I get really excited about is thinking about how well these models have to understand content to make these predictions, and that's generally what I like to exploit. I think a lot about the applications are fun and interesting and they're engaging, but what I think is going to really last the test of time is good ways to represent content, so that we can start to create real personalization or real models that can engage people. Because when we're sitting here in this attention economy, everybody's time is so precious and we're all fighting for the same minutes. So, well, I think it's very exciting how we're able to advance within AI and how the intersection of all these things coming is advancing the field. So, I am more interested in how we can exploit that for really practical use cases.

>> Well, and I think that's awesome, and at least the way that I'm hearing that something I look forward to is even more hyper customization. So, rather than waking up to my three different emails or five different emails, it's probably seven different emails from Condé in the morning. I would wake up to one that was my best bits of each one and be able to toggle out, I would imagine, maybe.

Tim Shokair, Ph.D.

>> That is something that we would like to get to, be able to think about how our network can really enhance engagement-

>> And cross-pollinate, so you're keeping everyone in the Condé family, even if they're swapping publications. How long does the average person read an article?

Tim Shokair, Ph.D.

>> Depends on the publication. Not that long.

>> That's what I thought. So, I studied communication. That was kind of right when the jolt economy was coming out, and the stat when I was a researcher in this was 17 seconds was all you had on a video or within a soundbite. So, I'm curious if it's similar.

Tim Shokair, Ph.D.

>> I think what we are seeing is a little bit more than that when you think about New Yorker, maybe a little around that if you think about some of our other shorter publications.

>> Yeah, that makes sense. Okay. Do you think we'll read for longer moving forward? Do you think we'll have more time for reading maybe because of AI? Now, I'm really just going out there.

Tim Shokair, Ph.D.

>> I don't know. What we are pretty excited about is audio and how AI can help with creating narrated audio. While people maybe don't have as much time to sit and read or they don't like to read a publication on their phone, people have AirPods in all the time, and so being able to listen to our stories we found is pretty engaging. So, I'm pretty excited about that. Even if people aren't reading as much.

>> Well, yes, they're still learning and allowed to be... And frankly, with the audio there, it makes it more accessible, and that's a big part of it. It makes me so happy to see so much more accessibility at all of these shows. And is that something that's a big conversation for you guys?

Tim Shokair, Ph.D.

>> Yeah, I like to think about that in different ways that we can make our content accessible to everybody.

>> Yeah, and it makes it more fun and interesting. And you're basically turning the articles into a podcast almost automatically.

Tim Shokair, Ph.D.

>> Exactly.

>> I mean, AI can essentially already do that. Wow, that's really cool. What's the future for Condé? What's the future for you?

Tim Shokair, Ph.D.

>> Future for Condé, I think we are thinking a lot more about how we can leverage all this technology to enhance our offerings and make our content more accessible and easier to discover for our customers. For me, I want to keep pushing forward on bettering our content understanding, expanding beyond text to audio and video and starting to really come up with good recommendation models.

>> What's the top of the funnel for a brand new reader for you guys? How are you getting them?

Tim Shokair, Ph.D.

>> That is a good question. Historically, social was pretty big for us. That's not as big as it used to be. Search was another big one. Organic growth is really the most engaged users that we see. People that come either by referral or come through-

>> Word of mouth, baby. It's always been the best form of marketing. It always has been. And I mean even with referral engines, nothing beats a good friend telling you, "Hey, oh my gosh, I read this super cool article. Here, check out my past episode," or whatever. "Check this link." That's great. Wow, Tim. Okay, last question for you, since we're going to definitely have you back on the show. What do you hope to be able to say a year from now that you can't yet say today?

Tim Shokair, Ph.D.

>> That we have really explored what AI can do to enhance our offerings. We've been toying around, we've been putting on hackathons, we've been getting a lot of good ideas, but really, having something in front of our customers that leverages this while maintaining the integrity of our brands and our journalism.

>> I love that. Okay. Wait, so I have one more follow-up question because that was an intriguing answer. Let's say I'm someone who's a little nerdy, but has always dreamed about working for a publication. I think when people think about GQ or Vogue as examples, or even The New Yorker, you don't always think of high-tech as a role within those organizations. What would be your advice to them?

Tim Shokair, Ph.D.

>> Yeah, so I also found myself in the same position. I never imagined being able to work in media or in publications.

>> It's super cool. Yeah.

Tim Shokair, Ph.D.

>> I was a science nerd. But we live in a place where smart people who know how to solve hard problems can find the applications that really interest them. And I find the best data scientists are the ones that care a lot about the product that they're offering, and they spend time understanding what it is that they're offering. It really helps them drive the models, it helps them understand how to develop and what's going to appeal to the users. So, my advice, find the things that you enjoy, find a way to do data for them because now everybody is trying to exploit data.

>> That was such a good answer and I love that. It really is the passion that will drive you to ask the questions of the technology or prompt the AI or whatever it might be to drive down the rabbit hole and get to the real solutions or get your dream job.

Tim Shokair, Ph.D.

>> Yeah.

>> That was awesome, Tim. Thank you so much and thanks to the team for always creating excellent content that I enjoy reading in my free time. And thank all of you for taking your free time to tune into theCUBE. We're here at Databricks' Data and AI event. My name's Savannah Peterson from San Francisco. You're watching theCUBE, the leading source for enterprise tech news.

Tim Shokair | Condé Nast

search

>> Good afternoon, nerd fam, and welcome back to sunny San Francisco. We're here at Databricks Data + and AI Summit. My name's Savannah Peterson for theCUBE, and I am particularly excited about our next guest because when I'm not reading about AI or sitting in front of this wonderful camera, thanks to the team at theCUBE, I am reading a Condé Nast publication, whether that be GQ, Vogue, Traveler, you name it, I'm a fan. And we've got Tim from Condé Nast with us right now on the show. Tim, thank you so much for coming to hang out.

Tim Shokair, Ph.D.

>> Thank you for having me.

>> They actually let me have this segment because I'm such a fan.

Tim Shokair, Ph.D.

>> Oh, that's great.

>> I know. I feel really lucky and I'm super curious. I'm sure everyone's familiar with Condé Nast, so we don't need you to talk about the company at all. But I really want to know what data is most important to the company? You've got so many different properties, so many different touch points, subscribers, online viewers, break it down. What's your job like?

Tim Shokair, Ph.D.

>> So, we think about really two kinds of data. There are two pieces that are essential to everything that we do. There's data on our users and data on our content. And from that, we can build everything else. So, if we have a good way to understand our content and we have a good way to understand how our users interact with that content, then we can start to build more models that draw more engagement, that can get as much value as we can out of our properties and our time with our customers.

>> How many touch points do you have, say, on your reader's side? How many different vectors are you looking at to make someone's profile, so to speak?

Tim Shokair, Ph.D.

>> Yeah, there's quite a bit. I mean, there's a lot of how frequently they interact with our content. We have what type of content they interact with. We, to some extent, know how many different brands people interact with. And then, there's the usual, what devices are they on? Where are they in the country?

>> Is there anything that's really surprised you since working for Condé Nast about some of the readers and the patterns they have?

Tim Shokair, Ph.D.

>> I'm sure there is. Let me think for a second.

>> Take your time, seriously.

Tim Shokair, Ph.D.

>> So, I think what I've maybe struck by is how loyal New Yorker readers are.

>> Oh, I like that. Okay, that's good.

Tim Shokair, Ph.D.

>> New Yorker content's pretty engaging, and it is not for everybody, but it is something that demands a lot of time.

>> It is, yeah.

Tim Shokair, Ph.D.

>> And I think what I was shocked by was how loyal those subscribers are, how once they commit to The New Yorker, how nothing changes that commitment to The New Yorker.

>> I am curious because I feel like loyalty is very much a New Yorker trait. I'm a former New Yorker myself. Do you think it's because New Yorkers are loyal, that they're loyal to The New Yorker?

Tim Shokair, Ph.D.

>> Oh, not all of our New Yorker subscribers are in New York.

>> Right, right. Okay. So, it's that demographic, it's that long-form, very well-written researched type writing?

Tim Shokair, Ph.D.

>> Yeah, and I think it's almost like the variety of the content, but just the depth of the style.

>> It is, yeah. Yeah. Ooh, that's super fascinating. I love that. I like that as a little nugget in my pocket. What's your favorite Condé Nast property?

Tim Shokair, Ph.D.

>> The one I probably go to most often is New Yorker, but I go to Bon Appétit quite a bit myself.

>> Oh, yeah, I can get down. God, I forget you own some... How many different magazines is it, 50?

Tim Shokair, Ph.D.

>> In all our brands and markets, we're about 60.

>> It's about 60? Geez, that's a lot. We're all reading it. Okay, so we're here at Databricks. Sorry, I just had to geek out on some Condé stuff there to start off with. How do you use Databricks at Condé Nast?

Tim Shokair, Ph.D.

>> We use it for a lot of things. So, I lead the data science and machine learning engineering teams here. So, we use Databricks. Our data engineering team is processing all our data and storing it within Databricks, making it accessible for us. My team uses it for model development. We do a lot of model exploration. We use a lot of the ML tools that they built in, and then my machine learning engineering team uses it now to take those models, turn them into services. We use feature stores, we use model-serving endpoints. We use feature-serving endpoints. We are big into ML flow. We use it quite a bit.

>> Wow. So, pretty robust across the board. Do you use it for a lot of different use cases? Yeah.

Tim Shokair, Ph.D.

>> Yeah. So, we build models really in a few use cases. The first is really around segmentation, so some of that is segmentation for advertisement, some of that is segmentation on how our users consume content. We also do work for our subscription business where we are understanding propensities and people's propensity to subscribe, people's propensity to churn, and then what their price tolerance is, essentially, for renewals. And then the last kind of more broad areas really around user engagement. So, we do things like content recommendation models. We do that personalized content recommendation models and more generic content recommendation models. We also think about content discovery. So, we've been getting into search both on articles and within images.

>> I bet that could get really interesting, especially across property too.

Tim Shokair, Ph.D.

>> Yeah. So, where we have a good image use case is within the Vogue Runway app. So, in the Vogue Runway app, what we have are about 1.2 million runway images that were from different runway shows, not necessarily labeled and not necessarily searchable for our end users. So, my team built from ideation to finished product, a tool that could take in text and search the images, and it's a really fun, cool tool.

>> That is a really fun, cool tool\, because I'm sure, I, as a user, when you see something in the magazine or on an app and you're like, "Ooh, I want that," or, "I want something like that or something more from that designer. What else is in that collection?" It's a really fun way to deepen that engagement. I bet it's really sticky.

Tim Shokair, Ph.D.

>> It is sticky. We've been seeing now most of our sessions on that Runway app include a search.

>> Oh, nice. Cool. That's got to be kind of validating when you help implement that.

Tim Shokair, Ph.D.

>> Yeah.

>> It makes you feel like you're really reading the data right. You must be a scientist or something.

Tim Shokair, Ph.D.

>> In my former life.

>> Yeah. Why did you choose Databricks to do all this for you?

Tim Shokair, Ph.D.

>> Yeah, so we've been using Databricks since before I came to Condé Nast. So, it was a nice structure that we had built into place, but I've used it in a number of places. It started for me as a data scientist, a tool that we could easily use to collaborate and building off of Spark, being able to process massive amounts of data or build models on top of massive amounts of data was really useful. And then, they've just continued to build features that make it easier for us as data scientists and machine learning engineers to put models into production.

>> It's nice to be able to carry over that toolkit. How long have you been at Condé?

Tim Shokair, Ph.D.

>> Almost three years.

>> Almost three? And where were you before? I'm curious, you mentioned your other life now.

Tim Shokair, Ph.D.

>> Well, I was a data scientist for many years before at CCC, at NBCUniversal, a few startups, but before that I was a particle physicist.

>> Casually.

Tim Shokair, Ph.D.

>> Casually.

>> You were a particle physicist and then you went into TV and now you're at Condé?

Tim Shokair, Ph.D.

>> Yeah.

>> If you don't mind me asking, because I think it's fun and I do think we're at... Well, I like telling stories, but I think that we're at an interesting time in technology where if you have a passion or love of something or a curiosity about something, you can actually career pivot and end up in an industry that's really cool. Can you tell me what guided a little bit of those career pivots?

Tim Shokair, Ph.D.

>> Yeah, so I think at the time data science was a new-ish field where there weren't programs that you could go and study data scientists. I think people were coming from different backgrounds. While working as a physicist, I heard people talk about how, "I was a former astronomer," or, "I was a former statistician or economist, and here's what I did in my work." And I found out how to apply that to tech and get to see really fast results. I was influenced, I got to say, by a talk by someone from Netflix talking about how they were able to use algorithms to predict what people would watch, and I thought that was a really cool use case of the same math that I was doing in physics to something that was kind of a fun application.

>> So, what's harder, particle physics or predicting what people will watch or read?

Tim Shokair, Ph.D.

>> Particle physics we have to be a lot more accurate, so-

>> Yeah, I would imagine there's a little more ramification if you're slightly off.

Tim Shokair, Ph.D.

>> Yeah, I mean they just don't accept your results if you can't prove them to high accuracy-

>> Right. You can't just serve them Tiger King anyway?

Tim Shokair, Ph.D.

>> No, no.

>> Oh, that's great.

Tim Shokair, Ph.D.

>> But in some ways predicting human behavior is pretty difficult.

>> I bet it is really difficult. If I'm someone who still reads pulp magazines in the flesh, does that make me old? Do the kids read magazines?

Tim Shokair, Ph.D.

>> We still have print magazines in my house and there are some kids that still read the print, but I think it's few and far between.

>> It makes us on your elder end of the subscriber base. What excites you most... I mean, we're in the middle of this kind of crazy technological revolution. You obviously have a robust scientific background. What excites you most about this intersection of tooling science and potential and velocity with AI revolution?

Tim Shokair, Ph.D.

>> For me, I think what I get really excited about is thinking about how well these models have to understand content to make these predictions, and that's generally what I like to exploit. I think a lot about the applications are fun and interesting and they're engaging, but what I think is going to really last the test of time is good ways to represent content, so that we can start to create real personalization or real models that can engage people. Because when we're sitting here in this attention economy, everybody's time is so precious and we're all fighting for the same minutes. So, well, I think it's very exciting how we're able to advance within AI and how the intersection of all these things coming is advancing the field. So, I am more interested in how we can exploit that for really practical use cases.

>> Well, and I think that's awesome, and at least the way that I'm hearing that something I look forward to is even more hyper customization. So, rather than waking up to my three different emails or five different emails, it's probably seven different emails from Condé in the morning. I would wake up to one that was my best bits of each one and be able to toggle out, I would imagine, maybe.

Tim Shokair, Ph.D.

>> That is something that we would like to get to, be able to think about how our network can really enhance engagement-

>> And cross-pollinate, so you're keeping everyone in the Condé family, even if they're swapping publications. How long does the average person read an article?

Tim Shokair, Ph.D.

>> Depends on the publication. Not that long.

>> That's what I thought. So, I studied communication. That was kind of right when the jolt economy was coming out, and the stat when I was a researcher in this was 17 seconds was all you had on a video or within a soundbite. So, I'm curious if it's similar.

Tim Shokair, Ph.D.

>> I think what we are seeing is a little bit more than that when you think about New Yorker, maybe a little around that if you think about some of our other shorter publications.

>> Yeah, that makes sense. Okay. Do you think we'll read for longer moving forward? Do you think we'll have more time for reading maybe because of AI? Now, I'm really just going out there.

Tim Shokair, Ph.D.

>> I don't know. What we are pretty excited about is audio and how AI can help with creating narrated audio. While people maybe don't have as much time to sit and read or they don't like to read a publication on their phone, people have AirPods in all the time, and so being able to listen to our stories we found is pretty engaging. So, I'm pretty excited about that. Even if people aren't reading as much.

>> Well, yes, they're still learning and allowed to be... And frankly, with the audio there, it makes it more accessible, and that's a big part of it. It makes me so happy to see so much more accessibility at all of these shows. And is that something that's a big conversation for you guys?

Tim Shokair, Ph.D.

>> Yeah, I like to think about that in different ways that we can make our content accessible to everybody.

>> Yeah, and it makes it more fun and interesting. And you're basically turning the articles into a podcast almost automatically.

Tim Shokair, Ph.D.

>> Exactly.

>> I mean, AI can essentially already do that. Wow, that's really cool. What's the future for Condé? What's the future for you?

Tim Shokair, Ph.D.

>> Future for Condé, I think we are thinking a lot more about how we can leverage all this technology to enhance our offerings and make our content more accessible and easier to discover for our customers. For me, I want to keep pushing forward on bettering our content understanding, expanding beyond text to audio and video and starting to really come up with good recommendation models.

>> What's the top of the funnel for a brand new reader for you guys? How are you getting them?

Tim Shokair, Ph.D.

>> That is a good question. Historically, social was pretty big for us. That's not as big as it used to be. Search was another big one. Organic growth is really the most engaged users that we see. People that come either by referral or come through-

>> Word of mouth, baby. It's always been the best form of marketing. It always has been. And I mean even with referral engines, nothing beats a good friend telling you, "Hey, oh my gosh, I read this super cool article. Here, check out my past episode," or whatever. "Check this link." That's great. Wow, Tim. Okay, last question for you, since we're going to definitely have you back on the show. What do you hope to be able to say a year from now that you can't yet say today?

Tim Shokair, Ph.D.

>> That we have really explored what AI can do to enhance our offerings. We've been toying around, we've been putting on hackathons, we've been getting a lot of good ideas, but really, having something in front of our customers that leverages this while maintaining the integrity of our brands and our journalism.

>> I love that. Okay. Wait, so I have one more follow-up question because that was an intriguing answer. Let's say I'm someone who's a little nerdy, but has always dreamed about working for a publication. I think when people think about GQ or Vogue as examples, or even The New Yorker, you don't always think of high-tech as a role within those organizations. What would be your advice to them?

Tim Shokair, Ph.D.

>> Yeah, so I also found myself in the same position. I never imagined being able to work in media or in publications.

>> It's super cool. Yeah.

Tim Shokair, Ph.D.

>> I was a science nerd. But we live in a place where smart people who know how to solve hard problems can find the applications that really interest them. And I find the best data scientists are the ones that care a lot about the product that they're offering, and they spend time understanding what it is that they're offering. It really helps them drive the models, it helps them understand how to develop and what's going to appeal to the users. So, my advice, find the things that you enjoy, find a way to do data for them because now everybody is trying to exploit data.

>> That was such a good answer and I love that. It really is the passion that will drive you to ask the questions of the technology or prompt the AI or whatever it might be to drive down the rabbit hole and get to the real solutions or get your dream job.

Tim Shokair, Ph.D.

>> Yeah.

>> That was awesome, Tim. Thank you so much and thanks to the team for always creating excellent content that I enjoy reading in my free time. And thank all of you for taking your free time to tune into theCUBE. We're here at Databricks' Data and AI event. My name's Savannah Peterson from San Francisco. You're watching theCUBE, the leading source for enterprise tech news.