How Deep Learning Connects Marketers to Customers

Have you ever wished that your entire video production workflow lived in one place? Well, you're in luck because that is LTX, the creative suite that was built by LightTricks, all for AI video production. It takes you from idea to final 4K video, all in one workspace. And under the hood, you've got LTX2, the next generation creative engine, powering native 4K, synchronized audio, and cinematic quality fast. So if you make content professionally, this isn't just another tool. This is your new creative home. Go explore it at LTX.Studio. This is Marketing Trends, your number one source for exclusive interviews with chief marketing officers and executive marketing leaders in the Fortune 1000 and beyond. This is Jeremy Bergeron and I interview, collaborate, and partner with world-class CMOs and marketing leaders across industries. Hey everybody, welcome back to Marketing Trends. It's 2023. We're well into 2023. We are in the studio again. And just like the week before, we have another amazing, amazing leader in the virtual studio. I want to tell you a little bit about Yana Yacavlevich. Yana started her career. We're going back to 2003 when she was working for a record label in Australia. And then started to veer and toward more digital media rather than music. And then she went to London and began this kind of digital media career. And started to work for one of the largest social networking sites. You may have heard of it. We'll get into that later. In 2009, she worked at MagNight where she was employee number three. So she's seen a lot of growth, I mean, from startup to IPO. And in 2015, she moved to New York City to oversee the launch of programmatic solutions at Spotify. Shout out to Spotify. So Yana has a lot of deep experience. She joined Cognitive in 2018 and now is leading partnerships and leading client success there. Lots of interesting things happening in your world, Yana. But thank you for being here on Marketing Trends. Thank you so much for having me and a big hello to all your listeners. And that was amazing work on the pronunciation of my name. I think it was the best pronunciation I've ever heard. Nice. Fantastic. Awesome. Well, tell us a little bit about Cognitive. Like for those who don't know about Cognitive, tell us about the company and what you do there in your role. Yeah, absolutely. So Cognitive specializes in deep learning. So we're using deep learning to bring intelligence to advertising. And if your listeners are not familiar with deep learning, it's really sort of all of that AI that touches our lives like Siri, Alexa, facial recognition, self-driving cars. All of that is powered by deep learning. It kind of gives computers that are like the human superpower of generalization. So if you remember when Siri came out, it kind of just happened overnight. Like all of a sudden we could talk to our phones. And the way that happens is these models, these deep learning algorithms are trained with lots and lots of data. So we take this technology and apply it to marketing to predict consumer behavior. So we'll take people who have performed that desired action, combine it with a lot of other data sets, and then use that to try and make accurate predictions for our clients. So of course you've hear all about the chat GPT, you know, hitting everyone's talking about chat GPT and it's hitting it's literally everywhere. All my friends are talking about it and trying it. But I wonder like how you view that because to me you've been in this world for a while. Do you use when when chat GPT kind of hit the mainstream? What was your view of that? Did you feel like that tech is was that impressive to you at all? What was your kind of perspective there? I mean, at first it was not all that impressive, but I would say, you know, the latest release. I am definitely impressed. I think, you know, for so many years we heard about how AI would replace certain jobs. And I think some of them were easy to imagine like when they said, "All right, personal assistance will be replaced with AI." That's easy to imagine, you know, but you know, when they spoke about lawyers being replaced with AI, for me that was a lot harder to imagine until I play with chat GPT. And you know, it can write a contract for you. You can insert the terms and it'll spit out a contract for you. Like, that's pretty powerful. So yeah, I'm definitely excited. Has there been any kind of utility or connection to chat GPT with the offering of cognitive? Like, is there a team thinking about your tying into that some way in new ways at all? Yeah, absolutely. I mean, we're always thinking about large language models and how that can help us understand the context someone is in or the context or sentiment of a page, absolutely. But that's just really sort of a small element of what we do. Okay. Okay. And now before we dive deeper out, I do, I do want to mention this because, you know, just just I think under a year ago, ad week named you, one of the publications, young influential for highlighting innovators who changed the way people think about branding. And it's a, it's a who's who list and you're on that list. So I want to just know about the accomplishment. Like, talk to us about that. What projects or campaigns do you think really helped you like achieve that recognition? Sure. I was hoping you weren't going to bring it up. Obviously, a huge, huge honor. I was incredibly humbled. And of course, my background being in programmatic and ad tech isn't a very glamorous background. So not just a huge honor for me to be recognized, but for someone in the industry. I think, you know, the entry really spoke about, you know, me coming to cognitive and, you know, cognitive was the first to sort of, you know, deliver these deep learning custom algorithms for marketers to make them really powerful. They need a lot of data. So part of my role when I started at cognitive was to license data. But we needed to use data in a little bit of a different way. Most people will buy audience segments, they'll buy data segments and I'll target those segments. For us, we just want to license sort of data in its raw form. So we can use it as an input into our model. So that was a little bit different. Sort of the way we apply data is different for the industry. You know, prior to that, you know, leading programmatic at Spotify, when I joined Spotify, you know, we were just launching programmatic. We started, you know, with display and video, but at the time, obviously we really wanted to run audio as programmatically. So, you know, helping pioneer that, I think, you know, we ran the first audio private marketplace, you know, at the time the open RTV spec didn't have anything around audio. So it was like, you know, how do we standardize this? And, you know, prior to that, when I joined Magnitor at the time, it was called Rubicon Project. And I was employee number three, but employee number three in Europe, Europe just to clarify. You know, when I joined there, Rubicon Project was just doing yield optimization. There was real time bidding hadn't really been invented. So when it was, we supported it, we built RTV technology. So, you know, I was out in market, you know, educating the market on what programmatic was or what real time bidding was, you know, how it worked, you know, and why they should adopt it. Because there was a lot of fear around programmatic at the time. Publishers was scared that, you know, buyers would get access to the inventory cheaply. A lot of people thought, well, this is going to get rid of, you know, a lot of jobs. But, you know, it didn't create it. A lot of jobs, in fact, it created an entire industry and I sort of view AI in a similar way. Wow. So, I mean, cognitive is this well-established, you know, leader in deep learning and technology. Can you talk a little bit about, I think it's called Neural Mind. Yeah. And how it improves upon industry standard marketing data acquisition, which I think is really interesting. Yeah, absolutely. So, Neural Mind is, you know, it is our platform. It is our DSP. And the way it works for market is, say, yes, we want to predict consumer behavior. So, again, what is that desired action you're trying to optimize towards? If you can provide us a signal for that, and it could be quite simple. It could be a pixel on the page, but it could be offline data too. But as long as we can stitch it together and get that positive signal, we can train a model. So, we take their data and there's not really much heavy lifting from the client side apart from providing us with that first party data. Those people that they, you know, they want to find more of. So, we train a model. We combine it with our in proprietary data. So, our contextual features, behavioral features, all of our sort of vetted third party data. We only want a sort of license deterministic, qualified data. So, we combine it all together and we train a model. So, before we even purchase media, we're gathering that data, training a model and making sure we have an accurate model. And then, once we're confident the model is accurate, it will actually purchase media in real time. So, we run these deep learning algorithms in real time. So, when we get an add opportunity, we're evaluating a thousand different data inputs and in predicting the probability of conversion. and then we will bid and then we'll record what happens. Like did this person take an action? Did they not? And the model uses that feedback to sort of retrain itself and to get smarter and better over time. - Wow, can you give us an example of maybe a recent win? I know you've worked with some big brands. I saw some of the real case study examples, but what's something like that's been really exciting in recent kind of months? - So I would say for us, we really sort of excel at solving complex KPIs. So of course, if someone just wants to drive people to their website, absolutely, we can do that. But where we really excel is prospecting so finding new customers. So we work with one of our clients is Adam and Eve. They want to drive a low cost per order on their site, but they also want to drive a high percentage of new customers. So we're very successfully able to do that. So retargeting can be quite easy, but finding prospecting and finding those new customers is a little more difficult. And the other area where we really excel is incrementalities. So if you think about it, if you're a big brand and you spend a lot on television, on outdoor, people know your brand. When you run digital ads, you don't necessarily want to waste your money on the people who know your brand and would go out and buy your product. So one of the things we do is we predict not only the users who are going to convert, if they see an ad, but we're also predicting the users who are going to convert regardless of whether they see an ad. And it's that group of people we don't want to show an ad too, because we don't want to waste our clients' money. So we're very good at that as well. Wow, that's super interesting. So what would you say is the buyer, the persona, of who's the customer for cognitive? So yeah, I always get asked this. Is it are there certain verticals where we perform really well? But honestly, again, it's like we're really good to optimize in towards human actions as long as we have that signal, as long as we can continue tying back those conversions and importing it into the model, we can be successful. I think what's important is we can't train a model of, let's say, two signals. If two people buy a product, it's very hard to see what is the pattern in those two people. If you Jeremy Buy a product and I buy a product, what's similar about us? It's hard to see a pattern. But if 10,000 people buy a product, then the deep learning algorithms start to see a pattern in these people, in their behaviors, that we as humans may not be able to see. So this is like industry agnostic, vertical like gnostic. I mean, this can work really anywhere. Yeah, absolutely. Wow. Has there been some utility in kind of the local SMB world, like the small business world yet? That is a little bit more difficult. OK. Often clients will ask us, how long will it take to train a model? And it's not really about the length of time. It's more about the number of signals as I was saying before. So the more signals, the more accurate our models can be. So like just example, like a dentist office, right? Like if a dentist office has, let's say they're serving a city and they're serving hundreds or maybe even thousands of patients, would that be enough signals for that to work in that kind of a market? Absolutely. They're serving hundreds and thousands of clients, then absolutely. That's so interesting. OK. Cool. So in terms of the gaps that cognitive is really trying to fill, right? It's got a bunch of products. Like what would you think are like if there's these big buckets they're trying to fill? There's the sweetest things that cognitive does. But what are these gaps that they really see? OK, this is what we're solving for. Yeah. I think we really trying to solve the who, what, when, where? Right? And even today, even with programmatic technology is really difficult. And if you think about it, I mean, I feel with programmatic and real-time bidding, we were kind of building the airplane while we were flying it. And as a result, the industry is a little bit disjointed. Like if you think about a marketer, they have their ads server. They have their DSP. They have maybe creative, like a separate provider that does creative optimization. They have their brand safety, their viewability. They have their third-party segments that they're testing. And so they're kind of layering and layering on top. And eventually, they will get the balance right. And they can get performance. But the problem with that is it doesn't scale. Like, you know, you can say that your target demographic is females 18 to 45. But more than likely, there are people outside of that demographic that buy your product. And so how do you find them? How do you target them with an ad? How does this tech build custom algorithms autonomously? So full disclosure, I'm not a data science. But everything has been automated for us. So once we gather that first-party data, combining that with our third-party data and proprietary data assets is all automated and the updating of the models. So once we're live, that feedback loop, that backpropagation, so serving ads to people, seeing if they take an action, all of that is updated in real time. So people often ask, well, how often do you refresh the models? And we're refreshing them in real time, actually. So we have a pretty small team. Yeah. So after startup, is there a minimum effective dose of data entry for it to be truly autonomous? Like, how is there a threshold of we need this much data? Yeah, there is. And it kind of depends on the KPI. I like to say, once we have 10,000 signals, it can just run. But it kind of depends how complex that KPI is. OK. Seems like also this would be really powerful in the e-commerce world as well. Absolutely. Yeah. So we hear a lot about, especially ending 2022, and certainly now in 2023, we're hearing a lot about real time data. Do you feel marketers truly know how to access and analyze and act upon this real time data? You know, I think they know how to. I think they haven't really been given the tools to do so. And you know, you will often hear, you know, marketer say, I have so much data. I don't know what to do with it. And that's kind of what I love about deep learning is you don't really have to understand the data. As long as you're giving the algorithm clean data, quality data, you don't really want to give it probabilistic data. But if it's a deterministic data signals, it will decide what data is important. But it will do this dynamically. So it's like, what data is important at this moment in time? So if you think about someone who's going to buy a product, they're not buying a product based on the demographic they fall in, or they're not buying a product based on the time of day, they're not buying a product based on what location they're in. It's usually a combination of these things. And it's deep learning that sort of can adapt. And look at all of these inputs, apply a waiting to it, and then again, predict the probability of conversion. And it's doing this dynamically. And it's learning all the time and getting more accurate. So do you think there's such thing as too much data? Like in other words, are untrained marketers subject to this analysis paralysis where they have a ton of data and no guidance on how to use it? Yeah, I mean, absolutely. And again, it's usually fragmented, disjointed. But of course, as a human, how do you analyze that data? Let's say it was all stitched together. And you can see, these are the people converting. But how did they convert? Were they on their phone? Were they on their PC? Were they traveling? Were they at home? I just think there's too many factors for a human mind to compute that. And then of course, activating that in real time and human behaviors are changing all the time. And we saw actually during COVID for our clients, whoever was the cluster of the main cluster of people performing, that really shifted their behavior's shifted. But because our algorithms are adapting all the time, it was able to capture those new audience, those new people. Wow. I want to know about my space. Yes. I want to understand what that's like. I'm sure that's-- I'm sure that's come up at many a cocktail parties. But what was it like to work there? What was your experience there? Yeah, I'm just curious. I know it's been a few years. But-- Yeah. So the beginning of my career was quite colorful. As you mentioned, I started at a record label in Australia. It was a home of Kali Mineric, ACDC. And that was kind of-- we had a department called the New Media Department. So I'm really showing my age now. But I became interested in the New Media Department. And then I moved to London. And while I was in London, I was actually working for one of the largest social networking sites. But it actually-- it predated my space. It predated my Facebook. And it was actually called Face Party. I'm not making this up. And so at the time, I mean, we did over a billion impressions of mine, but that was back in 2005. So we were-- Wow. --the third most visited site in the UK. It was like Google, MSN, Face Party. It was quite crazy. We had this office that was a rainforest. We had a giant tree house. We had a waterfall. When I talk about this, it kind of sounds like I'm making it up. But thankfully, if you go on the internet and Google Face Party office photos, the photos will come up. I would use an incognito browser, just one tip for you. And so I did. I spent two years at Face Party. I had a wonderful time. And in my space launched in Europe. So they were hiring. So I went to my space. And so my experience is a little bit different because everyone was like, well, my space must have been so cool to work for. But obviously, after Face Party, it was a little bit more corporate than Face Party. Because I joined my space after the Rupert Murdock acquisition. But I had a great time at my space, too. I was selling my spaces unsolved inventory to ad networks. So I think that was from 2007 to 2009. I did that. - Wow. - And face party, I think face party was eventually bought by Yahoo, right? - No, no. - Wow. - Face party was never acquired. But face party, it was, I think it was founded back in 1998. It was one of the first social networking sites to combine photos with profile data. - Okay, that's so interesting. Yeah, I wasn't familiar with that, but it was so huge. I'm like, and I don't want to go down the rabbit hole and figure out what the heck happened to face party. - Well, actually very interestingly, they earned face, earn the trademark of face for social networking sites. So I believe Facebook had to acquire it from this party. - Oh, that's an interesting story. Okay. (laughing) That's so, so cool. So, where did the, I guess for you, like where did the love affair with like tech and AI start? Like what was your experience in the moment of like, you must have seen something, you're interacted with something and you like, you got your full attention? What was that moment like? - I mean, certainly, you know, I was, I enjoy working for companies that combine sort of music with tech. I mean, after my space, you know, unfortunately, so, you know, while it was at face party, my space started stealing users from face party. And then I went to my space and Facebook started, you know, taking users from my space. And then Facebook launched in Europe. And I was like, no, enough is enough. Like there's no longevity in social networking sites, which was a huge mistake. (laughing) You know, I'd say, I'd say it's a wonderful thing. But, you know, the GM at my space, he went to Rubicon Project. I actually, you know, knew nothing about Rubicon Project. I was in London. I needed a visa. I'm like, I need a company to sponsor me. Can you sponsor me? It was like, yes, I need someone to sell to the ad network. So it was just, you know, a logical fit. And I think from there, it's, I really love working at startups, you know, building something from the ground up. You know, honestly, it wasn't a conscious decision for me to move solely into ad tech. It just happened and I really enjoyed it. And I worked with some really great people and those contacts and connections I still have today, you know, Jeremy Fein, I'll see your cognitive. He used to work at Magnite. And that's how we know each other. While you're in luck, because that is LTX. The creative suite that was built by LightTrix, all for AI video production. And under the hood, you've got LTX too. Next generation creative engine, powering native 4K, synchronized audio and cinematic quality fast. - We know AI is not, it's not a new technology, right? It's certainly changing and evolving. We talked about JGP, JGPC. And it's certainly a lot more accessible in mainstream. And I think it's going to move even quicker. Are there any more kind of like wins that you have been excited about longer term success stories from your customers, just things that are really kind of fueling the evolution of this that you can share? - I mean, yeah. And then it goes back to solving those complex problems. Like I said, the prospecting, the finding the net new customers, finding incremental customers, not wasting your dollars on targeting people who would have converted anyway. It's also finding high value customers. It's like, yeah, I don't want to just target the people who are going to churn and burn, find me, the loyal customers. I think what we do really takes a lot of the headache out of, you know, AdTech. And you want people to focus their time, you know, especially in marketing on that creative side. You don't really want people having to do repetitive tasks every single day. You know, I think if AI can do that and it can do it more efficiently, you know, why not? - With such this kind of high end, every evolving technology like AI and machine learning, it likely takes like a special type of marketing team. What in your experience, like what kind of unique skill sets and personality traits are you looking for in the team? - Yeah. So obviously, yeah, we're not really hiring marketing people and saying, hey, you need deep learning experience. You need to talk about artificial neural networks. So, you know, I think it's similar with any company. You want people who are passionate, people who want to learn, people who are really good storytellers. You know, what we do, you know, the AI, the science can be quite serious. So we keep our, you know, our marketing tactics and initiatives really fun. And, you know, the same way we're using AI to connect marketers with their customers, we kind of do the same with our marketing team and we want to, we want to use it to connect. So we keep things quite fun and light. Like, for example, the can lions last year, we had an app, which was Rose, not Rose, which I don't know if you've seen Silicon Valley, but when they come up with the app, that's hot dog, not hot dog. - Ah, okay. - And, you know, you know, the investors were really excited because they thought, oh, you take a photo of food and it tells you what food it is. But it was like, no, it just tells you if it's a hot dog or not a hot dog. So we used that. So people could take a photo of their drink and it would tell you, - Smart. - If it is a Rose or not Rose. So that used deep learning. So that's a fun way for people to connect with deep learning. And also, you know, if we sponsor events, depending on what city it is, we'll do goody bags, you know, that are themed, that will take, you know, whatever the sort of local speciality is. Maybe it's a hot sauce or something, you know, something in there, you know, because generally you don't have time to explore the city. And so we keep it fun. We keep it light. - Well, obviously we look around, especially in tech, the world is changing, right? I mean, we're heading into this really interesting time where I think brands of what kind of all shapes and sizes are having to like do more with less and figure out how to, you know, navigate these scenarios. That's all today. Google alphabet had a big layoff. I mean, Salesforce has had this lot of big layoffs that have happened and possibly even more coming. As you kind of look out across, you know, the industries that you serve and support, you're seeing the world change quite rapidly. You're seeing this technology that's really helpful. And you're seeing a lot of companies having to make a lot of moves. Like, what's your perspective there is these companies are trying to do more with less and trying to continue to grow and serve their customers and having to do it in a much different way? Are y'all positioned even more now because the way things are going or you feel like you're having to fight for more market share because things are shifting so interestingly? - Yeah. And, you know, for all those people who are being laid off, I really, you know, do hope they find work. I mean, here at Cognitive, we are hiring. So I guess that's a good sign. - That's huge. - That's huge. - That is a good sign. A lot of especially brands and marketers, it's hard to find the talent in, you know, and you don't have unlimited resources. So, again, you know, I think, you know, when you hear sort of that hands-on keyboard, you know, if you remember back in the day where there were sort of some bad players taking really huge margins in the industry, not providing value. So brands and agencies are like, hey, we have to take this in-house. We need to build teams. But that's not always scalable. And I think that's what we're seeing now. And a lot of people just, you know, have too much to do to do everything 100% effectively. So, you know, I do think AI companies are a well positioned. Also, you know, there, again, going back, you know, I was saying, you know, you don't really want to do a repetitive task every single day, you know, if a job isn't enjoyable and it can be done, but, you know, with AI, let people, you know, go back to being creative, follow their passions. And again, you know, AI doesn't get tired, you know, it's, you know, it's not prone to human error. It's working 24/7. So, you know, if you can let it purchase your media and you know it's doing it more effectively and at much greater scale than you could, you know, I think that should be very exciting. Well, you've already gone through an IPO. And I'm curious if that's, is that kind of where cars that have said it, is that where we're going here or we can't share extra strategies? I can't share. We're trying to company. Okay. But your answer is I don't know. You know, our focus is definitely, we want to continue to grow. You know, we're hiring lots of people. We want to continue expanding our client base. We want to, you know, keep arming people with data. So, yes, we do dope learning algorithms, but that doesn't mean it's black box. So, we're very transparent in everything we do. All of our clients will get access to an intelligence dashboard, which shows them here are the people who are converting. This is how they converted. It's updated all the time so they can see shifts week over week. So, they can use that in their other marketing tactics. You know, I oversee the client's success team. We're very big on service. You know, we provide weekly reporting, month reporting, quarterly business reviews. But we're very transparent in, hey, this is what we tested. Here are the models that we tested. Here are what the models are valuing as important, you know, whether it's demographic or whether it's more the context of the page. Or, you know, these are new, you know, the new customers we drove for you. This is where we found, so we can really deliver sort of unlimited insights that just, you know, depends on the client's appetite. Some love to get into the data. Some are like, no, just hit my KPI. You're all good. What is having this technology do for, you know, cognitive's relationship with growth, right? Because to me, you can use the same technology as your own dog food, eating your own dog food to go generate more interest and get more customers and get more leads. Is it, is it that to that level where you can turn it up and down and kind of dial up growth? And, you know what I mean? Like, is that, is that where we're headed? Because to me, anybody's good at using the tech, it's going to be your actual business. But what does that look like up from the inside? perspective. Yeah, I would say B to B is a lot more difficult because you know, if you think about decision makers, you know, what is their online behavior? And is it any different to someone who's not a decision maker? I'm pretty sure like online that I look like a 15-year-old sometimes, like, you know, at night, I am on TikTok like for way too many hours, and I care to admit. So, you know, I said, no, we're not we're not locating our clients via AI. It's very much through, you know, through our sales team, the traditional way through building those relationships maybe one day. Yeah, I don't think we're far off from from them. I really don't. So, I do want to touch on your going IPO experience because then that's rare, right? To be at early early stage employee and be a part of growth all the way to IPO, tell us that story. I mean, what you learned during that experience, I mean, again, usually the people that start a business are not always the same people that you're there after an acquisition or after an IPO. So, to be part of that core crew, to go all the way through the levels of growth, talk us about, tell us about that because that sounds like a really interesting time and a roller coaster ride potentially. It's definitely a roller coaster, right? It was funny because when I took the job at Magnitle Rebacom Project, as it was known, my manager was like, yeah, this company is going to be sold in here. I was like, wow, that's exciting. And I was there for seven years from when I started to the IPO. But honestly, like, what a journey, I made so many friends there as well. It was, you know, at the beginning, we're in this tiny office. It was a lot of late nights. And, you know, what we were doing, it, you know, it was like, we're doing yield optimization. These publishers who are working with two net and ad networks, but they're not really optimizing and they're not working with multiple ad networks. So in Europe, these ad networks had very close relationships with the publishers. So we came in and as American company and we were like, hey, buy the traffic from us and we said to the publisher, work with us, we'll bring you more revenue, we'll bring you more ad networks. And a lot of the ad networks are like, well, no, no, we're absolutely not going to work with you. And unfortunately, a lot of those big ad networks don't exist now. But the ad networks who were like, and especially when we started building RTV technology, it was like, all right, you know, we can't resist this because one, the publishers wanted to these partners are delivering publishers lots of data inside some more revenue. We need to buy from these SSPs. But the ones who saw like, you know, who were interested in that at the time was called real time bidding technology, they adapted really quickly and they built their own DSPs. So those ad networks went on to be very successful. But, you know, you know, being early stage startup is a lot of work. And like I said, it was a lot of late ads, there was a lot of traveling, it was a lot of being on the road. And it was a lot of sort of, you know, trying to convince these partners who thought, you know, you were going to maybe like eat into their business or illuminate them on, you know, why, you know, why you should work with us. So and then, you know, eventually, you know, publishers became to trust us more and more, they were making more money. They're like, actually, this is actually a very efficient way to buy and sell media. And especially in Europe, where the markets are smaller, the budgets are smaller, but it still takes as many hands to service that IO and to run that campaign manually. So it was really like a no brainer. So it was super nice to see sort of this industry that was a little bit wary of us to being really sort of collaborative partners with us. And like, for example, the trade does, look at huge the trade desk is I remember integrating the trade desk into Magni. You know, that was small, un-nearned companies. Yeah. What a moment in time. It's cool to be a part of that experience. I mean, I think it's so formative to be to anytime someone can go from, you know, early stage to IPO or to some sort of acquisition. Just the, it's like invaluable, the things that I think that you learn as a contributor to them becoming a leader. So now, you know, where you sit and what you're, you know, doing now, what do you feel like you're cultivating as a leader now? Like, what are you really working on? You've done a lot of things. You have a lot of perspective, a lot of deep experience. But like, now as you kind of look out and where you're at and where you're headed, what are you, what are you cultivating now? Yeah. So I think like, last year, you know, my focus was the client success team, you know, which resumed sort of existed within cognitive for a couple of years. And really, you know, we're very good at the tech, you know, our our custom algorithms perform really well. But we don't want to just do that. We really want to also differentiate with our service. So building out a team that works closely with our sales team that works closely with the data science team and can provide this insight and intelligence and service to our customers has been really important. So this year, continue expanding that that team. For me, you know, we started and this is kind of interesting. When we started the company, our vision was never to be the pipes. We just wanted to give people their own custom AI, this custom brain. But very quickly in the beginning, we realized the industry didn't have the infrastructure to support these deep learning algorithms. So we built our own DSP. Last year, we actually started integrating with a couple of SSPs, Andrew and Pubmatic, who can make our custom algorithms available in other DSPs via a deal ID. So if you're a marketer and you want to try deep learning, but you know, you don't have the bandwidth to test another DSP or maybe the client wants to keep everything in in one place. That's okay. We can send you a deal ID. All you need to do is target that deal ID. Don't add any sort of audience targeting on top of it. And you can test our technology that way. And we can still deliver all the same insights for you. So, you know, I would love to see this year us integrate it into more platforms. So, you know, people can test the technology without it being a huge sort of overhaul on their side. Wow. Awesome. And I think we're going to focus a lot this year on probably contextual as well. Okay. Okay. So, Yana, what would you like to see in the next five years? Take a second of the future. What do you see happening in this really interesting space that you're serving for marketers? Yeah, absolutely. Apart from the obvious for everyone to work with cognitive. Now, but what, you know, what I would like to see from marketers is a few things. So, one is the attribution methodology that they use. So many marketers today are still using last touch. But with the cookie going away, they really need to move away from that. So it's one looking at a multi-touch attribution platform. Two, it's like, all right, the cookie list future is coming. You know, looking at other sort of ID solutions, you know, so often we get asked, hey, do you have a cookie list solution? It's like, yes, yes, we do. We can run a Safari traffic for you. We can get our ID on a portion of Safari traffic. We can serve these people with an ad. And we can see if they convert. But, you know, we have to explain to them, you're using DCM to report on your performance. That data is not going to be within DCM because that is cookie-based. So, you know, again, for marketers, move away from last touch, move away from, you know, having everything live within DCM, especially if you want to, you know, test new technologies. Three, I kind of really hope that overreliance on Meta and Google goes away a little bit for marketers because really, you know, you're handing over all of your data to them. They're using that data to fuel their algorithms for your competitors. And you just don't want to be over-reliant on them, I would say. Finally, you know, looking at contextual as well. But, you know, we have chat GBT. Obviously, the technology has improved immensely. You know, moving away from just targeting key words, you know, just because a article might say restaurant three times, that doesn't really give you them, you know, I can't tell you what that article is about. So, you know, adapting to more advanced sort of contextual technologies that are considering all the words on the page, the order of the words, and you know, understanding that sentiment. Good. Awesome. Wisdom and insight from Yana. Pay attention marketers. Let's get into some fun questions. Are you ready for the lightning round? I'm ready. Okay, here we go. So lightning round. Try to be quick, you know, with your answers if you can. And let's have some fun. First question. American or Aussie rules football. I love that you know Aussie rules football. So, I obviously have to say AFL. And if you listen to this, have no idea what AFL is. I would suggest five minutes. Check it out. Yeah. Five minutes on YouTube. It's worth it. It's worth it. It's worth it's worth a child. What's the best city in the world for foodies? I would have to say Melbourne and not because I'm from Melbourne. I have lived in a lot of different cities and countries, but honestly, the food that I'm always surprised when I go back every year at the quality and the standard. There is a huge amount of European and Asian influence. And you know, I live in New York. Obviously, you can get great food in New York. But with these larger cities, usually you need to know where to go. And I feel like in Melbourne, you can't lose. Anywhere you go, the food is amazing. It's fresh. It's delicious. Okay. Here's a doozy. Tom from Myspace or Mark from Facebook. Tom, Tom all the way. I mean, you know, Tom tortoisell, basic HTML. He sold Myspace and he was like, peace out. I'm going to go be a photographer. You know, I don't need world domination. Love Tom. Actually, my claim to fame, this is what it said. But when I was at Myspace, I was friends with the real Tom. You know how everyone got the football Tom. I was the real Tom. That was my claim to fame. That's as good as it gets with me. Oh, that's fantastic. What's your least favorite like business buzzword or marketing buzzword? Friend of me. Because I came across it so often when I was at Magnoi again because of networks. So like, oh, you're a friend of me. We kind of have to work with you. So. Okay, okay. If you could use marketing to send a message to the entire world, what would that message say? Honestly, I would have it say eliminate single use plastic. Yes, okay. That's the only thing that keeps me up at night. It's awesome. What's currently in your Spotify playlist? Who are you listening to these days? Oh, it runs the gamut really. In excess, obviously, I need to have some Aussie music in there. I'm a big Prince fan. I mean, there's some Beyonce. There's some Anderson Pog. No. I dance salsa badly, so there's a lot of salsa music in my playlist as well. Okay. What is one popular activity that you wish you enjoyed more? That would have to be running. I hate running. It's the worst. I could walk a day. I'm with you. I can't even run one block in Manhattan. What would be the title of your unauthorized biography? Spear. No way. That's taken. I would probably call it blunt because I'm very blunt into the point at times. Yes. Blunt. What is the best business advice you've ever received? It would be don't rent out your time. Either earn something or earn a piece of something. Well said. If you were to devote the rest of your life to philanthropy, what cause would you choose? When I started in AdTik, it was a very male dominated industry. I'm quite passionate about getting women into more male dominated industries. I'd probably start with the industry. I know like AdTik and then expand it from there. Great. Okay. Last question. What items are on your bucket list? Well, I'm trying to win Spanish. I've been taking classes for over a year and that is moving very, very slowly. In a sort of business related way. One day I'd love to start my own company. I didn't sell that company. That'd be very nice. But I've been very fortunate. I've lived in a lot of different countries. I've had a lot of different, you know, both for some great companies. Career wise, I feel like, you know, I'm fairly content. Awesome. Well, great. Thank you so much for being a part of marketing trends, Yana. Conversation. Lots of interesting twists and turns where we're headed with AI and deep learning. So excited to see where cognitive is headed. Congrats to you and the whole team. And thanks for being a marketing trends again. Thank you so much and thanks for having me. You know how creative projects usually means juggling six different tools and dozens of different version histories of the project. Well, this is exactly why LightTrix built LTX. All in one creative suite for AI driven video. It handles every phase from conception all the way to final delivery. And it's all powered by LTX too. They're latest and greatest model. It creates 4K, synced audio and video with studio level detail. So no exporting, no waiting, just pure creation. Go try out LTX and see what it actually means to create at the speed of thought. Visit LTX.studio.