Alright. So let's get started. So before jumping in directly into our agenda for today, I wanted to highlight that COveo does go beyond commerce, and it's in fact a composable AI search and generative Experience Platform, and it's being used across many, many use cases. It supports semantic search, AI recommendations, generative answering, and of course personalized product discovery. So we have native connectors as well as custom connector framework to be able to connect to any index type. In order to, index content as well as product data. Our unified index also does both keyword search and semantic search with the ability of keeping vectors or embeddings directly within the index. And you'll hear a bit more about that from the team today, and Simon. However, a big call out here is that, of course, it's AI across the core. So we have multiple AI models around machine learning, deep learning, and now also the ability to deal with a generative AI and large language models. And we also have, of course, you know, a set of API frameworks, native integrations to allow us tomorrow. Fluidly, I would say, integrate with the various use cases we support. So website calmer service workplace. You see those all along the top hand, the top part of the screen. And we also have, of course, use case extensions like our merchandising hub that we have for commerce. So all that is a little bit about our platform. So, today for our commerce release, we're going to be focusing on a subset of the latest and, greatest functionalities of the platform. So most of those are marked off here. You see with the the blue diamonds. And in terms of an agenda, First, we're gonna cover the latest and greatest in AI and machine learning. So we'll do a focus on that. Then some of the newer merchandising functionality then we're going to finish it off with data and reporting and some connectivity updates. And with that, I'm going to pass it over to Simon to kick us off. Here we go. Alright. So for AI and machine learning, so we use, within an AI company, since our very much very beginning or at least our beginning in the cloud. We have developed several models for several use cases, ranging from commerce to self-service or which workplace and website. But no one will cover some of our newer, machine learning models, that are used through that journey. I will take a bit more time to introduce generative answering and mostly where, we plan to use it into the shopper journey. So it has been used so far, with, actually, quite impressive results, into the self-service use case. But we are introducing it now into more of the engagement education part of the shopping journey. It's really what we are aiming at. So here, you know, you can see a bit of a different would say more zoomed in into the e commerce journey, where we would put generative AI power powering knowledge answering at the very start into the research education phase. So mostly, you know, if you are kind of looking at, okay, where can I use GenAI in e commerce? Can be using several, several flavors, several areas, In the case of product discovery, we see it at bringing a lot of value very early on. Actually, you know, when you're searching, for the best product, for example. One of the first thing you'll do before you even reach, the e commerce solution, application, website, will be to search online. A lot of people still, for example, watch YouTube video or search for, you know, what are the best cell phone in twenty twenty three. And you'll end up usually on the blog, and that blog will most likely redirect you to, Amazon, Walmart sometimes, depending, you know, on the country. But most of the time, you know, it end up into that kind of Amazon. And we want to kind of I backed a little bit, with this and bring that expertise, directly into your e commerce solution so that you can start a journey there and avoid to bounce off. Obviously, you know, the questions that are expected. We're targeting, both b to b and b to see. So more at BDC questions, you know, I need to build something. I want to buy into a DIY store. You know, where do I need to buy? On the b to b front, obviously, some more complex questions, you know, I am on a project. I need this specific product. What should I buy with it? Is there something I should know about before applying it? So, obviously, you know, we are a search company at our core. So what we offer is mostly a search experience from predictive query suggestion to, ranking personalized result ranking, as well as personalized facet order. From multiple data source, while, you know, a more typical generative AI solution, we'll look more at, you know, longer queries, natural language queries, and we'll, and we'll offer, you know, longer responses responses are usually quite rich, and that will contain, you know, a lot of different attributes, that will explain a little bit more around the why you should choose a product. And with this, with these kind of two ingredients, what Cavell has done, you know, since, actually, before even the chat GPU preys are because we've been working with LLM for, case deflection smart snippet for quite a while. We have integrated, embeddings and vectors, which are used to power semantic search, which are used to power LMs engine AI, directly into the index. So usually, the way that it works is, you know, the search engine as an index with secure connectivity to bring the content into that that index research, and then you have your vector database that is used to store your, embeddings. In our case, we combine everything together. So mostly we bring vectors directly into the index without necessarily a use for a database. And we have product embeddings that happen directly into the index. Product and word embeddings, that are used mostly for long sentences and to understand what is the connection, what is the semantics connection between keywords, and what is the connection between products, mostly through their usage. So we've been using this for quite a while to do, to do product vector search, which Benoit will cover a little bit later. Which is used mostly for intent detection. And we use it also for, word embeddings to understand how words correlate to each other. And create these long sentences or make sense of these long sentences. So we are offering, relevant generative answering for commerce so it is now open into early access for commerce customers. So it was offered before and more to our service line of business. We have several customers who have joined this program, a, product release, sorry, a PR was released recently on zero using it, for their self-service portal. They have seen great result when it comes to customer satisfaction and case inflection, and overall improvement in CSAT. We do We do now open it for a commerce customer, hoping, obviously, but quite confident that it will yield, great result when it comes to bounce rate or even overall top line revenues. So the timeline for it, if you decide to join today, contact your CSM, or contact your, your account executive you can join our early access, program, early adopter program, I should say, at a discounted cost, to have access to the feature. The goal live will be at around, mid December. For the price to come back to a normal and for the, model to be available, into the admin console. Before that, it will be activated by our team, and will be obviously totally tested. This service comes with, a professional services engagement, which is included in the price, as well as a BVR, program. So we will make sure to test the outcomes that are important for you. So we have a set of predefined metrics, but we are absolutely open on metrics that you might want to push forward. The type of customer, obviously, that we are targeting, are those that have rich content. This is, quite important. You should have mentioned it before, what's gonna drive that education journey is obviously rich content. So if you have a marketing team that produces a lot of blogs or you are more on the b to b front. You have tech sheets, or anything related buying guide, that are related to the content, to the products that you sell. This is really where these, models will bring a lot a lot of value. So we are targeting, obviously, customers who have rich content who want to push that expertise forward and compete with the giant on the market using that expertise. And on this, that's it for generative answering. If you want to join the program, please, reach out to your CSM account executive. I know it will be a pleasure to get you into the early adopter program. And on that, I'll give it to Benoit. Thank you, Simon. If you allow me, I'll share my own screen. Gen AI, you said the the high per bar pretty high. Thanks thanks for that. We we still have a few, more incremental, improvements that that we've made to our our core AI functionality. So we'll present three of them, starting with, a quick reminder on personalization as you go. Because as Simon mentioned a few times, vectors, it's it's like if vectors are everywhere now, So, let me try to be a bit more specific about about this one. Personalization as you go is, built with, vectors, but not to confuse with the more the LMM space and and the words embeddings and so on. This is all built from, the shopper session vectors. So how people interact with the different products and create those journeys, create those those vectors in in your storefront. And by accumulating all of these sessions, this is how we can, build our our under standing or our representation of of the the product space. So all of the product, from your catalog, and and build sort of a map of the relationships between these products from how people interact with the the or or create a sequence of how they interact with the different products. And Like the this leverages the whole power of of big data, sort of this of the the flavor of the month, buzzword, but but it's it's very powerful, even more powerful than the information contained in the the product metadata itself because of this volume of data and of of the all the different shoppers interaction with with these, products. And we use that basically when a new shopper comes in, even if he's anonymous, and that's a whole power of it because most most shoppers are anonymous, is use all this previous successful journeys that are similar to the path that this new shopper is taking and try to predict which outcome is is most likely to be successful next. And this is where the the intent detection comes in, what we call intent is what's that next move? And and how can it, how can we maximize the chances that it it's successful? And we combine this layer of intent detection to the other AI algorithms that that are in place on your website that are more, looking at the overall popularity and so on. So a few layers of AI to to give the final ranking and the final, score through to all these these products. That's a bit what we see on the the right here in the image. On the top and the bottom, you see two customer coming in with different context. One is a golfer in in Florida, and at the top, we have someone shopping for winter gear, in in New York City. But they both search for gloves, and they get different products because we understand their intent. We understand where they're coming from, and we match this with all the previously success full events of people that had similar context and and behavior on the site. And that this this foundation of behavioral data and and vectors applies to, that's what we see here at the bottom. All three of the main product discovery components that could deal powers on your website starting from the search box with the predictive query suggestions when someone enters, or starts typing in your search box, intent to where product re ranking will be on all your result pages and session based product recommendation will be all those those parasels, that could be, in in multiple spots on on your, storefront. So it powers all three. Now why this reminder about, personalization as you grow it? Because there's a problem with the behavioral data that we're we we we fixed. And it's that the traffic is really concentrated, like, the popular products on your website, they they get most of the views and so on. And we already have algorithms that focus on the popularity of products, but now it's really to detect intent. So what we're trying to maximize is is the coverage. And I'll introduce this this chart here where you see that curve where, like, the the up viewed product get most of the traffic, and there's a big percentage of the catalog that that gets very little traffic and views and interactions from clients. So this is where we make the difference between what we consider warm products and colder products or even frozen product if they had zero interactions at all. But we want to increase that coverage So how do we do that? We do that by, using or combining both the the metadata of the products and the descriptions and the images and compare them to these warmer products. So the the warmer products, they build the the background of the map, and then we we come in with all the other products and fill the voids. By comparing the similarities with products. So this way, we have a very complete map. And, in general, we will see numbers going from around that, like, thirty percent mark where products have enough traffic to to gather a reliable vector and increase it to the the ninety nine like, some some very high number where it's just the the small turnover of products that that maybe doesn't have, vectors. So it's really a a coverage of objective. And yeah, that's what we see here. We would come in and whether it's a new product or what we call long tail product, we would, introduce them in, in your, product, catalog representation, the space of of these vectors. We have a few results from from this already. But quick reminder first on, intent aware product ranking, because that's where most of the uplifts comes from, and we see it on on the commerce attribution metrics AOV RPD conversion rate, massive increases. If if you look carefully at these numbers, average order value increases by in in that range of one point five percent to five point five percent of the different clients that we've tested RPD in uplift, around four point four percent conversion rate, two point two percent. And that's across all the the session. So insane numbers, on top of that now, with the the this cold start solution, brings is a small, improvement in the relevance metric. So you sit with the the click through rate metrics and the click run, those two metrics improve. And this way, you have a much better coverage and you don't few more of the traffic toward popular data. And in all cases, all three cases at which we've tested it, we've reached at that ninety nine percent threshold of of catalog coverage and and, suggestions or visibility to these these products. So, very positive. If you're curious to learn more, we have a bit of documentation on this. Next quickly, we have two more, features to cover. One is about those no result pages as you see up there, oops, we don't have any, content for cryax. We have a few features already that tackle this problem and try to reduce the number of null results. It goes from, like, did you mean, attempt correct, questions, which means we will detect that there's more result for a similar query and suggest the user to change its query and and then go to another query. And, there's other ways to to do this as well in Kavil, but now we're trying to make it more robust. And how we're doing that is by leveraging existing ML models So you're all familiar with what query suggest is doing. We just covered it previously. There's all all all also now, the predictive query suggest that includes some personalization, as you go. So even more powerful than before, and we leverage that for, increasing the range of null result pages that we we avoid. So, when a user doesn't select one of those query suggestion and still enters a query that leads to no result. Now the behavior will be that it falls back on that top suggestion that appeared in the list when he interacted last with the search box and will default to this content instead. While notifying the user. And so showing less null result pages more robust and leveraging the existing ML and all the power that we encapsulated into the the query suggest model that you you're already using if if you're, if you're powered by by Kaveo. So really good stuff. We have few, results yet from this. We have a few, early access customers that have done this, and we've We've seen positive results, around ten percent of reduction in total null result pages. So, what I mean by that is let's say you had ten thousand null result pages per month, now it it goes down to nine thousand and not, like, eleven percent to one percent. Not there yet, but, it's quite a a massive, improvement in no results. So showing more products and keeping the the shopper engaged. Lastly, an upcoming feature, around listing pages and optimization of of our AI models again. So listing pages are a slightly different beast than search, there's already some intent that is, part of the the action of the shopper of of reaching that page of navigating to that page. And the products are already scope. Right? So we don't have, like, a keyword to evaluate relevance compared to the whole catalog and so on. So the process is is different. The the scope is more narrow. So we have to adapt a bit our AI to this reality. And there's two internal initiatives that combine to make this possible. One is around, building this specialized model for listing pages only. So decouple it from the the other search experiences. That's for one. And also, the second initiative is trying to introduce more criterias to evaluate, the the what to optimize. So you see two examples up there. R p v and profitability. So what we mean by profitability is really like the the margin of each individual project each individual products. Sorry. And the overall profitability of a session as well when those products combine, but that's an another milestone in in the this research project. So combining those two initiatives to really, offer something specialized now for listing pages, and we don't present it today, but we also have another feature that goes along with this, which is the PLP manager, which offers a a new UI to manage these listing pages. So you can see a trend we're trying to specialize the two experiences as much as possible to to optimize the performance of each and adapt it to the the context. Again, we're listing pages. There's less of this intent. So the the previous models that we've seen for personalization as you go, where they focus on intent, Well, this is less true for, listing pages. So we're trying to tweak this and make it more specialized. It's early, but, we accept, registry for, early access to this. So don't hesitate to to contact us. And, yeah, that's that's it on my side. Jason, your next. I'll explain what. I'll just take home this screenshot. Great. So I'll be taking you through some exciting updates across merchandising dating reporting connected to integration all the way through to security and infrastructure. So let's get started on the merchandise side. So we've helped our businesses, business users drive a ton of success with relevant product messaging through badging on product details pages. And we've seen our clients regularly improve their revenue per visitor metric by one to three percent by deploying iterating on tactics like social proof. So what we've done is extend this capability to multiple product areas on your site from areas such as listing, search results, fast and recommendations carousels. So we announced this capability in our last, new in webinar. When we were in early access, but now this is a generally available capability. So we wanna go through the up to date, flow. And share what we've learned from our users leveraging this feature. So if I head over here to our merchandising hub app, and we want to get started on a new badging campaign. So head over here to our badging flow. Let this start step of the process. I want to select the area of my site. So I want to target an area where there are lots of products, like my product listings page, where a developer has set up. A PLP image badge in placement for me. So if I click through here and the first step that I need to do is to add a badge, I'm gonna go through here and create a best practice, bad, just social proof. I have the option here to only show this to a specific audience, of visitors, but actually, for this case, I want to target everyone. So I'll go over ahead with that And with this new feature of targeting multiple products, a new capability that we've guided here is the ability to control badge and visibility. So you'll notice here at the top you can either control how many badges are shown on the listings page by a threshold or by percentage. To make sure that your badge is still sound out to your shoppers. So I'm gonna go ahead here and select limit of three. And to select to set up my social proof badge here, I'm gonna do this based on purchases, and I only wanna show this message when there's been more than twenty perches of the product over the last twenty four hours. So once you've set up your rules and your strategy behind your badge, The next step is to tailor the design and how you were gonna present this to your shoppers. So one learning that we've we've we've heard from my users is combining badging wording with social proof counts helps to give that credibility and drives a lot more value than just ambiguous messaging such as editors picks on its own. So I'm gonna combine a message like best seller with the number of purchases that a product has had. And, so you can see here the preview as I'm typing, So best of luck, twenty one customers, I'll put purchased this today. And then what I'll do here is I can also if I want to upload an image to go with my badge. So I'll upload a fire icon there. To show that it's, a popular product. And this is another great, feature that we've introduced for this capability is the the way for you to work with your developer and actually build a placement. That's not just gonna support your first use case for badging. For future use cases as well. So this is where you can add additional custom fields to really configure, your campaign. So in this case, I want to make sure that I can control the text color and the badge and the background color. But other good use cases we've seen here is being able to control things like fade in and fade out times of badges being able to control that as a setting or even be able to add a a link and make badges clickable that takes the the shopper to a dedicated listeners page that has other products as part of that promotion as an example as well. So I'll go ahead and create that badge. And what we'd also recommend as well with badging is to really start simple with use case like this. But then thanks to the ease of iteration with the with our campaign flow, you can really start to test and build your way up. By adding complexity and dry making sure you're driving your value by doing so. So within our flow, you can replace what visitors are shown by default on the site known as the control. And we can copy over what we have here. And what we're doing here is essentially setting up a head to head test. So within here, we can control things like be able to test and find what's the sweet spot of the threshold of social proof badges that actually show in this listings page, or maybe I can change, the the count and that this badge is shown to. But where we've seen also another value is users being able to actually pay around the wording. And this is a really great way to drive more revenue through your badge by trying to create more urgency through the wording. So be on to test things like best seller against something like don't miss out is is a really good way to see if you can drive more value, through your badges. But I'm gonna return back to my best practice setup here. And another thing that you can do in this flow with a collection of lots of products, you might want you might also want to incrementally test the impact of adding other badge styles to an area like your listings page. So I'm gonna go ahead here, add another type of badge. And this time, what I wanna do is call out to my shoppers products that stand out because of their sustainability features. So one way that we can do this is if you have a specific list of products, and you now have that capability, but for the purposes of this demo, I'm going to target some specific product IDs that I know have that feature. So I can go here enter my product IDs, and target these products like so. Save that on the next, I'm gonna call up that these products is sustainable, and upload an icon there as well. And save this. So we've now got two badging strategies that can fire on our listings pages. The next key step to do here is actually to go ahead and preview before publishing a change like this to make sure it all looks good on the site. So we'll go through and preview through to our demo site here. You can see on this specific page how our best seller badges is firing on the relevant products in the page. And if we head to a popular page, like the surfboards page here, we can see, and this best seller badge also mixed in with this sustainable badge as well. So that was the demo for multi product badging. And next up, we have, our self serve insight dashboards. So at Koveo, we collect lots of important signals. About how our shoppers are engaging with product discovery, from which search terms of popular to products with the best kick through rate on a listings page, to interactions on a recommendations carousel. Now with this feature, what we've done is we've taken all of that granular raw event data, and we've joined them up in a model that structures your data in a way that makes it easier to analyze the performance of your conveyor product discovery solutions. Such as things like being able to see which search three terms are driving the most revenue. You can access, analyze, and visualize these insights through your conveyor provided Snowflake reader account. So let me show you how we can go about doing that. So if we jump over here, I'm now the Snowflake interface. And what you can see here is I'm starting to to write a new query, but let's break down what's happening, on this screen here. So on the left, you can see We have our call model. This is where we model all of those events, and we have some folders here like common, which give here different types of views that have been created. So as an example, insights is a really is a really useful view here, because this gives me information about every time conveyor returned, a ranked list of products shown to a visitor. I also have commerce specific views. So things like carts where I can see by user who interacted, and added products to their cart. I can go into the cart items, and this gives me information about the exact products that are added from the SKU, the price, and so on. And also my transactions views that allows me to calculate really important business metrics such as the number of purchases that might have happened. So looking to do hearing is write a query to analyze the state and trying to get more insights by taking information from the transactions view dining that up to cart items all the way to insights so I can attribute a specific transaction all the way through to the conveyor solution that, that led to that transaction happening. So when I run that drill within Snowflake, you can see you get the dates within here by day and biker value solution, the total revenue being driven there. Also, in this interface, you can go step further and start to visualize this data. So you can go to the chart option here and I have lots of different options to be able to visualize this data. I can come in here and add conveyor solution to my chart and I wanna make sure that I'm plotting this as a series here. So now what I can display is how each solution is driving revenue for my business over time. I've also got some display options here where I can label this chart as well. Now this is just one example of an insight I can add this to a collection and create a dashboard that will show me things like revenue by solution, AOV by solution, and even the revenue per visitor by search terms as well. For more guidance, any information on how to go about doing this, can head over to our docs site within our conveyor for commerce section. Another reporting we give you a tutorial of how to go about creating these custom dashboards. And ideas of insights you can pull. So if I wanted to get more information about my revenue per visitor, I'm giving a few ideas of the type of analysis that I can do, and if I check out that analysis, we also give you Andy, SQL certificates to help you get started as well. Now moving on to data and reporting, there's been a a lot of focus around data health. Now data from event tracking is critical, activated. It's through that that we're able to fuel our AI and it's through the data tracking and reporting, we can give those sorts of insights that we just saw there to show how today's product discovery solutions are affecting user behavior. And business metrics. We've upgraded the data health dashboard, which aims at helping our clients validate their event tracking initially and also post implementation, making sure they maintain that data quality over time. So Koveo can so Koveo AI can continuously perform at its best Let's have a quick sneak peek of what that looks like. So I'll head over to and our demo organization here into the data help you here. And as soon as I land here, I'm given an overview for my organization. Now I might have lots of different websites set up in my organization, and I can drill down to a specific one through here. So let's drill down to and my sports site as an example. And what I can see over here is a data health score. Now this ranges from zero to one hundred. This is our demo site after all. So I'm quite happy to see that it's at that higher end. But this gives you a sort of indication of where you're currently live. And this score is a measure of of the rules we have set up is your data passing, the quality that's needed. And it's really influenced by Also, for each rule that we set up in here when we're checking the validity of your events, how important. So this is a really key thing to look out for is making sure that you have a look at and keeping those critical errors down to zero percent. And this allows you, of a way to prioritize that as well. So something to note here is things like data helps or can fluctuate over time that can either be changes to your site or presentation, but also as we start to work with our customers and learn about specific edge cases that might be happening in their data tracking, we may look to introduce more rules here. That, can also change the the data health score with the addition of these numerals. The other key thing to note here is what this this view is giving us is the validity of the events that have been sent to Kadeo. But of course, there could be journeys on your site or your application where we aren't receiving any events. And this dashboard won't be able to inform you that, which is why it's really important to make sure you have the comprehensive validation process to make sure all of those journeys are tracked. So for more information in guidance now, you can head over to our documentation site, but we give more hints and, and best practices and what a really good data validation process looks like. But back to the view here, you can see that of the rules of how we're checking the validity of events, we group them into different categories here. There's an example, the group has event syntax. So this is checking for every event that we receive is the content of that event valid, for that specific event type. Are we collecting the sort of information that we need to drive our now and reporting. You can click through and get more information of what may be failing and what you're doing really well at at passing as well. And and you can always expand one of these, get more information about what the cause impacting solution is, and an example of an event that is failing as well. But we can see here that we need to dig a bit deep into our event browser to to know the specific reason that the event payload is failing. This is where this tool comes in, which is really good. Either if you want to deep dive and look at the a specific event that's being sent in your site, what the content of a valid event looks like versus an invalid event, and you can go through here and filter by different event types and so on. If we wanna go in here and check our invalid events and to address that issue we we saw there, we can then go through and deep dive into a specific event. So for our car tab event, we're being flagged here that we're missing a parameter. See you, which is the currency parameter that's needed on this event. So that's the information I need. I can go ahead, and address that specific issue. Full connectivity and integration, we've invested in our Salesforce b to c integration, So we're excited to announce the release of a cartridge, which makes it easier to index your product data to power your Caveo product discovery solutions. This connector this cartridge, sorry, also provides UI components for search to help you get started on building the design of your new search interface on your Salesforce powered store front. So let's take a sneak peek of how this integration fits into the Salesforce commerce cloud platform. Once you've installed the cartridge. So if I head over to the business manager, on my side here, I'll now quickly just log in. There we go. Good thing that we've got the the security checks in there. So once you've got in within your demo site, and with when you're in your business manager and you've installed that conveyor cartridge you can expect to find a Caveo configuration section, which allows you to link, your Salesforce powered store with your your Caveo and organization. So things from, like, the API keys that are being used to the organization ID and so on, you can make that connection. For more information on how to check this all out, we've got our documentation site, that really takes you through step by step how to go through, and set this all up. So you can then end up with a Salesforce powered store front. Here's an example here, of our search results page, where we're using conveyor to power this search results page from the filters that you're seeing within here displaying development products from my search query, and also being able to do really nice features here such as product grouping, where I'm grouping very, similar products but of different columns through color swatches there as well. So finally, we've got some great news on the security funds and also on the infrastructure side as well. So we're proud to announce that we are twenty seven thousand and one certified. This certification gives our customers confidence that covid his to the internationally recognized security stand standards. It it goes to show, that we have measures in place to really protect our customer's data, from software infrastructure, development best practices, and information management processes. And we also have, another feature to call out here on the infrastructure side known as active. So at Kaveo, we have a strong track record of search uptime, and we've hit our commitment of at least ninety nine point nine nine percent. Dot time in the last four years. However, if issues were to occur in a region, we want to be able to offer extra assurances to our clients that need it. This is now generally available and active active provides enhanced infrastructure resiliency for product organizations whose primary deployment region is in the US. So with active active, you can see here Koveo distributes the load between two US regions that work in pair with each other. So if an issue does occur in one of the regions, as here in this example on the right, Kaveo automatically reroutes traffic to the healthy region to ensure your implementation continues running. This gives clients that five nine reliability, which is equivalent to less than six minutes downtime per year. This is a a premium uptime and that can be purchased and added to your account. If this is something that you're interested in, you can contact your accounting for more details. And that's everything from us, and I think we'll open now to any q and a. We shall. Thank you, Jason. You can leave it right on that slide. So we did have a couple of questions that came in. I'm going to start with the first one. I think this is more of something you can answer, Simon. Surround generative answering or generative AI, is there a limit to the amount of content they can index or use So I guess it's the scope of it. And Yep. So for the early access, we have a limit in the amount of vectors. So, obviously, that doesn't tell much. But let's say, it's under a million document. More or less depending on the size of the document. So bigger document will create more vector, more, like, bigger embeddings. So if you have, for example, you know, hundred page PDF, you know, and that's pretty much all of your content is gonna be less than, than a million. Otherwise, if you have very short content, up to up to a million. Okay. So it's a mix of the number of documents and the size of those ones. Yeah. Exactly. Exactly. Okay. Perfect. Thanks. And there's a question that came in that I saw you answered, but maybe others, are interested in it. So we showed an example of standard products, but How does the system support something like event tickets and subscriptions? So can aspects of a one time event be entered to provide AI recommendations to the customer? So maybe you wanna repeat your answer as well. Yep. I, yeah, answered that one. So I will I will just repeat it for everyone. So you can always and behavior all custom context. If you have a a user that, for example, log in, so a user where you have past behavioral data, like purchases in this case, past purchase, you can send this as context to the Cavell models and it can all be contextualized that way. If you have more anonymous user, then things that happen on the site can be sent also as context Obviously, there's always a balance between the amount of context that you send and how much you give to, or or how free you let the model be. So you always have to be careful. You know, the the goal of our deep learning model is to do one to one personalization. If you want to add the segment on top of this. Just make sure you stay within a certain reasonable amount, you know, six, seven, ten segments, but not two hundred, because then you know, what's the point of having a one to one engine if you if you start to to create a lot of segments? Yeah. Interesting. I was at an event this week that was focused on b to b and not kind of question came up because you have customers that have very specific catalogs or reordering patterns. So how much do you wanna scope them into that kind of set of your catalog to ease the journey while at the same time exposing them to maybe other products that they're not buying or that they may need that they haven't purchased in the past. Right? So it's a balance there. Yep. Okay. Perfect. There was a question that came in around having a need for Snowflake license to be able to create those reports. And I think the answer is it comes with Koveo. If once you have Koveo, you have a reader access to Snowflake, so there's no licensing issues around that. And one that came for you, Jason, around setting up social proof badge across different categories. So example, if you're where the purchasing threshold can vary, I guess, because you're looking at different, yeah, you're looking at different categories of of products. Yeah. That's a really good question, especially if you're in the retail. That's that was diverse range from clothing all the way to expensive TVs. You can have different thresholds. There's a few ways you can tackle that within, our merch closing file. So one way that you can do that is actually take on a fallback strategy So you could set up your purchase threshold only for those products that meet, a certain amount being bought. You can then fall back to, a certain number of add to baskets or views as well, and use those other types of badging, or another way that you can do is through the audiences as well so you could set up a social proof badge to be shown for visitors on a specific category page and set your own threshold that you want for that category page as well. So that's two different ways you can cope up that. Okay. Thanks much. So I'm very conscious of the time. I think we've gone over just a little bit. I realized we need to wrap up. Thank you, Simon, Ben, and Jason, and thank you to those who joined us today for a new in Coveo. If you're a current customer, shout out that we have the twenty twenty three Covayo relevance awards, which are now open for applications. So, if you've you're adopting an innovating the use of Cobeo and AI, then get your application in. There's always a very prestigious award you can, get And also for those that might have, if you move I think there's one more slide here. If you might have multiple, COVIDO use cases, or simply interested in hearing about, you know, other new features from other areas than tune into the known Coveo to their service and workplace happening next week. One is on the Tuesday, October twenty fourth, and the other one is on, the Wednesday, the twenty fifth, and you can register via our website. Again, thank you for joining us today, and have a good rest of your day, everyone. Thank you. Excellent.