Hi, everyone. My name is Kerri-Ann Beach. I'm on the Product Marketing team here at Coveo. I am joined by Mathieu Lavoie-Sabourin from the Product team. And we are super excited to talk to you today about conversational search. What is it? What does it take to make it work? And of course, we're actually gonna see it in action. Of course, I'm talking about a live demo. Looking forward to it. Should be fun. So let's dive in. So I wanna start off by talking a little bit about what's happening in the market. We know that customer expectations are changing. They're not just comparing you to your competitors anymore. They're actually comparing you to ChatGPT, to Clode, to Google AI overview, to all of these tools that they use every single day. They're bringing those same expectations to your site. What's interesting is it's not just their expectations is changing, it's actually their behaviors. Indeed. Gartner's actually seen that fifty one percent have changed their research habits because of Gen AI. And seventy percent of those have actually changed the way that they're phrasing queries. So they're asking more longer form question based questions. Yeah. And we've seen this in real life with some of our customers who have deployed Generative Answering on top of their search deployments. Who suddenly, not overnight, but suddenly see changes in behaviors with their users where they'll slowly start to ask longer and longer and longer questions as opposed to your traditional search queries. Yeah. It's super interesting. So you know their expectations are changing, their behaviors are changing. So this is really something that you can't avoid. And because of that, we're seeing search kind of follow suit. Right? Search is in yet another stage of transformation. So we've seen it go from results to answers and now we're on full blown conversational discovery where they're asking questions the same way they would, you know, to a to a person. Like us having a conversation. Like us right now they wanna follow-up. They wanted to remember the context and keep going. So let's look at an example. Say someone comes to your site and they ask a very long complex type of question. It's not just keyword based. So maybe they're asking about buying a first time home, what are the mortgage options that are available? Maybe it's something in manufacturing, where they're looking for services, facilities, regulatory support. The answer to this question doesn't lie only in one place. It doesn't even only lie in one team. It's actually going to be found across multiple different systems, multiple different teams and be in a ton of different formats. So you have all of the content but if no one can access it or they're only getting bits and pieces of the answer, all of that value is essentially gonna fall to zero. And suddenly it lives in multiple repositories. It's also that the search system itself has to know that the answer comes from multiple repositories. Has to decipher multiple intents from a single query, make sense of it and actually go out and find the answer where it might live. Exactly. So not only is the answer across all of these repositories but you also have to consider the different types of people who are coming looking for different answers. Different intents. Yeah. Maybe it's a prospect, maybe it's a customer, it could be someone who's super technical, it could be someone in marketing. You need to be able to understand who that person is and what they want. So the question is, if you get this type of question on your site, how would it react right now? Would you be able to search across all of those different repositories, understand who's asking question, and then serve up the right type of answer. This is really where search is heading. And we understand that not everyone is here at this precise moment. Maybe you're kinda starting smaller. Maybe you're still unifying search to have a consistent experience across all of your sites. Maybe you're just starting to get into generative answering. But ultimately when you are making these types of decisions about a search solution, you need to know that you're going to be able to meet this kind of criteria in the Involve towards it. Exactly. Like call call, walk, run. Yeah. That's exactly it. So the reality is most actually aren't ready for this. We conducted a survey called the State of Websites and we actually found that seventy eight percent of leaders said that their search was good. But then once we dug a little bit deeper, we asked them to describe it or to talk about the challenges, we actually found that they were drowning in manual tuning. They had static rules. They didn't have any personalization. They weren't able to scale. Or my personal favorite, it feels like the nineties. So there's really a lot of effort that's going into maintaining these kind of good enough type of experiences. And now that we've gotten into conversational with all of these new types of things that people expect from the website, that's just no longer cutting it. Won't scale. It won't scale and it won't meet the expectations. So we do know that the ambition is there. We saw that seventy two percent of businesses are actively exploring conversational AI. But the ambition is outpacing the readiness. They're building AI on top of platforms that were not created to support it. And as you scale across all of these different domains, multi site language location, you're really gonna see the pains there. Definitely. We've seen this with some of our customers who are really ambitious and they want to go straight into conversational. But the reality is they don't even have Generative Answering on their website. We've seen other customers as well who want to do this on their own, who think they have all the tools and some of the tools are available out there but it takes a lot to develop an end to end solution that will work, that will scale, that won't require a lot of effort. So some have tried this themselves and then come back to us and asked us to actually help them make this something that scales at the enterprise level. It really isn't that easy. For sure. And a lot of times these are kind of issues that you only run into once it's too late. That you're not really seeing the manual tuning and all of that effort that goes into maintaining these type of experiences. And you know, the costs are real here. Right? There's a lot that's actually on the line, for you as a business. Every time that someone's coming to your site and they're not able to find, the result or an answer, maybe they get no results or honestly, even worse, maybe they get irrelevant results or answers, they're gonna leave. And that's what shows up in, you know, prospects leaving, it shows up in churn. And if your competitor is able to provide that experience, then that is kind of one new prospect that you've just given to them. Indeed. Indeed. We've seen this across our customers that they feel that they're like falling behind if they're not doing what their competitors are doing. Because what your competitors what your customers, I'm sorry, will see on your brand on your website will differentiate you from others. Right? If they see that you're falling behind, well this may affect the way your brand is perceived. And there's really real costs that are on the line for your business here. Right? So every time that someone comes to your site and they maybe hit dead ends, they get no results or even worse they get irrelevant results and answers. They're gonna take their business elsewhere. They're gonna go to your competitor perhaps. So that's gonna show up in lost prospects, maybe even customer churn. Indeed. You know, what our customers are telling us is they don't wanna fall behind. They don't wanna be perceived differently than their competitors because what your users will see on your website is going to echo and is going to make them see this as who your brand is, who you are and how you differentiate yourself from others. So if they see that your website and the experience that they're having with your business is falling behind, this will affect their perception of your brand. For sure. And to take that one step further when you talk about generative answering, obviously a wrong answer is going to damage trust. But then there could also be significant business risk if you're in like a highly regulated industry. So there's definitely a lot of guardrails that you wanna be thinking about as well. There's also missed opportunity. Right? So if your search is only going to be showing what's in that one repository that you're searching across, you're only bringing a little portion of your organization forward. So maybe you're not able to cross sell a different product. Maybe you're not able to deepen knowledge, help people find things that they didn't know that they needed. You're just really leaving a lot of money on the table. The other piece that you have too is call centers. Right? So every time that someone can't answer something on your site, they're gonna call in and then that's gonna be a much higher cost as well. So real cost. Real like quantifiable cost that can occur from having a poor experience. Exactly. So these are really not nice to haves and as we were saying before, these good enough search experience can actually be costing you quite a bit. So there's no such thing as, you know, a free lunch. That's fine. So what do we do with all of this? I think it's just about understanding what it is that you're trying to build for and having the right foundation to get there. So if we're talking about creating these digital experiences that your visitors are now coming to expect and also that your teams need, and the support that they need to create those experiences, kinda boils down into three main categories. So first we want to unify search across all of your your your sources, your content. We need to be able to intelligently learn and adapt and like we're saying not drown in all of that manual tuning to really Indeed. Keep up to that. And then The scale. The scale. Exactly. Really important piece. And then lastly, being able to answer and then follow-up and hold the context. So really and engage. So really just kind of walk, crawl, run. Having a solid foundation of which you can build all of it. Awesome. I guess it's good time to have a look at it now. I think we should. Let's go. Awesome. So let's look at those foundational pieces that we just discussed with Carrie Anne. Let's we're gonna look at essentially how the magic happens behind the scene. How do you bring all this content in into unified index to power those search experiences. But then from there, how do you then evolve those to actually scale, to become better and better over time, and then move into generative answering and conversational? So to start the demo, we're going to head to the Coveo platform. We won't dive into the entire platform today, we're going to look at the sources section. This is really important and I love to start showing this first because this is really where the magic happens. Where all the content comes in to be made available in a single place. So we're now Barca Group Production or Barca Group is a fake company that we love to demo. They sell surfboards, both accessories, both GPS. They have both software and hardware. They do a bunch of things which is very convenient to them with different things. And this org here of Coveo is going be able to power multiple use cases from the same index by bringing all the content that exists across all different repositories, CMS, products, sources of knowledge, articles, tech document into a single place. Whether you're using Adobe, WordPress, Contentful, Optimizely, we're able to bring all of that in through dedicated sources for specific sources like Salesforce for example or YouTube that we index frequently. As well as generic connectors. So if you have an API or a website or a sitemap we're also able to bring this in. As you can see in our connector repository we have both cloud connectors, on prem connectors. We're even able to use those old school crawling modules and actually go and crawl data into databases for example. And ultimately we also have a push source. You can also just push stuff at us. We'll be able to absorb it breaking into the platform. The key with this is that once all of this is available in the platform, you get exponential value from Coveo. Right? Because you're gonna then be able to use this and apply all of this content across different use cases. Whether it's a website use case, like the one we're gonna look at today. Or whether you're interested into growing with Coveo across other use cases like commerce or service for example. And of course, when we index all of this, we normalize it. So that one single query can retrieve a single source for a single use case or five sources for that use case or ten sources for that use case. We're going to make sure that we can retrieve across all of these different repositories from a single search experience which we're gonna look at just now with our Barca corporate website. Yeah. While you bring that up, I think that's so important especially the queries that we just looked at before. You might be in marketing and you're creating things on maybe Adobe or Sitecore or Optimizely. You're in your CMS. But the answer might have little bits that are in support or even product catalogs. They could be throughout the organization. And the user shouldn't be kind of punished because on the back end, all of your things are kind of siloed and in different departments. And it's almost an impossible challenge to try to bring it all into one source. So this is the challenge that Coveo solves and bring it all into one place and making it available as a single source of truth. Yeah. Without needing the, you know, very time consuming and costly kind of migration. Migration, duplication, etcetera. Yes. So we're here in the Barca corporate website. Barca is as I mentioned our demo company. Barca does many things. And as you can see there's different sub properties that exist like Barca support, engineering parts or sports. But the Barca corporate website essentially did like unified entry across all of those brands that belong to Barca, the Barca corporate company. So to make search experiences great, we're going walk through some of the basic things that are always part of search experiences that allow to make those experience better. So the first thing here if I click on the search box you're going to see like some recent queries I did. If I start to type you're going to see query suggestions. So we learn, right? As you mentioned earlier, we don't want you to have to manually tune and manually have to suggest every single potential query that might occur. So we learn from successful search queries that resulted in clicks, that resulted in a user consuming your content and we make those available to do a query essentially from those and to be able to get essentially the same success that previous users have gotten by doing this query. So query suggestion, we learn from successful outcomes and we make those available so a new user who maybe doesn't know your brand can very easily as they start to type navigate and get suggestions of things they might want to search for and look and learn about your business. We even can do things like fixing typos. So if I search for GSP and I'm not actually looking for the famous French Canadian MMA fighter, I'm looking for a GPS because I'm on a boat company here. Well we can actually correct those queries. So here we can see there was no results for GSP because this is not related to MMA, this is a boat company. We actually automatically corrected this query and we're able to find content for GPS in this specific scenario. That's really useful. I think especially because when people are are searching like you said, they don't know your taxonomy and they shouldn't need to know your taxonomy. I think a really good example of that is healthcare where people are gonna be searching probably not using the correct medical terms. They just don't know it. Maybe they're searching, you know, pediatrician. Maybe they're searching kids doctor. You need to be able to give them the the most relevant results regardless. Indeed. That's a perfect segue into this next example I have which is how to listen to underwater sounds. So I'm looking for sonars. I'm looking for underwater acoustics but I don't quite know the right term for it. Right? So as you can see here, I search for underwater sounds but the search was able to intelligently identify the meanings of what I'm talking about and to return results for underwater acoustics for example. Underwater sound technology. Underwater acoustics, what are the basics, right? So we have what we call a hybrid search where we understand queries lexically. What are the words that were in that query to find the right content but also semantically. What does query actually mean? And then to add an additional layer on top of that, we have here recommendations where we're going to be able to learn from what past users have found the most useful for that given query. Even with those misspelling. Even if they call it a kid's doctor instead of pediatrician for example. And go out and actually promote or recommend the most popular information. So using the power of the crowd to avoid those manual tuning, right? So we understand synonyms, we understand words that are roughly the same or have the same meaning but could be written differently. We can learn from that over time. And then as users use the site, as users consume the content, we'll learn from those behaviors to continue making the experience better and better over time. That's a lot of time saved. Indeed. And of course we kind of skipped over it but as you can see here we generated a generated response. So we've seen this more and more over the last few years where our customers are moving from search, standard search to Generative Answering, providing at least a first answer, right? As we mentioned earlier, content might live in multiple places. So in this example here, if I look for like what are the latest sonar technologies, there's a lot of different articles that actually talk about that. What we want to do is we want to take pieces and content, you know the best bits from these different articles and bring them together. Coveo has found a very advanced and very powerful way to do this that keeps the relevance of your search that is powerful, that is optimized and that improves over time, and is able to then bring that power and those events retrieval and relevance capabilities into the Generative Answering. So we do what's called a two stage retrieval where we'll retrieve the very best results for a given query using everything I've just mentioned across lexical, semantic and behavioral models. And then from those very best results, which consist of articles, which have a lot of details, we'll actually go and find the very best passages. The very best bits of text across those documents and bring them together even if it comes from multiple sources as this example here. And bring that into an end to end generated response that will give the user what they're looking for. This example here where the latest sonar technologies, I get a detailed, thorough and grounded answer that will only use your enterprise content, will not make anything up so that you get a proper answer that is accurate, that is insightful for the end user with a lot of ease of use so the user doesn't have to navigate across three, four articles to get this exact answer. Yeah. Honestly, there's a couple things that are interesting from about this from my perspective being a marketer. The first one is you don't need to create a new piece of content for every query that could ever possibly exist. Right? Because it's pulling from across multiple. So you're already kind of making the most of the content that you're putting out, which I think is obviously really important. You also kinda mentioned the security aspect. Right? I know with Generative Answering, it is definitely widely adopted at this point. But there is some still some maybe hesitancy around it, especially in like highly regulated industries. I think a lot of that is due to kind of a lack of understanding about all of the guardrails that do exist or at least should exist. These are definitely things that are really important to look into. One thing for example is that you don't need to have all of your content available for Generative Answering. Right? You can kind of You can pick and choose the ones that you want to. Indeed. That's a great point. As we've seen this actually ourselves with our own Generative Answering deployment a few years ago. Where we have things like documents, article, but then we also have community posts. And community posts are obviously very insightful for search. You probably still want to have them in your search results for a user to be able to navigate across past posts that were posted on community. But a community post is, with the name community, means it's open. Means anyone can contribute, ask questions, they might formulate their question wrong. Someone might contribute an answer that's not perfectly accurate or maybe becomes outdated over time. So there's a differentiation between making something available for search which community posts can be part of. But then we can actually ensure that what is made available for generative answering is only the source of truth by defining which of the content that's available in your search actually is made available to use for generated responses. And of course talking about security, we even apply additional layers with Coveo being at the root built to support security. So items can have even one to one permissions for example. So we make sure that only what people are allowed to see is made available to them in search but then we also give our customers control over whether that content that is possible for user to see can then actually be used for generated responses. Which will typically content that you know is grounded, true, accurate and up to date of course which community post might not always be. Yeah. That I think you definitely want to have control over what it is that you're being able to feed in. That's really important for a lot of people. So I think it's a really great thing to point out here. Absolutely. And then of course, we're kind of glancing over the follow-up here so let's actually go and try that. As of course you know those experiences have evolved to now support things like conversational. So here I asked a question about sonar technologies and I want to know a little more. Like what does Barca offer for that? And as you see here, I didn't actually say what does Barca offer for sonar technologies. I've actually said what does Barca offer for that? So this is really where the Agentic Retrieval comes in. So we've developed an agent to power those conversational search experiences where the follow-up question might just be like tell me more. Tell me about that. What does Barca offer about that? What is that? Yeah, is that? So as you can see here it's very interesting. We have a stream of thought that will actually show what the agent is doing underneath. Like you might have become accustomed with when you work with Claude or ChatGPT if it thinks. So the same happens here where the agent will actually take that, understand that it means sonar technologies and then do a search internally for Barca sonar technology product offerings and solutions. And then we've added a fun thing to this demo where we've added custom behaviors because that's another thing our customers tell us. So yeah, your generative answering is great, it's accurate, it's grounded, but I also wanted to speak like my brand and use my tone. So here we've purposely added some friendly boat captain tone where it says ahoy, you know Barca's got you covered before giving your answer. Which will be grounded again, same principle as the first answer, only using the content that was made available. And then to ensure that we help users navigate across that conversation and ask for more, we even added with those custom behaviors here follow-up questions, right? So this can be tailored from one customer to another to adjust and narrow or customize this experience to fit your brand and tone and how you'd like it to be. In this example here we even have suggested follow-up questions which again will help guide the user down the path of what they should want to do, they might want to do next. And then finally you know if I completely change topic here, well you know the agent will keep track. The agent will know that okay well I'm actually going back to the acoustics topic we discussed a bit earlier here. Those two previous questions were about Sonar Technologies and Barca. Acoustics is kind of related but the agent here will understand. Well the latest query was actually about acoustics. So okay, let's talk about acoustics. Let's tell you about acoustics within the context of sonar system. So this example here, it's actually talking to me about hydrophones, about sonar system, about systemic serving, explain the differences between all these things. So again, I'm getting a great generated response. The LLM and the agent here has thought about it properly and made the proper query in the context. Understanding that I'm kind of changing topic but I probably still wanna compare that to previous topic to make sure there's no no confusion anymore. And then generated a nice end to end response with again some suggested follow ups to continue that conversation going. It's honestly really impressive when you also just think about all of the content. Like you you already had all of this content before but it's just bringing it to life in a completely different way and really turning your content into conversations. Like that's that's really cool. Yeah. And there's an element to it that we haven't really touched on here but there's lot of ease of use in deploying that, right? We've built this at scale to work for large enterprise. So it's actually possible if you have content indexed in Coveo which some of our solution engineers can do quite rapidly actually. Half a day they can index a website, bring some content in to do a POC. We've seen you know some proof of concepts get built and set up in less than a day. Yep. Where you can bring content in, throw our search agent on top of it, enable conversational search on a test page and suddenly be able to test this with your own content which is always pretty impressive. I mean we're working in this industry and I still get impressed when we're able to put together a POC, bring some content from a customer, put put it in our index, you know have our agent learn from this content and then start to answer questions very accurately with that content for this brand new customer we've never never worked with before. So ease of use has been something very key that we've added and made this so that our customers can very rapidly and easily benefit from the value of conversational search. Yeah. No, for sure. And that's why it's even, you know, before we talked about how people are kind of at all different levels of maturity. So even if you're not there before, we've seen a lot of success in just kind of starting with unifying search, having you know query suggestions, recommendations, automatic relevance tuning. And they're able to kind of prove success really quickly within their organization. And then when the time comes to, you know, add in generative and conversational that is, you know, happening at every board level saying to go and do this and do it like yesterday. What agents today. Exactly. You're able to kind of set that up and run with it quite quickly and of course we're there to be a partner as well and Absolutely. Be on the journey together which is something that's also really unique. We're talking like one line of code to add this whole block and then a toggle to add the follow ups. Right? So we've made this very easy to turn on. Once you have everything in the index like we've mentioned earlier where the magic happens, the rest becomes very easy to deploy and to grow across those phases. Awesome. So I think that kind of covers it for the demo. So let's perhaps go and do a bit of a wrap of what we just looked at, how Coveo can help you in deploying these capabilities across your business. Awesome, so let's review some of the things that we just looked at that make those experiences work at scale for enterprises. So the first is connectivity. So as we mentioned at the beginning of the demo, all your content, one index, no migration. So we connect to any source, we bring it all in. We're built for enterprise scale. So whether you have a couple thousand document or you have a couple millions, we're able to bring all that in whether they're PDFs, FAQs, videos, long form guides, those ugly thirty page PDF, not a problem for us. We can bring all of those in. Full document indexing, so not just the metadata. So again, those thirty two page PDF, we digest all of it and we make it all available for generative answering. So even if the answer lives at the thirty page of a thirty two page document, we'll be able to go out and find it. And finally, this is all obviously a foundation to grow. Once you have it in our index, can power your search. But we can also power generative answering and then your conversational and agentic experiences. All from one secured permission aware index. The other element that we looked at is AI search and discovery models. So to make the experience better over time and avoid having to do all this manual tuning, we have generative and behavioral ML models that will improve the experience over time. So understanding the intent, semantic, lexical and behavioral signals all working together to get the best experience. Self learning. So again, as your users interact, behave, we want to learn from the power of the crowd. What's the best way to learn from all your different users? What did they actually search for? What did they click on when they searched for a given thing? How did they actually behave? Learning from the positive outcomes that they have with your site and making the outcomes of the next user even better through the power of the crowd. And then those models compound. So they work together, whether it's the query suggestions that we saw earlier to help guide the user and ask the proper queries, automatic relevance tuning that improves the relevance over time, or a semantic encoder to understand the meaning, and other models that we have to understand intent or improve towards outcomes, those all work together and compound over time to make the experience better. And then of course it's proven to work at scale. We don't work with very small companies. We tend to learn with large enterprises and we're able to ensure that these work at whichever scale you might be operating at. For sure. And also these same models can be applied across use cases without really doing anything different, right? So we're talking about the website use case right now but where you're gonna see the most success is if you use it across and you use it across the full user journey. So they start as a prospect, they do some research, maybe they make a purchase, they become a customer, maybe they have a support question later. Even even workplace, right? Because your employees are also your own customers. Absolutely. All of these models are able to be applied across your whole cross pollinate. Right? So you get to choose whether you wanna cross pollinate or not. But of course we've seen great value for our customers to use those insights from one use case so then make the other use case better. It's all powered ultimately by the same index. So this at the root provide a unified experience across those different use cases. But then those signals that we learn from across experiences can also be cross pollinated to further improve the experience wherever wherever the user might be. So some of the models that we just discussed that we're showing here on on screen are here are things like query suggestions, automatic relevance tuning. We have content recommendations as well where we'll proactively recommend a piece of content to a new user. Even the facets and filters, we haven't really touched and spent a lot of time on that. But over time we can even reorder facets and filters based on what's the most popular. Smart snippets and of course generative answering and conversational search are now the key pieces that our customers are most interested about. But these on their own, they're not going be as powerful if they're not powered by the proper underlying retrieval layer, which is what the other models in our unified index allow to bring. To actually make generic conversational search work at scale and answer properly. Another big element that we've touched on is security and governance, right? So that you can trust your search especially when the stakes are high, especially in regulated industry as you've mentioned earlier. So enterprise grade security is there at the root. We're ISO certified, HIPAA compliant. We have all the built in security that's possible to serve large enterprises. We're also permission aware, so every piece of content that's indexed can have item level permissions. So if Carrie Anne does a search and I do a search, we do the exact same search on the exact same use case. Our permissions might allow us to see different pieces of content. Anything that's outside of permission, it's as if it doesn't exist. You cannot make a mistake essentially with the manual rules. Those permissions are built in to be bulletproof and things that cannot be circumvented by rules or anything like that. Carry on sees what she's allowed to see. I see what I'm allowed to see only. Anything else doesn't exist. Certified and audited of course, as I've mentioned. And secured by design, right. So encryption, single sign on, access management, security and control built in. It's not bolted on, it's there by default. So if you grow across different use cases from a website search to a support search, maybe a commerce search and different teams have to start to work on the same platform, All this is built into scale at the enterprise level and support different teams working in the same platform across different use cases from once again the same unified index which is really where those economies of scales happen. And that's also what allows you to grow within the highly regulated industries because once you've been kind of certified for one use case you can very easily start to grow and not have to go through everything. Exactly, the same motions every time. And then ease of use and flexibility, we've mentioned that but we've built our technology to allow you to have faster time to value, right? So ML models that can deliver results from day one, as I mentioned. We can do POCs in a day, right? So give us a week or two, we'll have something fully ready end to end for you to actually test and try and touch what we're able to deliver for your business. It's very business friendly, right? So as you mentioned earlier, marketing teams would all run for them to have to spend tons of times doing manual optimization. So things like Search UX Builder, Pipeline and Rules are very easy to manage, Very quick to deploy, plug and play, almost slow or no code tools that allow you to go live faster. An easy path to grow from your standard search to Generative Answering to conversational. And of course, flexible, right? So we do want to make it easy for you to deploy, but ultimately enterprise will have their own brand, their own colors. So we do allow for you to customize your experience so that it suits your brand, your tone, and your needs. Yeah. This part is interesting too. Right? Because a lot of people do just kinda want something that's kind it's easy, it's managed. Right? They feel like they don't have to someone once told me I don't have hours to sit and manually tune everything. Right? Oh my god. Or they want marketing to be more efficient in their rules and not rely on IT all the time. So having those components that are easy to use to get up and running quickly. But then other people actually want that flexibility and that customization APIs. They wanna really get in and make it their own. So being able to offer kind of both of those things is important. Absolutely. And so you have an easy path to get started very easily with almost no code, but you do have the flexibility. Ultimately, we're an enterprise grade platform to scale and to add more custom or more particular use cases you're depending on the needs. And then finally, course, generative and conversational search, right? So turning content into conversation. So we can answer complex questions with clear conversational responses even for long form questions. We can synthesize across sources so, like as you mentioned earlier, no one has the time to write a new piece of content for every single query that will happen, right? So with existing content we're able to bring that existing content together and essentially pretty much create new content at runtime when you get a new question that perhaps we never got before. Every answer is grounded, right? We think of like this temperature level that LLMs can have where they can be more creative, hallucinate a little more and be more grounded. At Coveo, no hallucinations. We focus on making sure your LLMs and Generative Responses are fully grounded within your content and never go out of track or out of brand, with giving you nonetheless the flexibility to still tweak customize how you want the LLM to answer and how you want the LLM to use this grounded content. So we'll always use the content. We'll always stick to what you have made available for the LLM to use to answer. But you can still tune it to represent your brand, your tone, not talk about your competitors, for example, or talk like a pirate, like we did in our demo earlier. And then finally, continue conversation, right? So with those follow-up, we allow end users to continue that across multiple turns. To power conversational search, we've seen customers then of course bring this into Chatbot and other use cases as well, as ultimately we give you the right tools with our APIs, with our solutions and models for you to scale this across multiple use cases across the enterprise. So that was essentially the key things that we saw and this I guess is a kind of a one page recap everything we looked at, right? And this is where we've seen experience and we've seen the industry evolve, right? So what we see to be the modern search experience today is you can ask questions naturally, right? Your search understands your intent, not just keywords. You get answers that you can trust. You get ease of navigation to navigate across this. And it's funny, we've heard customers tell us that hey, we want conversational but we still need those filters because we have so much products, so much variety across our brand and across our products that we still need to have those filters. We want to allow for you to use or ask follow-up questions to continue the conversation, still be able to view relevant results. Again, customers want follow-up. They've been telling us they want the full conversation. But what if someone just wants to go look at a very specific article and read it? That should still be possible. And then of course guide users to relevant content with proactive recommendations and ML powered improvements over time. So the experience get better and better and your customers can discover your content, get great answers, self serve or buy more from you. Ultimately improving the business outcomes that your business can realize with Coveo. Exactly. Honestly, this is kind of where search is going. Like we said, we understand that for some it can take longer to get there. It's walk, crawl, run. But at the end of the day, it is really about the foundation. I think that's just kind of the key point that we wanna make here is making sure that you have that solid foundation from which you're able to grow to go into those conversational experiences. This is truly where it scales and where, you know, customers get economies of scale, where customers can grow and expand across multiple use cases and ultimately get the very best value for their enterprise. Exactly. Well, that kind of wraps it up for us today. Hopefully, you've enjoyed this session. If you have any questions, feel free to reach out. We're more than happy to chat. Thanks. Thanks everyone.