Hi. Good morning, everyone, and also good afternoon for those of you joining us from Europe today. So welcome to Nuance Coveo for Commerce, and this is the spring edition. My name is Shareen AI. I'm part of the product marketing team here at Coveo, and I'm going to be your host today. So welcome. I'm also joined, of course, by my colleagues from our amazing product team here. We have Olivier Teisseur and Anthony Delage, also known as Adi. And they will be doing, most of the heavy lifting today, spending the next forty minutes or so diving into, you know, the latest and greatest functionality for our commerce product, and also providing just a sneak peek into what's coming next. So a quick reminder because, everyone is on listen only mode, due to large amount of attendees that we have on the webinar today. However, we do wanna hear from you. So if you have any questions at all, please pop them into the chat, and we will have or the q and a window, and we will have some time reserved at the end of the session to answer those. The session is being recorded, so that's always a popular question. And you should get, or receive a copy of the recording in your inbox within about twenty four hours of the end of the session today. AI. So, everybody, let's get started. And there we go. So before I hand it over to Olivier, I just wanted to provide a bit of context of what's going on in the market right now. And each year, we commission a research firm to survey four thousand consumers to better understand, you know, shopper expectation, what's frustrating people, our buying behaviors changing, and all that good stuff. So we've been running this survey now, I'd say, for about five years. And if you're interested, we we do summarize the findings in what we call a relevance report. And if you're interested, you can go down download a copy from our website. But in any case, the last couple of years, as part of the survey, we began asking questions to explore how shoppers are interacting or thinking about Gen AI experiences. And what we're seeing, you see some stats appeared, but the first stat comes from our survey. And, it says that sixty five percent of shoppers stated that they're more likely to buy when supported by Gen AI driven guidance. So that's a really, really interesting, stat there. And then the trend is not just what we're seeing in our data, but also, you know, in other areas. So a recent Capgemini report that we found, saw that fifty eight percent of consumers now use Gen AI tools instead of search engines to find product information. And we probably see that reflected in our own behavior. I know myself as a shopper, it's the same thing. Also, interestingly, I found a Boston Consulting Group article that highlighted that the number one reason that consumers are excited about Agentic and commerce. Now whether they call it GenAI or not, some kind of chat GPT format, its ability to answer more complex product questions. So the the the the key takeaway from that, I think, is that expectations are really evolving, and AI really has the potential to significantly transform the way people are discovering products and deciding on the products that they choose to buy. So you see here a beautiful image on the screen because at Coveo, we see the search box evolving really into something a lot more powerful. And what we're calling it is an intent box. So the search box, as you know, basically found all over your site, familiar spot, familiar tool where customers go and they type keywords, hoping to find the right product. But today, shopper behavior is changing. So they're starting to be more inclined to search the way they talk, so using natural language versus keywords and asking questions. So, for example, what are the best shoes for marathon training instead of running shoes for men? Or, what are the best office chairs for me, to help with lower back pain versus, you know, show me your ergonomic chairs? What type of drill can I use maybe for to drill into a concrete wall versus, oh, I'm looking specifically for a power drill? So the queries or the the way they're expressing themselves in that box is changing. Their intent needs to be AI. They have a context and expectation. They're not necessarily just looking for products. They're looking maybe more for guidance or validation or answers. And so that's why the shift from a search box to what we see as an intent box is really so transformative. So it takes search from, you know, just reacting to keywords to actually understanding what shoppers want and helping them, you know, in that discovery journey. So this is kind of how we imagine what it evolving to and looking like. Shopper asks a question, maybe gets a helpful educational answer, something that explains, you know, the what to look for and why. And then based on that, they might see recommended categories. They might see, you know, curated collections or maybe even specific products that match closely with what their intent indicates by the question. There could be maybe suggested follow-up questions to help them narrow things down, kind of like what a store associate would ask in person. Maybe you can even you know, help full videos are surfaced here. Highlight a sponsored product, which might make sense based on their intent. But, really, I think the most important part is that you don't add friction to the product discovery, journey that they're on. So if they type something like small base chauffeur for my living room, you present them with the most relevant products right away. So this is where shopping is heading, we think, and it's happening fairly quickly. And with that, context in mind, I'm gonna pass it over to Olivier, who's going to show you some of the really cool new AI and Agentic features, and bring some of this to life, hopefully. So, AI, take it away. Thanks, Shereen. So hi, everyone. Nice to see you all today. So let me share my and we'll dive right into it. Right? So, we've heard a bit of content for you today, starting with, our new semantic encoder, essentially. So the catalog semantic encoder that we wanna introduce, this is really our new semantic encoder offering that's pretrained and fine tuned on catalog data specifically. So it's really made to increase a breadth of search results that are returned as well as to make sure that we can, you know, truly understand natural language both from a keyword as well as full query perspective. What's important to know here is that we do this in full hybrid mode. Right? So for every query, we'll balance dynamically lexical versus Agentic. And this will definitely reduce the the the need essentially for manual rules and to source entries within search, allowing merchandisers to really focus on, you know, the rules that matter to their businesses. Right? And since we've done this in a multilingual way through multilingual encoders, we support a hundred and eight languages natively, meaning, you know, our customers both with sec both with segregated as well as with catalogs where languages are are combined can leverage this new technology, to make sure to get relevant results in both searching in any language as well as retrieving in any language. Right? So this is really interesting. And we'll go through a small demo of this a bit later on. Afterwards, we also wanted to talk about our new listing page optimizer offering. So this is really exciting. So far, most vendors, including ourselves, right, have been treating listing pages as just another search, if you will, and optimizing for, you know, the traditional KPIs of relevance as well as revenue and conversion. Now we wanted to change that paradigm and come out with an offering that was specifically tailored for listing pages, allowing not only to rank any product, both cold, meaning with no interact with no prior interaction, as well as those that are known, if you will, in the dataset, but also for every listing page. Right? So for every listing page, what we do for every product is we compute a preset of weights, such as click through, add to cart, conversion, predicted revenue, predicted profit, trends, newness, click, you know, etcetera. And then we're able to actually store these data points on products and dynamically combine them at query time for optimal results. This model has an optimization target for revenue per visitor, but we also plan to make this revenue optimization target configurable so that it can actually be configured for conversion, average order value, revenue per visitor, margin per visit, etcetera, and give full control to the merchandisers. The important thing is, as with everything else with Coveo, this is not a decision to be made between this and rules or this and other algorithms. This is combined with rules, allowing you to get the best baseline possible so that you only have to merchandise for your business strategy and not to actually solve for relevant results, which Caveo handles automatically for you. Next but not least, I wanna talk about also another exciting investment we're making around generative category guidance. This is really where we can augment generative answering technology through recommending relevant categories and interpreting the context of the answer to make sure that we can return relevant categories that can entice customers to go from knowledge to, you know, product discovery, if you will. We wanna get in earlier at the customer education stage to make sure that we can you know, to make sure that our customers can become destinations and not just shopping carts, and that their shoppers can actually get information and advice before their purchase journey begins. So here, what we do is once we get an answer back, either from Caveo relevance generated answering or other functionality, we can actually analyze that answer for some ethically relevant categories, and then we'll ground that in the catalog to return relevant categories to that answer as well as generating dynamic links to then end up on a physical listing page. Now that I've said all this through the slides, let me go in a demo and show you both Agentic and generative category guidance in the wild, if you will. So this is our barcode storefront. It's a fully generated storefront. I wanna give, that heads up before we begin. Right? Because not only have the products been generated using generative AI tools, but also the product images, descriptions, etcetera. So while it's a great demo and it allows us to showcase functionality, you will notice some products may look a little strange, and and look a bit impossible. That is because of the generated, demo environment, if you will. Starting with Agentic. Right? I wanna demo a few things here. So if first AI enter a semantically relevant query such as, you know, long wetsuit for paddle boarding, you'll see here that this query normally, typically, through keyword search would actually return a lot of noise either through partial matching or stemming or would return not many results because there's not a lot of wetsuits that are actually tagged for paddle boarding because paddleboarding is not a wetsuit attribute. It's an activity you can do while wearing a wetsuit. Right? So here, we can see the relevance of our semantic encoder on top of our lexical search, ensuring that we actually retrieve relevant results for these types of long tail queries that may be more complex for a traditional keyword engine. The beauty of Caveo not doing semantics as a fallback is that it doesn't actually impact negatively for any of the relevant searches that need precision, and it can expand the searches that are a bit in between where precision may serve a few results but may not be as relevant. If I search for something like a paddle board, right, a traditional keyword query, if you will, Here, we're gonna get relevant paddle boards because semantics understands that, you know, keyword is more relevant in this setting than in another. If, however, here AI search for, you know, wax for a surfboard, well, then we're gonna get a bunch of surfboards because this store doesn't have wax. So the semantics has understood to expand results based on this cord. This demo is readily accessible, so, you know, anybody can feel free to go and play around with it as well. The thing I wanted to mention as well and that I think is really cool here is what we're doing around generative AI with the generative category recommendations as well as, you know, with the intent box and the early versions of what that could look like in the future. So in that same search box, if here instead of typing for a query that leads, you know, dynamically to product, if I instead type for something like a question such as, you know, what is the best paddle board for a beginner, then here, Coveo automatically detects that this is a question and not a product query and will route it to a generative interface. Now, of course, this being a demo, it's gonna do this on me. But now, with Coveo's robustness, it was able to retry essentially and get to a generated answer still. Here you can see at the top, this is really our Coveo relevance generative answering. Now this does not have rich formatting in place, but it could for the purpose of this demo, though. This was not the most relevant piece. What's relevant here, though, to understand is that this whole answer now after its return is being used to actually see this carousel at the bottom here, which then interprets the answer, will parse it, will extract relevant concepts from it, and then match those with categories that are grounded in the customer's catalog directly. What you end up with is being able to recommend dynamically categories that are relevant to an answer to entice customers who may be more looking for education and information to actually go into the product discovery journey and start browsing products. If I click on stand up paddle boards here, for instance, we will dynamically generate then a category that matches with the stand up paddle board topic as a category, and we'll render that for the end customer. Do know that these pages can also then be added to your XML and index to add for more long tail category pages. But overall, this gives a very seamless customer experience where AI clicking on any product here, I can just dive straight into the product discovery journey where Coveo, of course, as usual, is providing recommendations as well as, you know, ranking on listing pages and etcetera. Last but not least, I mentioned the listing page optimizer. And before going back to slides, let me just show you a little bit what I meant. Here, this is a demo site. Right? So there's really no tuning or tailoring being done on any of these listing pages other than within demos to showcase our merchandising functionality. Meaning that if I go to any of these listing pages, right, I still want the link listing ranking to be relevant to the end user. And now remember, I did mention the weird looking products like this one here, but this is definitely supposed to be a surfboard in this context. Right? If I head to any other page, such as here the touring kayaks, which, assumingly, is probably a long tail page, Products are still ranked in a relevant manner because the listing page optimizer not only has knowledge of the full catalog, but is a also able to cold start even for products with no traffic on them. So this was my short demo. Let me head back to slides now for the last section before I hand it off to AD. So now that we covered how Caveo now does semantic understanding through the catalog semantic encoder as well as ranking listing pages and recommending categories, I wanted to talk about an underlying technology that we recently made available to customers to build generative experiences and agentic workflows. So here, just to take a second, what we see at the top here is really the conveyor relevance generative answering functionality. I haven't die I didn't dive deep into this because we have introduced this over two years ago now. But this is our our fully hosted question answering functionality that allows you to take rich content from any source and to get a relevant answer using Caveo as the retrieval engine in a rag scenario using, in this case, GPT four o. However, we have a lot of customers, both that are using CRGA or that, you know, are not using them, but that wanna build custom generative AI solutions such as AI Agentic workflows or through, you know, customer ag systems by using their own LLMs, what we've done is we've actually exposed the underlying technology that allowed us to build CRGA to end users to allow end users to receive trunks instead of answers. Right? Now to to drive this sort of RAG application, you need a few things. First, you need a generative model that's able to interpret, you know, an answer and to be able to summarize it in a relevant way to the end user within the context of the question. You need a source of truth because, otherwise, you have to rely on the generative AI model source of knowledge, right, which is a mistake in many cases because brands want control and deterministicness when it comes to the answers that they give. And third, you need a system that's able to orchestrate them. Coveo CRGA is a fully hosted version of that. In this case now, we're allowing customers now to integrate Coveo within their agentic workflows as well as AI applications by using Coveo as a source of truth to ground their generative answering and generative workflows into. For example, this is already available in Agentic and Jewel as well as in many many other types of orchestrator, as well as we have many many customers now using this in production, both for internal and external use cases when they need to bring documents and product passages within their agentic workflows. So we're really excited about this, and it's opening up brand new types of opportunities where customers can leverage our, you know, fully hosted and managed question answering service for many of their use cases and then build their custom use cases and custom Agentic workflows using the passage retrieval API, ensuring there's a consistency in the way answers are structured and in the type of information that you return to your end users. So with all that being said, I hope you're all as as excited as I am. I think this is all really cool new innovations coming from our teams, and I'll hand it over to AD to talk about and go over what we're doing on the merchandising front. Thank you, Ollie. Hello, everyone. It's a pleasure to see you guys today. Thank you so much for your time, and, looking forward to sharing this stuff with you. So, I'm gonna kick off talking about merchandising and the Merchandising Hub. We have a lot of new stuff that we added and a lot of great stuff coming, that we're super excited to show. And then I'll jump into some more, information about kind of integrations, that that we've developed over the last couple months that we're also super excited about. So first things first in the Merchandising Hub, one of the thing the key kind of investments we've made over the last couple quarters is this, transparency into AI with our ranking details. And we're happy to announce that this is now generally available. All of our merchandising hub customers have access to this. And what this allows you to do is now really understand all of the great AI that's influencing kind of our our decisions and how your merchandising roles are interacting with that. And we're really trying to create this world where, you know, business users are are are not looking at our system like a black box. They they don't think of AI as competing with with their needs, but rather as a compliment, they're looking into kind of a glass box. They see they they they know what's happening in there. I'll show I'll jump right into the product here. So we've added this new preview tab, and I can enter any query into here. So wetsuit for paddle boarding as an example to to follow-up on Ollie's demo. And I can quickly see what are the ranking factors that are actually deciding which products are in which position. And so I can see at a glance kind of this this simple scoring view, but I can also jump in at a specific product level and see, is AI influencing it or merchandising rules influencing it? Is it is it a lexical match? Do we have semantics influencing this? And this is really kind of all the tools you need to understand, hey. You know, why is it here? They don't we we get this situation often. You know, merchandising teams, it's a Sunday morning. Their leadership are browsing the site, and they don't like why why a certain product in a certain position based off of a career they were in. And pretty quickly, it goes up the email chain and and someone just gets an email said, hey. Why is this here? You know, in in the past, this would've been a quick a difficult question to answer. We've made it really easy, really intuitive. And and we hope that this really kinda helps, bolster this this confidence in our in our in our work. And where we see this going going forward in the future is as a constant area of investment, and that ties back into what Ollie just spoke about. So when Ollie's talking about adding, you know, this listing page optimizer that reranks based off of these factors, this is stuff that's gonna flow right into our ranking visibility feature. And so you can expect as we go on, yes, our AI is gonna improve, but you're gonna always be tethered down through these features in the merchandising hub. So we're really excited about this. The next thing I wanna talk about is, some advancements to our recommendation strategies. So, you know, in the in the last new in Coveo in the fall, you would have seen the recommendations manager. But since then, what we've added are new strategies available to merchandisers that are gonna be kinda quite situational. And what they're gonna allow you to do is personalize based off of a customer's behavior. We're we're Coveo. We do data at scale. We do AI at scale. Personalizing based off of users' kind of personal behavior is super important. And so if I go into this, this carousel, I can jump in and change my rules. And we're seeing here that we're using this intent to wear model. And so it's gonna actually, not just take in the product that AI user is currently looking at, but also all of their, behavior in that session or or even prior to to to personalize that that experience. But you can see on the left here and these might be more suitable for a homepage, recently viewed, recently purchased, but buy again, purchase with recently purchased. These are these are all these kind of fanned out strategies that have these great niche applications. We have a lot of customers who have frequent repurchase behavior, really loyal customer bases. Things like buy again, purchase with recently purchased. Those are things that are really gonna kind of encourage that that loyalty and and and and kinda get people back into to purchasing. So, we're really excited about kind of the the the addition of control that we've added, that we've provided to to merchandisers. And and we're seeing this kinda roll out on a bunch of sites and and and and kinda solve a lot of these problems really easily. The next thing I wanna touch on is facet management. So this is something that's currently in early access and we're really excited about this feature in the Merchandising Hub. One of the the the big investments we made, with the Commerce API and the Merchandising Hub was this move in our API structures towards, giving merchandisers more control, at the and and simplifying our customer storefronts. So storefront code gets simpler. Everything lives in the merchandising hub. And one of the big areas where this was gonna play out is facet management. We're super excited now that we're actually gonna be able to to deliver this and show it to people. And so I'm gonna jump in here, to an environment where I've got this enabled, and we're gonna build a facet collection for my search. And so collections are a group of facets that you wanna show in a situation, And you can combine, AI and and manual kind of curation to to define which facets go where. So when I start out with the manual side, I'm gonna pick some pin facets. I might want to always have my brand facet, and I might always want to have, my pro my price facet or my promotional price after, after discounts. And what I also might want to do is, add some facets managed by relevance. And so Coveo's AI is kind of uniquely capable of going in and based on the query that someone provides, actually, picking the right facets for that. And so we're gonna enable that. I'm gonna add five facets this. They're each gonna have four values. I can also manage these pin facets in a lot more detail. So I might decide that instead of sorting my values alphanumerically, I might sort them by by number of products or occurrences. And maybe I only just wanna show five. All of this control is super easy, super intuitive. And now that I've created my, my configurations or my collection, I can go test this on the site. So we're gonna go search for AI. And as you can see here, I've got my two pin facets at the top, but these relevance based facets are showing things like the water type, the audience, the number of people. If a user had come in and searched for dry bag, here, we're gonna get material, size, color, gender. So this is how we scale for search. Right? And when it comes to product listings, you know, which which is gonna be coming later, people tend to just go with manual curation. You know, if I if I'm on a dry bags page, I might know exactly which products which which facets should be shown. But when it comes to search, and this is why we started here, searches needs to scale. Our customers have hundreds of thousands of unique queries, and you can't possibly start to manually merchandise all of those facets. And so using AI is so important in this situation and having actual relevant facets rather than something generic that would just apply to everyone, makes such a difference to the user experience. So we're really proud about this, and we're really excited about what's to come here. We're gonna be continuously adding more control through this early access program using our customers' feedback and their needs to just, further empower our merchandisers to make this, you know, a a great, fasted experience. Next few things I wanna talk about are things that are coming up soon in the in the merchandising hub that we think are are gonna keep AI pushing people to be, you know, more and more efficient, more and more powerful, and be able to kind of drive more business value. The first is flexible targeting. Today in the merchandising hub, there's a lot of capability to create rules. But one of the kind of, you know, points that we've gotten on feedback is that, you know, sometimes, you know, going query by query and saying shoes and winter coat and this other listing page and this other listing page should all have the same rule that's very duplicative. And so with flexible targeting, we're gonna allow you to, a, pick multiple listings for a given, rule, and b, for queries, use operators like contains. And so, not just shoes, but anything that contains shoes. Winter shoes, summer shoes, men's shoes, women's shoes. All of those queries could be Agentic could be affected by the same rule. And what we really expect this to do for our merchandisers is have them create fewer rules, spend their time more efficiently, and move on to bigger and better things. The second thing we're gonna be doing is introducing this concept of audiences. So audiences in the Merchandising Hub are going to allow you to scope, your rules to specific user populations based off their attributes. And what we're really starting here is kind of these, we call these, like, these request time attributes. So things like the user's geolocation or their device type. All of this is stuff that we can rapidly infer. And we really see these, this first stab used as as, a way of tactically addressing merchandising problems. And I'll give you an example. If you wanna if you have this AI layout of, four products per per row, it's pretty easy to make a product salient by pinning them to the fourth position. And that way you save positions one through three for AI for more relevance. Well, on mobile, position four is pretty low and if especially if you scroll product by product. And so for a lot of for a lot of our customers, pinning to position four on mobile is very ineffective. You're not really increasing the visibility of a product. And so simply saying, on mobile, I pin to position two. On desktop, I position to position four. That that kind of difference in pinning can be a good way to view of using what's coming in audiences. And where we're gonna take it and and where where I think there's gonna be even more kind of value to the business is we're gonna start ingesting user profiles. So a lot of our customers have customer data platforms, whether they built them internally or they've they've bought software. And a lot of times when I talk to merchandisers, what I hear is that this data is very it's AI it sounds cool. It's a nice project. They haven't seen it. They've never used it. We wanna actually put this in their hands. If you have a CDP and you're able or, you know, some kind of segmentation framework internally and you want to be able to put in the hands of your business users, we're gonna be able to start ingesting these user profiles so that we know that a d at caveo dot com is, a gold star user with a very high lifetime value. And then you can then AI target those users with rules based off of that segmentation. So that's gonna add this overlay. And so from at that point, I think we're gonna move to a more strategic position, and we're really gonna be pulling in data that, you know, Coveo doesn't have. We have so much great data about the online experience, but what's happening, you know, in store or in in their kind of registration flow where they self segment. There's so much information there that's really rich and that can help us deliver a better experience. Next one I wanna talk about is spotlighting content. And so content spotlight or spotlight content, is is really gonna help people put in pulling content into their product discovery experiences. One of the first use cases we wanna cover is these kind of, like, inline, product replacement cases where you can then kind of replace a product with some content imagery, maybe a click through link. And so it could be something like, you know, if you're selling hardware and someone, is on a flooring listing page, you might wanna push them towards a flooring buying guide. Right? So what we see from our customers is such fantastic content that AI sometimes isn't really exposed to the user at the right, you know, at the right moment in time. And, man, these purchasing decisions that people are making just they're they're they're missing and we have the information. They just can't assess it. And I think over time, this will complement a lot of the stuff that Ollie talked earlier about, you know, generative experiences coming into commerce. But there's a lot that can be done here. And the reason we're taking on this problem and not saying, hey. Like, why don't CMS solve it? Is that we have, you know, unique, understanding of, the queries that users are entering. Typically, CMSs don't know that someone entered a specific query. And we also can make the product that the layout, you know, look right. Right? We can shift their products in the right way, make sure you don't have these kind of weird edge cases. And so, you know, the really within the scope of, a grid of products or right next to it, that's where kind of you get this overlap with with content management, and we feel that we can actually complement them quite well. So this is something that we're also working on. And that covers it for merchandising. So, just just to summarize, you know, a a lot of, operational needs that are that are that our customers have, we're we think we're making a lot of progress in making that really, really accessible, really easy. We think that, you know, we're extremely committed to great UX and any clean product experience. And over time, we're just making people super efficient at solving these problems, and independent. Now on the integrations front, there's two big ones I wanna talk to you about today. The first is, our our Shopify, application. And so we now have an official app that I'll show you guys in the Shopify, ecosystem that allows you to integrate Caveo, with, AI. And it's pretty cool. It's a pretty fantastic product. So I'll jump right in here. And so this is a a brand new, storefront on Shopify, and I'll show you how you can kinda run this app. And so I have this Caveo AI search app here. I'm gonna link my organization. We've made this super, super simple. And so we'll just call this QA install. Here we go. And we're gonna link it. And right away, this is gonna start we from this point on, we've got a AI of deterministic deterministic link between kind of Shopify and Kaveo. Next step here is to actually, ingest my products. So I've got a few here in, in in in my small Shopify catalog of thirty products. This is not representative of our real customers, but it makes it good for demos. And you could see here that, I can sync these products. And so we have kind of this automated integration between Shopify and Caveo. Now let's jump to a a fully set up storefront here. And so this is, AI of a a broader site. And here you can see here that the third step is to build the search experience on-site. And so you can use two options. One is using our Caveo atomic component library. This is for customers who want to be in the Shopify ecosystem and really leverage liquid a lot. So a lot of drag and drop templated, UX, it makes it, a lot you can get a a much lower cost of ownership doing things this way, but you do sacrifice some some customization. And so, we've kind of seen in kind of our our forays into the Shopify market, various, approaches to how customers use liquid a little, a lot, exclusively. Or the second option is to use headless. You know, this is kind of the the most common way customers, implement Coveo for commerce. And commerce integrates really well with, Shopify's Hydrogen, framework. And so, this kind of leads you to documentation and how to do it, but both are super well supported. The next step then is to track events. And so we can see here all of the the the the events that are key to making our system work, whether it's for ML, whether it's for attribution. We're gonna kinda bring them back here and let you know, are we tracking these things properly? And then the last step is you can actually start managing, or penultimate step is you can actually start managing your search experience here. So here AI got some I actually have a ranking rule created in the merchandising hub. There's this AI or this one way sync that allows you to jump from Shopify into the merchandising hub. And the last step is actually can check-in on my AI models. And so we've really tried to, make a a a clean experience to set up Caveo, get you started on the right foot for if you're coming in from the Shopify ecosystem. Really excited about this. In terms of future future for this, there's still, a couple investments that we're gonna be making to to make this integration work even further. One being AI of better connectivity. So handling, larger catalogs with more real time updates. We want to make AI model creation potentially completely obfuscated and just automatic. And we wanna create deeper links between some of the, the abstractions in Shopify to Coveo. A great example is taking Shopify's collections and turning them into product listings in the, the merchandising hub. All of this is just gonna make the the the experience of working within Shopify and Coveo more seamless, easier, and quicker to set up. So this is kind of a a huge step. And I think for even for our customers who aren't on Shopify, what this has done for us is it's really helped us, kind of build a lot of these abstractions to onboard quicker. So, or or to to to monitor health. And so if you if you have an existing storefront, you're not planning on expanding, a lot of these underlying investments we have in monitoring are gonna AI of benefit and accrue to you. Or let's say, you know, you you like Coveo and you're planning on expanding, internationally or you're planning on acquiring and and and implementing Coveo there. Whatever that might be, maybe these marginal implementations are gonna get easier and easier. AI I think working in this ecosystem has really helped us figure out great leverage points to make that easier. The next thing I wanna talk about, and this is kind of a quicker update, is some of our investments in our SAP integration. So our SAP integration is much longer lived, and has been kind of quite successful and leveraged by our customers. This, specific update is really kind of focused on, larger catalogs and and AI of more international, businesses. So, you know, the SAP customer base is obviously very large, very enterprise focused. And so making sure that, you know, you have these granular controls, for example, for, deciding on how you index multilingual content or, multiregional content from, SAP into Caveo. So these kinds of these kinds of investments have just, leveled up the integration. And for customers who kind of are in these situations, rather than falling back to custom code and implementation, what's there in the SAP integration is gonna let you go faster. And again, to my previous point, as you scale to more markets, as you acquire or grow, you know, these things become marginally easier. So this is kind of a a reflection of our our commitment to to making things easy, to making things simpler and simpler to to to work with us. And, hopefully, that just leaves more time and more energy to the more strategic stuff, that we've discussed before. So, I think that covers the content I wanna cover. Shareen, did you wanna say any words? Yeah. Sure. Hello, everyone. I'm back. Just stay there for a moment. Adi, I just wanted to do a just a quick, plug for a a virtual event that we have coming up next week, April twenty second on an Agentic AI master class. If you're interested in in at all in anything Agentic speaking, it's not specific to commerce. It's across Coveo. We have Sebastien Paque, who's our VP of machine learning at Coveo that's going to be on that event. So I highly recommend if you're interested at all in a Agentic AI that you, sign up for it. And if you can't, attend at the time that's indicated there, obviously, there'll be a recording if you're registered that you can access to after. And with that, you can go to the next AI, and I have some questions that came in. So I hope you guys are ready. So one of the first questions that came in, Ollie, is directed to you because it's about the semantic encoder. So the question was, will the catalog semantic encoder translate language, or does the content need to be in the foreign language? So it will not do automated translation. However, it does allow for search throughout different languages. So let me give you a concrete example. We had a customer, in the book industry in in in Europe, and they actually have books in over twenty seven different languages within their catalog. So when you search for books either in a different language in the search box and the books are in a different language within the catalog, we're still able to retrieve them because of the vector based functionality. So what we do is we actually create a vector out of the query and out of all the content. And since we do so in a multilingual fashion, so you don't need to specify a specific language, we're actually able to handle, translation dynamically. So while we don't do dynamic content translation, your content still needs to exist, AI, in the language that you're searching through. You know, it is very tolerant, if you will, to, you know, both mixing languages within the same query as well as typing queries in different languages. Okay. Understood. So you have to have the at least the catalog, in the different languages in order to understand what the Well, not for the semantic encoder specifically, but to benefit from the full functionality. Right? Because semantics itself is able to handle all languages together. However, for something like a lexical search, this is where languages do need to be specified, so that we can do tokenization properly. Alright. Okay. Now, in terms of, generative, Agentic stuff that we covered today. So, one of the questions came in is, can we add to cart and buy now in CRGA components? So CRGA, for those of you not familiar, is Coveo relevance generative answering. So that top part of, the screen that that came back with the answer, not necessarily the products that were recommended or the the categories that were recommended. So I'm trying to interpret this question. It sounds like they wanna know if you can buy now based on, I guess, the generated categories, which it's a little bit of a stretch there. Yeah. Yeah. I mean, you can't really add categories to cart. However, Kaveo does support you to add add to cart buttons and all that sort of stuff on any component, if you will, returning products. That being said, it needs to be an entity that you can add to the cart. Right? And in this case, we really chose to recommend categories. And, actually, I think that's a relevant that's a relevant point to make, actually. Yeah. So the reason why we are recommending categories in our products is really because generated generated answers tend to be vague in the sense that questions don't always relate to products directly. They can relate to concepts. They can relate to activities. They can relate to troubleshooting. So in this case, by recommending categories, it allows us to paint a broader picture and to ensure that we can actually link customers to relevant products in the end. Recommending a single product requires us to be very opinionated, and and, you know, it has a AI likelihood of being irrelevant for customers. And we can see it actually live today on many many, you know, sites where they're actually leveraging generative assistance and and and that sort of stuff today. Okay. If I'm if I might add to that as well, like, our perspective is that the the the leverage point for generative and commerce is not just a as a as a broad throughout the funnel application. When customers are keyword searching and they're saying, you know, I want the blue air force ones in size ten, they don't need a generative response there. They know what they want. We know how to respond to that query. Where generative is relevant is kind of higher up the funnel when we're talking about shopper education. Typically, this is activity that just happens outside of your storefront. This is activity that happens on a comparison shopping site or whatever. And it's kind of this leaky funnel because they might just go to your competitor or whoever wants to pay the most affiliate marketing. So, kind of using generative to to to to create that perception of a destination AI, to to train your customer, to just come back to your site to ask the questions, that's where we we think it'll have this kind of long term customer impact. And so at that point, is it okay to be vague? Yeah. And, to to add to that as well, we wanna make sure that it's grounded with the rest of your experience too. Right? Because today, we see a lot of shopping assistance being disconnected from search, from recs, from listings. So when you get out of that shopper education journey, you're having to start again. Whereas with Caveo, we want that to be tied in and really well integrated. So that's why we're we're doing it this way. Okay. Absolutely. AI tied back to the intent box. Why is the customer there? Follow is the search done okay. Another question. Is the search done in a conversational way so that follow-up questions are related to the original question? So this Yeah. So I can definitely answer that as well. So, yes, not only can you do conversational in the conversational way where you can actually keep the context of the initial question within follow-up, but we cannot we also have a feature now to recommend follow-up questions to end users as well. So end users have actually both options. Follow-up questions AI actually live on our documentation AI, for instance. So it's somewhere where you can go have a look, if you'd AI. But definitely, and also because the category recommendations are related to the actual answer itself and not the query, we ensure that every time you do a follow-up question, we recontextualize the category recommendations as well to make sure they remain relevant, to the answer being presented to the end user. Okay. Excellent. And I just wanna clarify something that came in through the chat feature and not the q and a. A sleuth was watching all of our screens, and they saw that when AD was, demonstrating the AI by query that, some shoes were showing up. And I know you answered this, Ollie, but I wanted to just go back to it to, you know, make sure that everyone is, aware of how our demo environment is generated and the issues that might occur from that. Yeah. For sure. And I'm really glad that I said something about that initially because if you go have a look at our environment, actually, you'll notice that the descriptions aren't always perfect. Barca and and these environments are always a work in progress, but over half of Kabelo's functionality is grounded using behavioral data, which brings, you know, really good performance on actual production sites with big data, but always face a challenge when working with generated products, descriptions, images, and generated data as well because traffic is automatically generated there. So, definitely something, we'll we'll we'll keep working on, on our end. AI. And there were a few questions that came in around, you know, licensing and, you know, additional costs for different components. And you need to really reach out for to your customer success manager or account manager to find out what's specific to your, environment and what you would need in terms of an add on. But I will say that, you know, the core feature is is is search listings, recommendations, and then anything that's generative will be something that's, an add on or put aside. So but if you have access to Coveo relevance generative answering already, you don't have to reapply that for commerce specifically. If you already have a license to that and then you have your your query set up for generative responses, then you won't have to repurchase that that part. But you need to for your specific, case, you need to Yeah. Reach out to your account manager. Yeah. And on that point, we don't see in in terms of generative experience, for instance, it's a new type of experience. So as Shareen mentioned, it is a new type of query for us in entitlement. That being said, all add ons that we make such as category recommendations now and any new ones in the future will be included within that generative experience that we're selling. Right? So we're not selling those piecewise by the feature. We're actually just selling the experience itself. Excellent. Alright. The same goes for the merchandising hub. So if you have access to search listings, recommendations today, but are using the merchandising hub. There's no entitlement gap. It's there for you. Okay. Thanks, AJ. And with that, I see that we're at the end of time. So I wanna thank everybody again for spending the last forty five minutes with us. I hope you learned something. I always learn something when the product team is presenting. So thank you, AD. Thank you, Ali, and thank you to everyone for joining us today, and have a good rest of your day. Watch for the recording in your inbox. Thank you so much. Thanks, everyone.