Newsfilter article unlocker

Everybody, welcome to KBOS and make some noise for Andrew Bialecki! All right, welcome everybody. Welcome to KDOS, welcome Klaviyo Boston 2026. All right. We've got an awesome two days. I think we have over 50 hours of programming. We've got Gwyneth in the house, Christina Tossie tomorrow from Milk Bar is going to teach us all about her journey. It's going to be a great day. So a couple of things before we kick off. First, I want to say thank you to all of our sponsors, Accenture, Arita, PostPilot, Shopify, and everybody else for helping put on this amazing event. And then I wantto say a big thank youto everybody that showed up here today. We have more than 2,000 people here at KBoss. And I'm so proud that over the last decade plus, Klaviyo has become a platform that's trusted by over 200,000, 205,000 businesses and brands, now in over 100 countries. And the thing I think that makes me the most proud, you can hold your applause for one second. The thing that makes you the most Proud here is how much growing we've done together. Since 2021, collectively, you all have built businesses, and hopefully Klaviyo has played a part in that, where you've generated more than $300 billion in KV, dollars that you've generated through building, delivering experiences on Klavuyo. That I'm super proud of. so I want to go back in time for a second when Ed and I started Klaviyo back in 2012 we had a mission and our mission was we believed the world needs more creators more builders more entrepreneurs more people with a creative spark to take a piece themselves and bring it out into the world. We were inspired by the creators that came before us. And one of the most humbling parts of Klaviyo has been the fact that now it's not us looking at those, looking at, looking at those creators and, you know, them having built things. It's the fact that now the most iconic businesses, brands, and creators in the world are building on top of Klavier. And I think this is only going to accelerate. With AI, I think we're entering an era of abundance. It's never been a better time to start a business or to launch a new product line, and the stats show it. I remember during the pandemic, we saw crazy amounts of businesses start, but it's actually accelerated with AI. The number of businesses started every year has more than doubled to over 5 million, And it's not just new businesses starting. Businesses that generate over a billion dollars in sales have more than doubled the size of their revenue in the last 10 years. So it's everybody that's starting things, founding things, as well as launching new markets, new products. Everybody's going faster. And we want to help you do more. But when we look at the stats, one thing sticks out to us. Despite the fact we've now had ChatGPT and Claude for a couple of years, still less than one in five of the marketers out there is really, according to themselves, when they self-evaluate, is using AI to automate their work, to go faster. We think that number is much too low. Everybody wants to do more, but they're not there yet. So I want to share our kind of mental model for how we think about making you more successful, you, you know, build more products, run more experiments. We break it into three parts. We think AI can help with these three things. The first is automation. We believe that it ought to be possible to create a skill, use AI, so that that next product launch, that next promotion doesn't take weeks or months to plan, but instead can be the kind of thing you can do in days. You can do it with a single prompt. We really personalization. We can now use AI to get very granular to every single one of our customers and have it not be a heavy lift and finally we believe that self-improvement is possible we want everybody to wake up every morning and have an inbox full of ideas from very smart intelligence that's been reasoning about their business coming up with ideas and maybe have already drafted some experiments in some campaigns that you can run so that's what we're building and we're going to talk about this morning. And to start, I want to talk about our data platform, but I want to share that we look at AI as it's going to help with the process. Anybody that's worried about jobs going away, we think AI and agents, they're there for the nitty gritty. But for us, it's all about unlocking creativity and saving the craft for ourselves. All right, so let's dive in. So the first part to us is really all about where Klaviyo started. It's our data platform. Now, for folks that don't know, the Klaviyo data platform, the way we think about it is we want it to be the brain for your business. And so we store everything you know about your customers. Facts about them, activities, actions, events they've taken, any kind of custom objects or data. Could be things like your product catalog, maybe if you're a restaurant or a yoga studio, it's classes or reservations, inventory, everything is there. And we look at it as our job is to make sure that collecting, aggregating all that information across hundreds of different sources, normalizing, making sure it's clean, making sure that data is governed so you have the right access, and then allowing you to query it. That's what it's all about. And we wanna do this at massive scale. We've invested a ton this year in making Scalable more, Klaviyo more scalable and more elastic. We're now processing over four billion data points a day, And we store information on over 9 billion consumer profiles. And what are we using this data to do when we talk about all this querying? Well, one is we're using it to stream or trigger various experiences through flows. You're using to do analytics. And you're also using this to do deep personalization through segmentation. But something's been gnawing at me for the last 10 years. We built this great database, and there's all this data inside of it, but you can't really access it, not at least not to its full potential. I get questions from customers that say, hey, I wanna look at my customers and cluster them, maybe based on which products they buy or which stores they frequent, or I'm running all these marketing campaigns and I have questions around whether, is it product launches or educational content or is it my promotions that convert best? All of these nuanced questions that you could kind of hack your way around in Klaviyo, but you couldn't quite do it. Today, we're going to fix all of that. So I'm very excited to announce that now, inside of the Klaviyo data platform, we've opened it up so there's SQL access to all of the data that you're storing in there. And we're making all of this available to start through Composer. We'll show you some examples in a second. And then through any MCP connector you have. So whether you use ChatGPT, or Cloud, or Gemini, or any other tool, you have instant access to all of the data inside of Klaviyo. So you can ask questions about it. You can reason about it, and then we're working on offering a SQL interface in the near future. So let me show you what this practically looks like. Let's walk through a couple examples. So we have our friends at Core Power Yoga, and I think everybody's pretty familiar at Klaviio. You can go to the individual profile page. You can see all of information that's there. And we're used to this. There's a whole stream of activities and events, things that people have done, orders they've placed. For Core Power Yoga, maybe it's a class they've reserved. And what you notice is, if you click deep in here, there's a ton of metadata about what's happening. You know, individual nuanced data about transactions. That's all here, but you couldn't get access to. So I wanna show you a couple of very simple examples of things that you can now do. And we'll use Composer for this example, but you can obviously do it through any MCP or any AI agent. So to start, I might have a basic question about marketing and the incentives I'm offering. I run different kinds of promotions, a percentage off, maybe it's a fixed dollar amount, maybe it's free shipping. I want to know what converts the best and what actually drives sales. That used to be the kind of thing that you have to export all the data out, maybe into another database, maybe to an Excel file. Now you can ask that question and get an answer in seconds. And as you'll see, we'll automatically compare it to industry benchmarks so you can see what's working and what's not. To give you more ideas and figure out where to aim. So then on top of that, it's not just about marketing information. Let's take a question that any subscription business has. I want to understand churn. You know, Core Power Yoga, I have people that are subscribing to classes, and I want to know how many classes does somebody need to attend for them to be a loyal customer. Now just in a few seconds, you can ask that same question to the Klaviyo data platform, KDP, and it will give you back an answer. Hey, when somebody only shows up to one class a month or zero, their odds of churning go way up. And I can instantly decide, okay, great, I need to build a flow around that to take action. And finally, a lot of us like to compare ourselves to previous trends, last year, last month, or to some of our peers. All those great benchmarks that live inside of Klaviyo, we're now infusing them into KDP with SQL and into Composer so that you can see those trends live, you can visualize them. And then critically, as we run all these queries, we'll show you the actual SQL, if you want to nerd out, if want to be very technical, on what the actual queries are you're running. so you can build trust that the data is accurate. We're really, really excited on what you're going to be able to do, the kinds of questions you're gonna be able to ask, and your agents are gonna be able to ask to help you understand your marketing and your customers better. All right, now as we launch this, it's coming out today in preview. So there's a QR code here for anybody who wants to sign up. We'll be rolling this out over the next couple of weeks. So, you know, get on that list. We can't wait to put this to work. I want to share a little bit about the data that'll be available to start. So in true Klaviyo fashion, the plan is that we're gonna have everything in there, all your profile data, all your custom data, events, segments, list memberships, campaigns, all your product catalogs, that's already there. We're working really hard right now to populate all of the other marketing objects, forms, flows, templates in there. And then any custom data that you wanna import. All of that will be available so you can do reasoning on it and then do marketing and build customer experiences on top of it, all inside of Klaviyo. And then finally, I mentioned a little bit as we'll explain the kinds of queries that we're building, it's very important, we think in the age of AI, to build trust. And for us, that means that the semantics of what your data means matters. So as part of the Klaviio data platform, we've introduced a semantic layer, which will intuit based on the types of integrations and connectors you have, what your data means, and then obviously you'll be able to customize it in places where you're doing something a little differently. So I want to give you a quick example. A very common question we get at Klaviyo, whether it's in aggregate about a business or for individual consumers, is, hey, who's the healthiest? How's my revenue doing? Well, here's the reality. Revenue can mean a lot of different things to different people. Are we talking about gross revenue or net revenue? So in this case, for Core Power Yoga, the way we've defined it is, Hey, gross revenue is nice, but what I really want is net revenue. I need to know folks that maybe they said they were going to go to a class, but they had to cancel, or they refunded a product. All of that math and those semantics, that nuance, is now captured inside of Klaviyo. So that when you ask who are my best customers, that concept of best is encoded in Klaviio, and you can expect that both when you're looking at the results or an agent's looking at results, it's using the right definition. Pretty cool, yeah? all right so that's the start for us when it comes to helping you put your data to work we also very much believe in personalization hey can i use the data that i have to build segments or to predict what my customers are likely to do next i want to tell you a little bit about our philosophy on how we see personalization evolving inside of klaviyo i talk to a lot of business brands and everybody wants to do personalization. The challenge is it feels like this big patchwork of different models, different techniques. Some of them work on some channels. They don't work on others. It becomes a big data engineering exercise. I hate this. There's all these creative ideas that people aren't able to take advantage of because of this diaspora of different, you know, models and what works where. Our mission at Klaviyo, whether they're models that we provide or our partners provide or you build yourself, we want to fill this in. We want to give you consistency so that if you have an idea, you or an agent, you can put it to work. So if you Have product recommendations, an algorithm that you have, whether it's Klavuyo or somebody else's, if you have an idea on, hey, who should be included in a particular campaign, a model that you build or one of our great partners. We want you to have consistency that it's going to work across all channels and be performant. So what we're doing is we're using Klaviyo as kind of a router for all of these models, right, and doing the serving, the inference. And we want to promise you a couple things. The first is for any model that's available on Klaviyo when you're doing personalization, we're going to be transparent. We're going to show you the results. With AI, I think a big way we build trust is we see the outputs. We get to sample that and test that so we know how that algorithm is making decisions. The second is we want to give you choice. So wherever possible, we want models that will allow you to optimize for different things. Most basic one would be, hey, do I want to optimize for near-term engagement, say, opens and clicks and messaging, or do I wanna optimize for conversion? Or do I want to test between those different models? You ought to have the ability to do both. And then finally, we expect personalization to prove its worth. We want to give you the ability to easily experiment, so you don't have to take it on faith that any personalization model that we provide or anybody else, that it actually drives performance. We think this is going to help usher in both more comfort with personalizing, but also letting agents help you do more personalization and let them experiment so on top of that as we do this we've realized you know it's time time for a bit of a name change well we've been calling marketing analytics for the last couple of years I remember when we were talking about this we said yeah it's kind of like analysis for marketing well one problem is we don't do just marketing anymore now we see people using you know personalization through customer agent on their websites in advertising it's everywhere it's not just in messaging we're gonna rebrand marketing analytics to be personalization and we're going to bring a lot of the the kinds of reporting features that were in marketing analytics obviously those are going to become more and more available through the kdp natively through composer and mcp with sql and so with that one thing we're really excited about we've been cooking hard on this for the last couple of months is we're adding a whole bunch of new personalization capabilities to klaviyo things that will help you understand who you should be sending to understand better the economics incentives and i want to show you a couple of examples of things that we're adding to our personalization suite now the first is we get a lot of questions from folks around discounts i just showed an example of the kdp you know sort of testing out and showing what kinds of discounting or what kind of incentive program works best we've now built models into klaviyo that will help you understand which customers typically wait for a discount to buy and maybe need that nudge versus which ones are are likely to buy anyways. This is available inside of now personalization today, and it's available across all channels. And then we're extending our audience optimization, which we launched a couple of months ago, which had the ability to remove people that were likely to be not engaged. They probably, the content or a campaign wasn't for them. We're now extending that where we will also look over your entire customer base and find other profiles that maybe weren't included in a segment or list, but for whom we think that particular content is highly relevant. And just like we promised, all of this will give you choice on what you want to optimize for, and will give the ability to experiment to see that this actually works better. For anybody that hasn't tried out what was marketing analytics or personalization, I told you at the top, we believe like this era of AI is gonna usher in way more personalization than we've seen in the past, and the results are real. I've talked to dozens of customers that are seeing double digit lifts across all of their marketing experiences as a result of doing better personalization. Our job is to lower the barrier to entry there so that every single experience gets tailored to that consumer. All right, and on top of the Klaviyo data platform and the personalization we're laying on top, we are incredibly invested in the marketing infrastructure, the channels, the experiences that really make Klaviyo what it is. We've done a lot of work the last couple of years to extend Klavii across many, many different channels. Email, text, now RCS, WhatsApp, mobile, social, and doing more with your website. And I'm very excited. We've invested a lot this year to make our marketing platform more scalable. We now have the ability for brands that are really scaling or have some very, very special launches. we can now send more than 100 million messages in less than 20 minutes for your burstiest workloads. And then if you have flows where, say, you want to trigger them all at once, we now support taking all that huge event stream and triggering more than a million flows an hour. But it's not just scale, reliability, and security, the infrastructure of marketing. We're also making a lot of future improvements. Really excited about things we've done around smart translation for internationalization, making video readily available inside of RCS. You can have deeper, richer experience on that channel. We've done a bunch of work recently with our Flows editor in Canvas. We're now, instead of having just two splits, right, yes or no, we now allow you to have multiple splits, which we're finding both everybody in the audience and the agents are very happy to have. It makes it easier to organize things. And we're trying to make it easier to connect with your customers. So for example, where we can collect more information, for instance, our integration with our great friends at Shopify and their shop auth and app, where people can log in and share their information, an email, a phone number, and their preferences, all without having to type it in. We're continuing to invest in this, and you're gonna hear more about that this afternoon when we go over the marketing roadmap and some of the cool new things we built there. Those are the core of Klaviyo. That's all the infrastructure that sets the table for what we believe comes next, how we're going to use AI to put all this to work. Now, to talk about Composer and Customer Agent that sit on top of all of this great infrastructure, I want to invite to the stage my buddy, Elias Torres, our new Chief Product Officer. Elias? Elias. I'm a very lucky individual three years ago I was at the New York Stock Exchange with Andrew and I had just turned my ACL on my left leg three weeks into the job at Klaviyo I tore my ACL again on the right leg surgery next week you know seriously Andrew I don't have any more knees left. I love working with you, but I got to stop. No, I really mean it. When I say that, I'm very lucky. I'm so very lucky because I've worked with great individuals in my career. And getting to work with one another again is an even amazing time, right? This is not the first time I worked with Andrew. This is us back in 2011, working at a company that David Cancel and I started. I do love his hair. But him and I took a different journey. He went on to build Klaviyo. That's the reason why you all are here today. Me, I took another route. I went with David to HubSpot where I got to learn how marketers do their work. We built the marketing platform in the CRM that you all use, that other people have seen before. I'm still getting over that. And then we went to build a company called Drift, where I learned the power of conversations with your customers. And lastly, I built a company called Agency, where i started to realize the potential of agents. That's all the experience that has brought me here today. And the reality is that I feel like agents just have not delivered. I know that they're amazing, they're intelligent, but we always have to go to them and ask them questions. We have to ask him, please help me accomplish something. And when they answer, we still can't trust the information. We still have to double check all of the work. And the worst part of it is that they still can do the work alone. So we have at Klaviyo an agent called Composer. What he does initially, he created objects and he created audits for you that allow you to double-check the work that you have in there. We already have launched this to 126,000 customers, and customers like Shields are already saving many hours per week. As soon as we got here, we started working. We added MCB connectors, we added skill support to customize with your business processes, and most importantly, we're giving access to SQL. But in the reality, I feel like I had another idea. In my experience, I want to take agents to the next level. I want a level-up composer. My vision for agents is that they need to be proactive. We got to stop asking agents questions. Oh, should I increase my revenue? Should you check this audit? Should you audit this flow? I think that's a complete waste of time. I want the agent to do the work for me and let me know what work needs to be done. Second, we need to start trusting our agents. We need to stop double checking the work. You want to trust a company like Cleviu to build the right agent for you that knows how to operate your business, that knows the industry benchmarks, that he knows the laws and compliance that your flows need to perform in order to give you the maximum revenue. And lastly, we want these agents to be independent. We want them to be able to operate without needing every single instruction, direction from you. Let me show you. You guys want to see it? So we are in a demo account for DemoLogica, and the first thing we want to do is we want to see how Composer is proactive. I had to schedule a campaign for Prime Day, and it was already ready to go, so I thought. But Composer went and noticed and analyzed that I was sending a 20% discount to a million profiles. Let's take a look. What does Composer have to say? So here is using this connection, SQL, and came up with an interesting tidbit. It realized that over 300,000 of our profiles have never bought with a discount. So why are we wasting our margins away? Let's have Composer fix this. It automatically separated into three segments, the ones that don't need a discount and the one that needs a lot of discount, and it's going to create all the variants for me to send this campaign. You guys like that? The demo is going a little fast here, but another thing that I've learned with marketers is that you have to capitalize on the social trends. You have to be ready and monitoring every single thing that is happening out there to to see if you can go and deliver more product to your customers, right? In this case, Dermalogica is gonna check for what is happening on TikTok, what is happenin' on Reddit. And Composer knows how to check all those things for you. But better yet, it's gonna cross-reference your product catalog to see what is the connection with what's happenin on TikTok. So here we have an ingredient called Ectoin. Everybody knows what that is, right. Nope, I use it. And so it says here that 71% of the mentions are for a segment 25 to 34, but the most important bit is that this ingredient is found in Dermalogica's products. So why don't we create a campaign automatically to send that right away? And this is what I mean by it being an expert. You're going to be able to see everything that Composer is checking for you, so you You don't have to worry about a thing. Lastly, we wanted to be independent. And so here, I have a running campaign that has been going on for a while. And Composer is sending me a message and saying, you know what? There are profiles that have acne that they're not performing well. We're ignoring their needs, and we're not sending personalized content. They're not preforming against the industry benchmark, and you don't have to do this anymore. You don't have to go look at the benchmarks and compare it with your campaigns. Composer will do this analysis automatically, and so it's going to create an A-B test for us, so it can run it independently, be safe, and tell us when it has reached the right controls. As you can see here, we're leveraging all the new Composer features of bringing Figma, your Brand Kit, the benchmarks in the industry and the expertise to create campaigns that that look exactly like how you want them, and be able to set that and forget it. I hope you can see the divisions already coming together. Composer is gonna be proactive, it's gonna be your expert, and it's going to be independent. But I have one more thing. If Composer's going be proactive it needs to talk to you. It needs to reach you where you are, and it needs to be reachable from anywhere you want. So we're going to be releasing an iOS app for you. You're going to be able to ask those questions and get all those notifications right in your pocket. So Composer is the agent that is going to help you run your marketing it's going to be the expert that you're going to trust but we need one more agent every business has a website that's all news but every business is going to have an agent you're willing to need a customer agent for every one of your customer for every one of you customers so what we have released already is already reaching 83% of resolution rates with customers like naked wardrobe sending millions of messages and we're not going to stop innovating. We're making it global for our customers. It's going to support already a hundred plus languages. It is going to have automatic locale detection and it's going integrate with Shopify markets to bring in your catalog that is germane to the region. We are also making it smarter. It will be able to connect with your profiles, read and write information as is having conversations with your customers to be able personalize those conversations. We We have auto-training so it can get you up and running much faster. And lastly, you'll be able to configure it all directly from Composer. Let's show it. So here we have our customer Core Power Yoga. And SQL has made an analysis, right, to say, after 10 classes, we should be able, they're more apt to buy some of the clothing and the other materials off your catalog. So we send an email campaign, and when the customer lands on the website, it's already been personalized because we have website personalization, and customer agent knows the customer right away. So when the customers having a conversation, we're able to extract any profile information from the conversation. In this case, we have a blue tank, and so that'll be added by color preferences, right, and sizing. And when the person buys, all that information will store in KDP. But then when the customer is communicating from Canada, it's going to know the location. When they're in London, it is going to another currency. And when they're France, they're going to the language. But you already know my spiel, right? I have one more thing. I'm under a lot of pressure. I have to ship a lot code. And so one of the things that I went, I went deep into the team and I said, what else can we show them? And this is something that we're working, this is very early, but we want to add one more channel to customer agent. Hi, Sarah. Thanks for calling Core Power. How can I help you today? Hi. I used to go to the Back Bay location, but I just moved to Charlestown and want to go there instead. I was wondering if there are any sculpt classes available before 8 a.m.? Let's see. It looks like Charlestoun has sculpt classes available at 6 a.am. and 7 a.a.m. Do either of those times work for you? Hmm. Let's do the one at 7 a.'m. Okay, I've got you booked at 7 a.m. You'll receive a confirmation message shortly. Do you want me to set Charlestown as your new preferred studio for future classes? Yes, that would be awesome. All right, I'm ready to go. I've updated your preferred location. Is there anything else I can help you with today? I'm all set. Thank you so much. Thank you very much, and I just want to first thank the team that has been building this. Appreciate that. And I'm very lucky to be working with you. Thank you. Andrew. Good job, bud. All right, what do we think? All right. So we are, as Aaliyah's talked about, We are rolling out Composer now to everybody. It's going to come out over the next couple of days, the next a couple of weeks. I've already been using a lot internally. We have a lot of internal team get-togethers on a Friday or during the week. And it's pretty great. Actually, the MCB connectors and skills are awesome for talking about automating a lot of the processes that we have. I've got one last thing. When we built Klaviyo, we wanted to build great software, but software really means great infrastructure, great building blocks. And we get a lot of questions from folks, not just about, hey, how can I query my data and get access to it, but then, hey I want to build my own dashboard, can I do that inside of Klavuyo? And we thought about this for a long time. Well, I don't know, now with AI, why don't we just let you go build that yourself? Or hey, I want to take the data that Klaviyo has and can I use it to personalize or drive some experience on my website or inside my mobile app? This is the kind of thing that we've spent a lot of time over the last couple of years building great APIs for, and we mentioned that we have almost 500 different APIs. Today we are officially declaring that Klaviiyo is headless. We have spent the last... Yes, I think this is awesome. We spent the last couple months going through literally everything you can do inside of Klaviyo and making sure you can the same things programmatically. This matters because everybody's a developer now. So having an API, just last week we launched our first CLI, obviously exposing all of this through MCPs. Whether you want to build it directly in code or you want an agent build it for you, we We believe this is the future. So we are investing in making sure that all functionality from Klaviyo here on out is available not just through a great UI, but also through a set of APIs. Let me show you real quick just an example of what you can do. Aliyah showed off some of the things you can through Klaviio directly. I want to give you an example what it looks like in a different spot, inside a cloud. So in this case, a basic report, no problem. We showed off what you can do now with the KDP, with SQL access, through MCP. You'll notice how you can query all of your different sources, merge all that Klaviyo data together. We can run all that analysis, and it will spit back out for you the report that you know and love. And then you can use this, obviously, from Composer, but obviously we'll give you your choice of tools on how you want to drive that. Then, on top of that, we've got a question here, hey, how can I use these insights on my paid media to improve Klaviyo? If you have skills or intelligence that you're using outside of Composer, we believe in Klaviio being open. We want it to be something where you can use Composer. We've done some fine-tuning and some real training to make it excellent at answering your questions, but we also know that a lot of businesses are investing in this in the the rest of the ecosystem and we want to support that too. So you can see here that it's, hey, not only can it come up with insights, but then you can use Klaviyo to actually go build those campaigns or draft them inside of Claude, ChatGPT, or your favorite AI client. Cool? We have a lot of businesses, Klavuyo brands, that are hard at work with us already. I'd encourage you to take out your favorite AI building tool. Everything from internal dashboards and tooling to understand the information, or actually, I've seen some people build some great workflows in the spirit of automating through AI to actually build out and draft campaigns to build that one skill that will set up a promotion or a launch. They've done that both inside of Composer and outside, and also seen some great things where Leah showed off some great agentic functionality we're giving to consumers, we've seen people build some really cool automations and agents that they're then giving to their customers too. Can't wait to see what everybody builds with this. Now, before we close out, I want to bring up on stage a company, Dermalogica, you've seen a bunch of demos with them today, and a guy, Frank, that is really leading on what we see the best companies, the best brands leaning into using their data and AI. But before we do that, we've got a quick video about Dermalogica for folks that need a little refresher. So let's roll the clip. Okay, so I'm looking at your face, but through here. Skin care is not one size fits all. Do you have sensitive skin? Do you Have fine lines? These are all these different pathways that we have solutions for. If I don't know that about you, I'm not going to be able to give you the best recommendation. And without Klaviyo, that insight exists in a silo. With Klaviio, it becomes a customer experience. The right product, the right message, the moment. Dermalogica is a part of the Unilever prestige portfolio. We operate more than 20 e-commerce sites across the US, UK, Australia, and beyond. At that scale, a CRM is not just a nice-to-have, It's the infrastructure on which everything else is built. And that's why we chose Clayford. Please welcome to the stage Frank Freeman. Frank, awesome to see you. All right, so Frank has the, Frank's role at Dermalogica inside of Unilever. He's the head of AI and digital disruption. All right. That's a pretty big job. Tell us a little bit about Dermalogica, Frank, and then a little about what you do inside the company. Of course. First, good morning, everyone. Thank you, Andrew, and to the Klaviyo team. We're delighted to be here. So Dermalogica, we've been selling skincare products since 1986, but we've been teaching estheticians how to give advanced skincare treatments since 1983. So our roots are in education. My personal roots are education as well. And so, yeah, we've got great products. And, yeah. You saw a little bit in the demo there. Awesome. And so Frank and I have been talking about Dermalogica is deep down using their data and putting that to work. Frank, tell me a little about, like, I've heard things like the term face map. Tell me a lot about, what does a personalization mean for Dermagogica customers? Of course. Yeah, so you may have seen in the memo that we're personalizing to the acne customer. There was some insight around conversion rate for the acne consumer. So how do we know that someone might be acneic or could be? So in 2018, we launched a product called Face Mapping. It was one of the industry's first skin analysis tools where the consumer can upload an image on the website. We analyze it with AI and then provide a customized recommendation in the moment. However, that moment is really just the beginning of a personalized journey for that customer. We've got something like 14 dimensions that we know about their skin, so I like to say that we model users in 14-dimensional space. Sorry, I'm a little bit of a data nerd. But yeah, going forward, then we can form hypotheses around different clusters of those users and how do they move together over time. I believe that X segment will convert in Y way if prompted with Z messaging. let's test that hypothesis. And so as a developer, as an engineer, I need to rely on a product that is API first at scale. And so Klaviyo has always been really part of our backbone. And so those custom attributes, that feature has unlocked so much for us because after we've done all of our modeling, we can bring those custom attributes into Klavii and then silo users and run tests on the go. That's awesome. All right, so yes, you're in good company here. I think we're all pretty much self-described data nerds here. All right. Now, we were talking about like, okay, so I love this idea of, hey, we have all this data, and hey, it gives us a lot of freedom to experiment. But I think one of the things I've had a lot to talk to these folks about is how do you build trust in, say, your team? And what's the kind of person that you like to work with at Dermalogica, Unilever. What kind of talent, what does that look like in AI where everybody's going faster, everybody wants to take more risk, but we really care about our brands. What's the kind of person you look for? Yeah, there's a vignette that always comes to mind. This was about 10 years ago, I was hiring for an animator intern. And I had a handful of candidates with strong portfolios. I had this woman come in, she bombed the interview. I mean, just the eye contact was off, very shy and sheepish, and, you know, so towards the end of the interview, I felt like I had sort of made up my mind that maybe she wasn't going to fit with the team, and I asked just one last question. And I said, do you know Adobe Illustrator? And she looked at me and she said, no, but I will by Monday. And I saw that fire in her eye. And I thought to myself, this is the type of person who's going to go the extra mile. This is the kind of person that's going work that all nighter, work through the weekend. Not that I try to do that all the time, but it happens when we need to ship something. Yeah, totally. Yeah. That person, Ashley, is now my operations manager. I rely on her for just about everything. So, yeah, I mean, that's an example. And as we look at, you know, when we use a tool like Klaviyo and we develop talent, one of the things that we really try to do is to, you You know, to give, especially junior operators of a tool like this, the freedom to test and to break things. We want people breaking things. Of course, guardrails are important, but I would rather, you know, all else equal, have someone who's willing to take a risk with something. And, you Know, the truth is, if someone has just a terrible, an objectively terrible idea for a campaign, the data will show very quickly that it was a terrible idea. And so with the right guardrails in place, that person will see, oh, like, that idea is going off a cliff versus control. And so, yeah, we really try to, you know, empower the people that are eager, curious, and willing to take a risk. I love that. Curiosity is surrounded with the Right Guardrails so that people can go experiment and, you Know, you can take risks and it's okay to fail. Yeah. All right. Speaking of building trust, I think we're all kind of thinking about, hey, where we're pushing the boundaries of AI. Tell me a little bit about what are some of the boundaries that you're pushing at Dermalogica, and what are Some of the things that maybe just aren't quite there yet, but you feel like, I don't know, they might be getting close? Sure. Yeah, so as a part of my role, I'm the de facto AI champion at DermaLogica and across the Unilever Prestige portfolio. and so guardrails are very important with AI you know for any whatever you're doing with AI guardrailes are important and you know, for us I like to think that they're important not just because they tell us what we can't do but more importantly because they tell us when we can do. We've got teams across the globe and in you know in places like Europe where people are you know our employees are naturally very risk-averse when it comes to what can I do with data what can i do with AI and I want to push the bounds as much as possible and so the guardrails that we have we've got a document we update it monthly because things are moving so quickly and we say these are the things you can and cannot do both for data inputs data outputs and what you can actually ship and you know so in terms of the data inputs I mean it's it's the intuitive things like we know PII we don't want touching the ai models makes sense um you know so we want to be very thoughtful about that from an output standpoint uh to answer your question specifically some of the things that you know we're not doing that's very clearly in our guardrails we don't like to do that we don' want human generated faces going out into the into the world right product you know if there's anything we can and should be precious about as a brand it's the way that the product comes out of the packaging, the way that it's dispensed, and the way the user applies it, anything that we put out into the world will be real humans there. That's sacred, yeah. Yeah, so I mean, but there's things that you can do with media that, you know, it might be, you know if we're working with video for example, there might be some opportunities to, you now, like old school green screen approach where you're swapping out the background and using AI to create new and interesting backgrounds to create more diverse creative that then gets fed into, you know, the meta-algorithms of the world. So, yeah, but that media front is an important part of our stack. And, yeah. It's, I think, an exciting time on that front, especially, I Think, in the coming months. I expect we'll see some interesting leaps from the tech. Awesome. All right, well, thanks so much. Thanks for the partnership, and thanks for sharing. Yes, we're all on a journey together. to be curious and see where it all takes us. Thanks, Frank. Thanks, Andrew. All right. Let me close with this. Lots of big announcements this morning. The thing I want you to take away is, as we all start to use AI more, three things that we want to help you with. One, automate more of your processes so you can get out of the minutiae and focus on the craft. the second is personalization deep personalization is now possible and we want to make it true across all channels and for all different aspects of what you can personalize for a consumer and third is you should expect AI to be as Aaliyah said proactive it should help you improve automatically that's our commitment to everyone here to all the businesses that are growing around the world we need more creators we cannot wait to see you build next thank you We're going to have a great two days. Thank you for being here. Cheers.