speaker-0: you
speaker-1: Welcome to this edition of the Infosys Knowledge Institute's Being AI First podcast, the AI interrogator. Today my guest is Cara Caruso from Sitecore. And Cara, I'm delighted to have you. Thank you very much indeed for joining us today.
speaker-0: Thanks for having me.
speaker-1: get started. What are you doing at Sitecore with AI? I mean, we talked about it in a bit, but I'd just like to hear how you're using it and why you've decided to use it.
speaker-0: Well, Sitecore as a software company, we're very focused on customer experiences. And our solutions focus on areas around content, engagement, and commerce. And so what we're doing around AI really focuses in those areas where you think about content and content supply chain. And for us, we are looking at ways to leverage AI in order to make our customers more efficient. and reduce the time it takes to do mundane day-to-day tasks, and instead be able to focus our brainpower on what we think is most important and strategic. And that's really our goal, which is bringing AI into our solutions to reduce time and effort and increase productivity for marketers.
speaker-1: That also implies having humans in the loop. mean, how do you manage the interface between humans and the AI and the automation there?
speaker-0: Today, nothing is still 100 % automated. I wish it was. ⁓ I wish I could be like Dorothy and click my heels and the magical land of Oz and AI all came together. But the reality is we've had to hire a director of product management specifically for AI and how we're going to integrate that in our products. And Musharif really is the one who's helped created that center of excellence in order to have all of our product management and product engineers really understand
speaker-1: Good.
speaker-0: what AI can do and how to integrate it into our solutions. No different frankly than what I think our customers are looking for. They also want to find AI talent, but there's only so much in the market with a new and emerging technology. I 25 years of doing this, we've had a few things introduced in my time. So our clients also still need to hire people who are going to look at these AI models and still be able to train them and have some human oversight to make sure that the actual execution is what we're all expecting.
speaker-1: you finding difficulties in finding the right people? mean, are you hiring outside? Are you partnering with people? Are you training people in psych? How are you handling that?
speaker-0: think a little bit of all of the above. So, Sitecore has obviously made an investment to hire some people and we're training. We are also partnering with companies like Microsoft, who have wonderful technology, who are leveraging things from Copilot and how we'll leverage that in our solutions to their OpenAI models. And we're also leveraging our implementation partners. Those large companies like an Infosys who bring the skill sets to bear. and are able to actually help implement our software and leverage the ecosystem of all these open AI tools from Microsoft and ourselves.
speaker-1: Yeah, because we did some research last year and how companies using particularly generative AI across the world. And one of the key things that they told us was, you know, the skills thing is a real challenge, getting the right skills from the right people at the right time.
speaker-0: Right, no different than any other brand new product or new opportunity you want to bring in. And there's only so much AI telling around. So part of our jobs out there is going to be also training people to get them up to speed about what's possible.
speaker-1: Okay, so I want to move on a bit to talking about personalization because that's also one of the big things that companies told us that they were using AI for. Not so much for content generation, which we were actually quite surprised about. We were expecting to see, ⁓ you know, we're using it for content for, you know, that kind of thing. And they said, no, it's all about personalization. So this is something you've been looking at, right? What does AI bring to personalization and how we do that?
speaker-0: I think what's exciting is we've for years talked about having large amounts of customer data available to us. And now the question is, how do we actually take advantage of that? Not just in pre-planning and saying, Hey, I know that Kara Caruso loves travel and she's interested in these things when she comes to our site. So we want to show her suitcases when she hits our retail site. But it's also about understanding exactly what I'm doing in that moment. And so what AI does is allowed us to build frameworks and to build decision engines that can look exactly what I'm doing in that moment. And maybe I'm on that website and I'm looking at a totally different category in buying clothing for my new niece. And it's great that you thought that I would love suitcases in your planned campaign for spring travel. But what I'm actually looking for is children's clothing. So how in that exact moment? We start to build the frameworks to say, this is actually what I'm interested in right now and how I present content and information and products that would be exciting and engaging to me and actually allow me to want to convert in box. I could give a B2B example as well, but I think you understand what I'm going for here, which is AI can start to make decisions in real time based on actions that are happening in the moment without having to have a person intervening.
speaker-1: does that play with privacy? Because I think, you know, if I thought my browsing, because I have nieces too and buying stuff for new nieces is just a joy. But if I thought I was looking at, I don't know, the Gap website, for example, which has gorgeous baby clothes, by the way, that my browser and AI were working in the background say, ⁓ actually Kate's got a new niece, how do I therefore leverage that knowledge? That feels quite creepy to me.
speaker-0: It's creepy to you, but it might not be so creepy in the moment when you're searching for a child's outfit and they recommend that these perfectly cute socks and shoes match the outfit. And ⁓ by the way, the next time you come back to this site, still being able to show you some products that are of interest to you. And I think that's really the essence, what I want to talk about here, which is building trust. Brands can only do so much as long as they have built trust in the brand. that you're willing to accept those cookies and be able to create that first party relationship. And so part of all of the GDPR and the California and Colorado Privacy Act and all of those things that we're doing ultimately means that you and I as consumers or even in our business day-to-day roles, we have to trust the company enough and believe that it's in our best interest that we'll be better served by agreeing to those cookies and sharing that information. And when you're in that space, it's a little less creepy in the moment if you find that your interactions are better, better for you, ease of finding what you're looking for, quickly have information at your fingertips to do your day-to-day job. I think that's what really makes the difference.
speaker-1: So, are marketers going to be impacted by the death of the third party cookie, which is certainly Google's working on right now?
speaker-0: There's no doubt that there's gonna be some changes. It used to be a lot easier to be able to go out there and leverage third party information in order to build a more interesting and broader customer profile. But we do think that there's still opportunity today because even if you don't know somebody who perhaps is coming to your website, but they've clicked through an ad, they've clicked through a social post, we still start to know where they're coming from and perhaps the last piece of information that captivated their attention and made them want to peruse our site and gain information. So I still think there's going to be a lot of opportunity with first-party cookies.
speaker-1: Okay, so how does AI improve that situation then?
speaker-0: it improves it to the extent that by being more relevant and providing content that you're actually seeking, right, when you come to a site, that you're more likely to accept those cookies in order to get that better experience. And that's ultimately what we're hoping for, which is you come to website and you can't find what you're looking for and you go into the search bar and it's a terrible search experience and the results don't bring back what you want. You may never go back to that website. You may never use that mobile app again, right? So, Ultimately, if you're having positive experiences, you're more likely to accept and continue and engage.
speaker-1: I was searching for something really specific the other day. It was ethernet cable at a specific length because I'm rewiring my flash at the moment. I wanted 10 meters of ethernet cable, Cat7 ethernet cable. It was showing me one meter lengths and 20 meter lengths and Cat5. And it's like, why am I even bothering with this? I suppose I stick with it because I know if I dig through it, but you know, if AI can help deliver a much better search result, that's a great thing, right?
speaker-0: Exactly. And so we're leveraging AI not just for personalizing content, but also helping with relevant search. It's just so important because if you're looking for something very specific, you don't want to point and click in 15 different places and navigate and go through all these different things. What you want to do is type exactly what you're looking for in a search box and get that result. And that again, AI helps.
speaker-1: That's really interesting. So what can go wrong when you're implementing AI in a business? Because this all sounds like a sort of fantastic new dawn for marketers particularly, but what can go wrong?
speaker-0: it's possible that you don't train the models well, right? And so one of the big things before you let a model maybe run, you know, ramp it and go live, you want to test it. You're going to test small cases and use cases. But we are finding customers who are finding success in four to six weeks of training a model and being able to execute a major project. For example, Voreal is a wonderful customer of ours, been a long time customer of SciCorp. And they have over 35 brands, over a couple hundred sites. And so what they're really focused on lately was an SEO project in order to make sure that people were finding the information they wanted and that it was above the full. And so they wanted to have all 200,000 of their pieces of content with better product descriptions and better meta tags. And that way the SEO rankings would be better. The search on their site would be better. And they were able to actually train a model within about a month. and able to execute that. And they think that that saved them over 120,000 man hours of manually processing content. And so that for us is something that, you know, shows the art of the possible in a matter of weeks instead of having to work months and years on projects. So take something small, make sure that the parameters are very doable, and then you're able to actually execute in a timely manner. And that's what we want to do is make sure we're working with our customers working with people to ensure success.
speaker-1: relies very heavily on having the data ready doesn't it? What's your advice there?
speaker-0: So it's having the data ready and having the content ready. But ultimately it's about putting together process and framework. So I like to believe that people process and technology is still what it takes to make a successful project. And so you still need to have those models and you still need to know what are the relevant frameworks for your business, right? So if you're L'Oreal and there's cosmetics and body products and all kinds of different things, you still have to train the model to understand the associations of words and content in order to actually make this process work. And so we believe that our customers can still do small and efficient projects that still have huge impact.
speaker-1: And that's something that's quite interesting actually because I think a lot of particularly smaller companies think, ⁓ AI is just too big for me or too expensive or I can't get it right. Is that true?
speaker-0: don't think so. think there's really small use cases that could be leveraged. As simple as how to start a campaign brief instead of starting from scratch and having to start filling in a template. Perhaps there's a way to leverage and look at all of your current content, your website and campaigns that are already in existence, all your past campaign briefs in order to generate something quickly. And I think those opportunities exist to take even if it's not a matter of a hundred thousand hours. Can I take a couple hours out of my week doing something manual? And what does the impact for me just having a couple hours of efficiency means me from a weekly basis? Because I think there's even small snippets of these cases that are possible.
speaker-1: But though the worry if somebody's, a small business is saying, I know I'll use ChatGPT or Bard or whatever and I'll just upload my stuff and I can have it summarized. What would you recommend to businesses there? Because that's, you know, a risk, isn't it? Uploading all your stuff to a big public database effectively.
speaker-0: Do you think that what we're seeing with our clients, and I've had the real pleasure of speaking to a number of people on different events over the last few months, what we are finding is people aren't doing that in the public domain. What they have done is purchased a private chat GPT model for themselves internally, and they're not connecting it to public data sources. They are keeping it to their own internal use cases, their own internal documentation. And they're not putting that out there to the public domain. So they are bringing the models into a private setting. And they're not just using chat GPT out there for writing their day-to-day content. It might be a little bit different for marketers who are writing letters and events that are eventually going to go on the website anyway. And you want to work on some sample language for a landing page. Maybe not such a big deal to put that in chat GPT because it's public knowledge and marketing content anyway. But at the end of the day, we're not seeing broad use cases like that where we're connecting to public data sources just yet.
speaker-1: That's actually quite reassuring, isn't it? So do you think people, even small companies as well as big companies, are aware of the risks and the problems?
speaker-0: think there were the risks. I think there's also been some concerns voiced around the brand still being able to maintain control of the voice. And yes, we have heard examples of that's too flowery or that doesn't quite make sense. And so I do think that what we're working on is making sure that there's more practical use cases and being able to leverage where you can and still have that human intervention and confirmation. want something has been created in order to ensure it meets what a company and their brand is anticipating.
speaker-1: quite reassuring because you worry about people diving in and using these public tools and potentially throwing away their data or making things worse. And also reassuring actually to hear that people are very aware of the need to have humans in the loop as well. I'm going to come to my final question, which is a question I always ask everybody. Do you think AI is going to kill us all?
speaker-0: Absolutely. I sure hope not. So as a cancer survivor, I will tell you that personally, I have great hope for the future. I mean, I'm here talking about marketing, but certainly in the medical community and what the options are here for us in thinking about research and treatments and potentially cures. I can only hope actually that it's the complete opposite of your question and that AI will eventually do so much more to help all of us.
speaker-1: Yeah, that's very much my feeling about it as well. mean, there's the all sort of the doom and narrative about, it's going to, you know, develop sentience and it's going to take us over. But actually, I very much see it as something that can help us do better and actually, as you say, help us save lives. Thank you very much for time. It's been a real pleasure talking to you today.
speaker-0: Thank you. Thank you, Kate. I appreciate the invite.
speaker-1: The AI Interrogator is an Infosys Knowledge Institute production in collaboration with Infosys Topaz. Be sure to follow us wherever you get your podcasts and visit us on infosys.com slash iki. The podcast was produced by Yulia Dabari and Christine Calhoun. Dode Bigley is our audio engineer. I'm Kate Bevel of the Infosys Knowledge Institute. Keep learning, keep sharing.
speaker-0: you you
speaker-1: Welcome to this edition of the Infosys Knowledge Institute's Being AI First podcast, the AI interrogator. Today my guest is Cara Caruso from Sitecore. And Cara, I'm delighted to have you. Thank you very much indeed for joining us today.
speaker-0: Thanks for having me.
speaker-1: get started. What are you doing at Sitecore with AI? I mean, we talked about it in a bit, but I'd just like to hear how you're using it and why you've decided to use it.
speaker-0: Well, Sitecore as a software company, we're very focused on customer experiences. And our solutions focus on areas around content, engagement, and commerce. And so what we're doing around AI really focuses in those areas where you think about content and content supply chain. And for us, we are looking at ways to leverage AI in order to make our customers more efficient. and reduce the time it takes to do mundane day-to-day tasks, and instead be able to focus our brainpower on what we think is most important and strategic. And that's really our goal, which is bringing AI into our solutions to reduce time and effort and increase productivity for marketers.
speaker-1: That also implies having humans in the loop. mean, how do you manage the interface between humans and the AI and the automation there?
speaker-0: Today, nothing is still 100 % automated. I wish it was. ⁓ I wish I could be like Dorothy and click my heels and the magical land of Oz and AI all came together. But the reality is we've had to hire a director of product management specifically for AI and how we're going to integrate that in our products. And Musharif really is the one who's helped created that center of excellence in order to have all of our product management and product engineers really understand
speaker-1: Good.
speaker-0: what AI can do and how to integrate it into our solutions. No different frankly than what I think our customers are looking for. They also want to find AI talent, but there's only so much in the market with a new and emerging technology. I 25 years of doing this, we've had a few things introduced in my time. So our clients also still need to hire people who are going to look at these AI models and still be able to train them and have some human oversight to make sure that the actual execution is what we're all expecting.
speaker-1: you finding difficulties in finding the right people? mean, are you hiring outside? Are you partnering with people? Are you training people in psych? How are you handling that?
speaker-0: think a little bit of all of the above. So, Sitecore has obviously made an investment to hire some people and we're training. We are also partnering with companies like Microsoft, who have wonderful technology, who are leveraging things from Copilot and how we'll leverage that in our solutions to their OpenAI models. And we're also leveraging our implementation partners. Those large companies like an Infosys who bring the skill sets to bear. and are able to actually help implement our software and leverage the ecosystem of all these open AI tools from Microsoft and ourselves.
speaker-1: Yeah, because we did some research last year and how companies using particularly generative AI across the world. And one of the key things that they told us was, you know, the skills thing is a real challenge, getting the right skills from the right people at the right time.
speaker-0: Right, no different than any other brand new product or new opportunity you want to bring in. And there's only so much AI telling around. So part of our jobs out there is going to be also training people to get them up to speed about what's possible.
speaker-1: Okay, so I want to move on a bit to talking about personalization because that's also one of the big things that companies told us that they were using AI for. Not so much for content generation, which we were actually quite surprised about. We were expecting to see, ⁓ you know, we're using it for content for, you know, that kind of thing. And they said, no, it's all about personalization. So this is something you've been looking at, right? What does AI bring to personalization and how we do that?
speaker-0: I think what's exciting is we've for years talked about having large amounts of customer data available to us. And now the question is, how do we actually take advantage of that? Not just in pre-planning and saying, Hey, I know that Kara Caruso loves travel and she's interested in these things when she comes to our site. So we want to show her suitcases when she hits our retail site. But it's also about understanding exactly what I'm doing in that moment. And so what AI does is allowed us to build frameworks and to build decision engines that can look exactly what I'm doing in that moment. And maybe I'm on that website and I'm looking at a totally different category in buying clothing for my new niece. And it's great that you thought that I would love suitcases in your planned campaign for spring travel. But what I'm actually looking for is children's clothing. So how in that exact moment? We start to build the frameworks to say, this is actually what I'm interested in right now and how I present content and information and products that would be exciting and engaging to me and actually allow me to want to convert in box. I could give a B2B example as well, but I think you understand what I'm going for here, which is AI can start to make decisions in real time based on actions that are happening in the moment without having to have a person intervening.
speaker-1: does that play with privacy? Because I think, you know, if I thought my browsing, because I have nieces too and buying stuff for new nieces is just a joy. But if I thought I was looking at, I don't know, the Gap website, for example, which has gorgeous baby clothes, by the way, that my browser and AI were working in the background say, ⁓ actually Kate's got a new niece, how do I therefore leverage that knowledge? That feels quite creepy to me.
speaker-0: It's creepy to you, but it might not be so creepy in the moment when you're searching for a child's outfit and they recommend that these perfectly cute socks and shoes match the outfit. And ⁓ by the way, the next time you come back to this site, still being able to show you some products that are of interest to you. And I think that's really the essence, what I want to talk about here, which is building trust. Brands can only do so much as long as they have built trust in the brand. that you're willing to accept those cookies and be able to create that first party relationship. And so part of all of the GDPR and the California and Colorado Privacy Act and all of those things that we're doing ultimately means that you and I as consumers or even in our business day-to-day roles, we have to trust the company enough and believe that it's in our best interest that we'll be better served by agreeing to those cookies and sharing that information. And when you're in that space, it's a little less creepy in the moment if you find that your interactions are better, better for you, ease of finding what you're looking for, quickly have information at your fingertips to do your day-to-day job. I think that's what really makes the difference.
speaker-1: So, are marketers going to be impacted by the death of the third party cookie, which is certainly Google's working on right now?
speaker-0: There's no doubt that there's gonna be some changes. It used to be a lot easier to be able to go out there and leverage third party information in order to build a more interesting and broader customer profile. But we do think that there's still opportunity today because even if you don't know somebody who perhaps is coming to your website, but they've clicked through an ad, they've clicked through a social post, we still start to know where they're coming from and perhaps the last piece of information that captivated their attention and made them want to peruse our site and gain information. So I still think there's going to be a lot of opportunity with first-party cookies.
speaker-1: Okay, so how does AI improve that situation then?
speaker-0: it improves it to the extent that by being more relevant and providing content that you're actually seeking, right, when you come to a site, that you're more likely to accept those cookies in order to get that better experience. And that's ultimately what we're hoping for, which is you come to website and you can't find what you're looking for and you go into the search bar and it's a terrible search experience and the results don't bring back what you want. You may never go back to that website. You may never use that mobile app again, right? So, Ultimately, if you're having positive experiences, you're more likely to accept and continue and engage.
speaker-1: I was searching for something really specific the other day. It was ethernet cable at a specific length because I'm rewiring my flash at the moment. I wanted 10 meters of ethernet cable, Cat7 ethernet cable. It was showing me one meter lengths and 20 meter lengths and Cat5. And it's like, why am I even bothering with this? I suppose I stick with it because I know if I dig through it, but you know, if AI can help deliver a much better search result, that's a great thing, right?
speaker-0: Exactly. And so we're leveraging AI not just for personalizing content, but also helping with relevant search. It's just so important because if you're looking for something very specific, you don't want to point and click in 15 different places and navigate and go through all these different things. What you want to do is type exactly what you're looking for in a search box and get that result. And that again, AI helps.
speaker-1: That's really interesting. So what can go wrong when you're implementing AI in a business? Because this all sounds like a sort of fantastic new dawn for marketers particularly, but what can go wrong?
speaker-0: it's possible that you don't train the models well, right? And so one of the big things before you let a model maybe run, you know, ramp it and go live, you want to test it. You're going to test small cases and use cases. But we are finding customers who are finding success in four to six weeks of training a model and being able to execute a major project. For example, Voreal is a wonderful customer of ours, been a long time customer of SciCorp. And they have over 35 brands, over a couple hundred sites. And so what they're really focused on lately was an SEO project in order to make sure that people were finding the information they wanted and that it was above the full. And so they wanted to have all 200,000 of their pieces of content with better product descriptions and better meta tags. And that way the SEO rankings would be better. The search on their site would be better. And they were able to actually train a model within about a month. and able to execute that. And they think that that saved them over 120,000 man hours of manually processing content. And so that for us is something that, you know, shows the art of the possible in a matter of weeks instead of having to work months and years on projects. So take something small, make sure that the parameters are very doable, and then you're able to actually execute in a timely manner. And that's what we want to do is make sure we're working with our customers working with people to ensure success.
speaker-1: relies very heavily on having the data ready doesn't it? What's your advice there?
speaker-0: So it's having the data ready and having the content ready. But ultimately it's about putting together process and framework. So I like to believe that people process and technology is still what it takes to make a successful project. And so you still need to have those models and you still need to know what are the relevant frameworks for your business, right? So if you're L'Oreal and there's cosmetics and body products and all kinds of different things, you still have to train the model to understand the associations of words and content in order to actually make this process work. And so we believe that our customers can still do small and efficient projects that still have huge impact.
speaker-1: And that's something that's quite interesting actually because I think a lot of particularly smaller companies think, ⁓ AI is just too big for me or too expensive or I can't get it right. Is that true?
speaker-0: don't think so. think there's really small use cases that could be leveraged. As simple as how to start a campaign brief instead of starting from scratch and having to start filling in a template. Perhaps there's a way to leverage and look at all of your current content, your website and campaigns that are already in existence, all your past campaign briefs in order to generate something quickly. And I think those opportunities exist to take even if it's not a matter of a hundred thousand hours. Can I take a couple hours out of my week doing something manual? And what does the impact for me just having a couple hours of efficiency means me from a weekly basis? Because I think there's even small snippets of these cases that are possible.
speaker-1: But though the worry if somebody's, a small business is saying, I know I'll use ChatGPT or Bard or whatever and I'll just upload my stuff and I can have it summarized. What would you recommend to businesses there? Because that's, you know, a risk, isn't it? Uploading all your stuff to a big public database effectively.
speaker-0: Do you think that what we're seeing with our clients, and I've had the real pleasure of speaking to a number of people on different events over the last few months, what we are finding is people aren't doing that in the public domain. What they have done is purchased a private chat GPT model for themselves internally, and they're not connecting it to public data sources. They are keeping it to their own internal use cases, their own internal documentation. And they're not putting that out there to the public domain. So they are bringing the models into a private setting. And they're not just using chat GPT out there for writing their day-to-day content. It might be a little bit different for marketers who are writing letters and events that are eventually going to go on the website anyway. And you want to work on some sample language for a landing page. Maybe not such a big deal to put that in chat GPT because it's public knowledge and marketing content anyway. But at the end of the day, we're not seeing broad use cases like that where we're connecting to public data sources just yet.
speaker-1: That's actually quite reassuring, isn't it? So do you think people, even small companies as well as big companies, are aware of the risks and the problems?
speaker-0: think there were the risks. I think there's also been some concerns voiced around the brand still being able to maintain control of the voice. And yes, we have heard examples of that's too flowery or that doesn't quite make sense. And so I do think that what we're working on is making sure that there's more practical use cases and being able to leverage where you can and still have that human intervention and confirmation. want something has been created in order to ensure it meets what a company and their brand is anticipating.
speaker-1: quite reassuring because you worry about people diving in and using these public tools and potentially throwing away their data or making things worse. And also reassuring actually to hear that people are very aware of the need to have humans in the loop as well. I'm going to come to my final question, which is a question I always ask everybody. Do you think AI is going to kill us all?
speaker-0: Absolutely. I sure hope not. So as a cancer survivor, I will tell you that personally, I have great hope for the future. I mean, I'm here talking about marketing, but certainly in the medical community and what the options are here for us in thinking about research and treatments and potentially cures. I can only hope actually that it's the complete opposite of your question and that AI will eventually do so much more to help all of us.
speaker-1: Yeah, that's very much my feeling about it as well. mean, there's the all sort of the doom and narrative about, it's going to, you know, develop sentience and it's going to take us over. But actually, I very much see it as something that can help us do better and actually, as you say, help us save lives. Thank you very much for time. It's been a real pleasure talking to you today.
speaker-0: Thank you. Thank you, Kate. I appreciate the invite.
speaker-1: The AI Interrogator is an Infosys Knowledge Institute production in collaboration with Infosys Topaz. Be sure to follow us wherever you get your podcasts and visit us on infosys.com slash iki. The podcast was produced by Yulia Dabari and Christine Calhoun. Dode Bigley is our audio engineer. I'm Kate Bevel of the Infosys Knowledge Institute. Keep learning, keep sharing.
speaker-0: you you
