Alright. Good morning, everyone, or good afternoon rather, I should say, depending on where you are. Welcome to this presentation, of the partner power hours. My name is Alex Moreau. You should already know who I am by now, but this is the day two of the RGA builder workshop. Let me just start sharing my screen, making sure that everyone can see. Doing that slide. Perfect. There we are. Hang on. I always have oh, there we go. Alright. Perfect. It works now. Yes. So my name is Alex Moreau. I am a partner solution architect at Coveo. So today, we'll be doing the follow-up for what we did, two days ago. So two days ago, just a reminder for everyone. We, what we did is that we created our RGA models as well as our semantic search models in our own organizations, and we index some content into it. And so the today, what we're gonna end up doing is just building those search pages, calling that RGA model that is now finished building, and and and seeing what we are able to get with effectively minimal, setup time. We'll also we'll be doing one, sort of a a a quick setup. We'll all I'll also be showing you how to do it in a sort of a more formal HTML page. Nothing crazy going on there. And then we'll talk a little bit about the importance of of, quality assurance. But I don't think that today is gonna take a full hour. So I just wanted to let you know. But still, I think I don't I don't think that you're gonna be angry to have a few minutes at the end of the the this session, the to yourself. That being said, there you know, if you have a lot of questions, we might go to the full hour, but, you know, we have the q and a. You can always ask questions. I'll make sure to I'll make sure to keep an eye on it and and make sure to answer any questions you may have. As per last time, this is meant to be a little bit interactive. So while I will be showing you exactly how to do things, I expect you to follow along if you wanna have a fully functioning search page. The steps are straightforward, so it shouldn't be too complex to do. And I'll make sure to to zoom in and and make sure that you're able to see what I'm doing. It's it's everything is in the UI, well, except for the, you know, the HTML part. Everything's in the UI, so that that's that should be easy to follow. So let's start with continuing with my slides. The disclaimer as usual. Right? We're a publicly traded company. So, yeah, the everything that comes with that. I'm sure you have time after this meeting to to read that into more details. Yeah. So as I said, this is the day two of the builder workshop. A reminder as well that next week on Tuesday, we're going to have a seller enablement. So, for the commercial teams, mostly how to build a a Coveo practice, how do you, how do you pitch a bit a buy versus build, how do you understand how that works? Just the pitch strategy, what are the key use cases, and the business value assessment. So, if you're interested into that, I highly recommend you you join next week. We'll you you're going to have, Liz that's going to talk to you about that, and I think that's going to be very interesting for anyone who's, especially people who are more into the sales side of things. Alright. So, already, we're gonna, we're gonna start by jumping straight into the, the organization. So, we'll and what we'll do is we'll create a test page where we can test that. So I'm I'll keep I'm keeping this shared. So I invite you again to go on platform dot cloud dot Coveo dot com. So you should already be familiar with that and join your organization. It should start with workshop followed by a number, the number being, dependent on who you are exactly. And you should already have access to that if you've created your models. So when you you get to the page, we get to the sources section where you you're going to have your your source that you created, the other day. And if we were just going to confirm in the model section of the machine learning that you do have, you do have these models in VR built. I am getting a warning, and we're going to address that right away. So, just wanting to take a step back to make sure that everyone's on the same page in regards to Coveo. Whenever, you know, when you make a query to Coveo, whenever you're trying to get results back from Coveo, be it for RGA or or, you know, recommendations or or listing pages or your or your normal sort of search pages or anything where you need to get results back from Coveo, the request needs to go through what's called the query pipeline. And it's inside of those query pipeline that we'll be able to to indicate what to what models to use, you know, what rules to use, your business rules that you'd be adding as as a as a business user, as a marketer. And so what we'll be doing for this session so even though I said, yes, create a test page, the first actual step we're going to be doing is creating a query pipeline. So we'll go into the search section. It says query pipeline. I just zoom a little bit to make sure everyone can see what I'm doing. As I said, I would. And and we'll go into the query pipeline section right there, and, and and we'll just add a new pipeline. By default, you should have only the one called default, and it's tagged as default. And that's just because, as I said, if you need to get requests from Coveo, you need to go through a pipeline. We have a default one just to make sure things don't break, but it is not best practice to use that. We normally wanna create new pipelines. So we'll go ahead. We'll create a new pipeline. I'll call it wikiHow because this is just what I call everything for this use case. I'll add a condition, and we'll talk a little bit about the condition. But I'll add a condition that that's called search hub. So I'll scroll down to it's in alphabetical order. Search hub is and I'll call it we I'll go in here. Wiki how search. And and I'll explain in a second what that means, but I'll for now, I'll just do that. You might wanna copy exactly what I'm doing. One thing you you wanna make sure is that you have the word search in it in there, though, technically, for this demo, it's not that important. Just remember, note that down. Well, okay. I'll enter it again. WP house search. Note that down because we will be using it, in the future and not long in the future. So I'll add a condition. What I'm saying here is that if my request has the search of value called wiki house search, go through that pipeline. That's all I'm doing here. So I'm I'm adding the pipeline, and now it's it already exists. So what I'm gonna do is I'm either you can double click on it or you can click and click edit components. It does the same thing. And you're getting into the the UI of how to to change your wikiHow sorry. Your query pipeline model, in my case, named wikiHow. For this, honestly, all we need is to is to leverage machine learning. So we'll go into the machine learning section, and we'll add we'll associate rather our two models. So I'll click associate model. I only have these two, so I'll add wikiHow. And what I'm telling, what I'm telling Coveo right now is I'm saying, when I'm going through this pipeline, I want you to leverage our GA. So that is what we want to do. We don't need to add a condition here because the condition is already on it's there's a super condition on the pipeline. We don't need to add a specific condition here. Though, I've seen a lot of people do it. For example, you know, one very common condition we may add here, and I'll actually even do it, is, make sure that the query is not empty. So query is not empty, and that means, you know, don't call RG if there's no query. If there's no query, nothing's ever gonna come back. So there's no reason to call it. But because, you know, if no if someone's not asking a question, don't expect an answer back. Then I think that's that's, easy to understand. So I just add add condition is query is not empty, then call RGA, and I'll associate the model. And the second one I'm gonna do is our semantic search. So I'll add the semantic search. This one doesn't need a condition, because it doesn't it it doesn't cause the same. So the reason why we added queries not empty to the RGA, is is just for the sake of consumption. Because one thing to be aware of and aware of, and I believe I've mentioned it last week, but one thing that's important is that when you when you, have an RGA license, when you get a CRGA license with Coveo, you are you're entitled to a certain number of, of queries per month for your, yeah, for your RGA. So yes. Or generative queries per month is what they're called. And so, you know, by removing all the queries that, the request for all queries that are empty, then we're saving on that. And I think that's that's normally normal. But for semantic search, you're not there's no limit to number of requests using semantic search. It falls into the bucket of requests total, and so you're you're you're safe in not adding a, a condition here. So I'll also sorry. I will associate the model, and that is all that I need to do here. Now let's go back to the why I added a search up value here. Just taking again a step back, making sure. So if you're if you're already familiar with Coveo, you should already know what that is. You know, if you've implemented Caveo in the past, you should already be familiar with that, but I wanna make sure to reexplain it, making sure everyone is on the same page. The search of value is the way that Twilio understands where your queries are coming from. It is normally a human readable value. Actually, you'd want it to be human readable because it's used for attribution, and it's used by analysts going on your site trying to or your own analysts, I should say, the customer's analysts going on on the organization trying to understand, you know, where those queries are coming from. It's used by our customer success managers as well to to be able to identify, you know, what is causing those queries. And and, you know, internally at Coveo, we use that a lot. And on top of that, it's also used by our machine learning models, that that are that are trying to verify, you know, where is that query coming from. Just making sure I'm giving you the right the right content for the for that page, making sure that we're getting the right the right suggestions and everything based on the data that we have. The reason why I I called it wikiHow search, is because there's sort of, there's a requirement that, you wanna have the word search, somewhere inside of it. It doesn't need to be capitalized or anything. It doesn't need to use spaces. You know, I've seen dashes used often. But we expect to have the word search somewhere inside of the query pipeline for a search page. If this was a recommendation, we would expect the the word rec, r e c, for recommendation. We'd expect rec somewhere in it. And that's just for it's for simplicity's sake so that we understand where where that coming from that's coming from so that it's it's the same everywhere. But, also, our machine learning models, looks look for those specific terms to, further optimize the the way that the machine learning, results are coming back. So just be aware of that. So this is why I call it WikiHow search. So WikiHow is my identifier. Search is telling me that it's for search. If I add, WikiHow in French, English, Spanish, I I would've we fixed it with, with that at the end. In my case, I only have English, so I didn't add it. But, you know, you you'll see oftentimes something like WikiHow search e n and then or WikiHow search e n US, for example, if you were, distributed globally. So this is just why I created that way. So now I have a now that I have that, I have a fully working query pipeline that has RGA on it, and it gets accessed when my search hub is Wiki WikiHowSearch. I'll copy this because now I get to go and create my search page. Thankfully, that's super quick. So I'll go into this section, nice little section called search pages, and I'll just click on it. And the search pages section is where you can create pages, inside of Coveo. And they there's multiple ways in fact, I'll click on add search page so you can see it. There's multiple ways of creating search pages inside of Coveo. In our case, for RGA, the simplest solution is to use a simple builder, and that's and you'll see why for in a second. But a simple builder is just a lightweight, you know, a lightweight builder that is just for prototyping. It is very limited in what it allows you to do because its purpose is just to show how, what a search page may look like. This is hosted inside of the Coveo instance. Anyone that has access to your, instance normally has access to that page. Of course, only being able to query the content that they're already able to query. And so they and so we can see that quickly. The advanced section is using our older framework. It is not recommended anymore. It does allow you to do, you know, a little bit more inside of the UI, elements, but it is it is normally not recommended for, production usage. And then our developer path is, if you wanna create the page locally on your computer, and then you can you have control over everything and then upload that to, to the Coveo, to your Coveo organization so people can access it there. So it's sort of the it's more complex than the simple builder, obviously, but it is a lot more powerful. It does require you to use the Coveo CLI, which is available for free online. The Coveo CLI, the command line tool, that allows you to to to push those pages. But in our case, you know, we'll just use a simple builder. We will call it wiki house search as, as I called my search hub. It's not required. It's just best practice, and I'll show you how you could change that. It it then the reason why I say it's it's best practice is it by default, the search hub is going to be that name that I enter, though you can change that. So I'll just call it wiki house search so we don't have to worry about that search hub. I it it's asking me to log in again. And so right now, I have a working page. RGA right now is not enabled, but how do I enable it? I go into the settings section, and I click, yes. I want relevance. And now I'm just gonna save it here just to make sure that, apply changes that I don't forget. And then if I wanted to change the search up, I could go in here and change it, or rather I can't change it. So I it's a good thing that I call it that. It just tell it's telling me what the search up is so I can copy it and add a pipeline based on it. So now that I have relevance generative answering working, you know, in here, we have things like how to do, household chores, things about moving, things about garden. And one of the queries that I like to do with just that that context is what is a sweet potato. Right? The same gardening, and I'm I keep seeing that thing, and I know what it is. But let's just say I'm I'm wondering what that is. I'll ask that, and there we go. Now we have a working RGA model that is telling me what a sweet potato is. So it's answering, you know, it's a root vegetable and a little bit more it's it's known for its taste, the various culinary dishes. And the reason why I made that request specifically is because there's no page on WikiHow that is just what is a sweet potato. Instead, it's it's taking this information from how to grow a sweet and in fact, it's right here. How to grow a sweet potato vine, how vine houseplant care and display tips. And and on that page, there's just information about, like, what is a sweet potato? And it's able to deduce based on the information that's there what is that sweet potato. There's a few other, you know, queries that we might try, things like how to move with a pet. So maybe you're you're thinking of moving. Right? We're in the summer, and I have a pet. And how do I move with a pet? So there we go. So now it's it's it's telling me a a few steps. Let's say I find that I want I have a little bit more bullet point summary. I can click on that, and it's going to give me, a bullet point summary. So those rephrase options are there by default. You know, I just clicked on that box, and it's it's giving me those options, and it's already working. So it it's telling me how to move out with the pets. So plan ahead, make sure they have your pet carrier, bring enough food, and I forget to bring a leash, everything like that. And, again, like I said, there's no there's not even a single page that's telling me how to move with a pet, but it's telling me how to move with pictures and how to move out of stick with pictures. And on those pages, there are points that talk about, you know, how do you go ahead and move with pets. And so Coveo is able to figure it out, just reading because it's aware of those those chunks of documents. You know, it took those pages and made chunks out of them. There's a chunk that says specifically how to move out of state with a pet. And so you might notice that the answer here is a little bit more towards, you know, long because it's out of state. There's long term, you know, long term thinking. You know, ensure the hotels you stay at are pet friendly. You know, might not be useful if you're just moving within the same city, but right now, I didn't say I'm moving within the same city. And so it's making sure I'm aware. You know? If you're going to move and it's gonna take a while, make sure the hotels you're staying at are there. And then you'll maybe if you're moving cross continents, we don't know yet. Consider some some of your things by air cargo, and everything. So there, we have a lot of information in there. If you go ahead and I hope you were able to follow along. If you go ahead and create those search pages like I did and make those same requests, if you had the WikiHow page, you should have the same type of content, that's coming back. So it's just a quick example of how, how we can expect that to to work. Now let's go in here, creating a local search page. So, there's a say, I don't have a lot of slides for today. So, how to create a local search page, you know, this is all good, but, let's say that you have a you wanna have it working on your on your own CMS. So this is using Atomic. Again, if I go back to that page, this is an Atomic page. Atomic being a framework made by Coveo that is a a you know, that is lightweight and easy to use. You know, you create the everything using simple HTML tags. So for example, that facet right here is a simple HTML tag, and I'll show you exactly what it looks like. I'm using Atomic for the sake of simplicity. If, if you wanna go a little bit deeper, say you wanna integrate with, with things like something like Nuxt JS or, you know, Vue JS or or or another framework like that that may not always be already be supported by Atomic, then you can use what's called headless, where, you do need to to tell Coveo or you need to to tell exactly how you want the HTML to be rendered. So it's a little bit more, it's it's it's for front end developers specifically, whereas Atomic can be used by non front end developers. The headless for libraries for specifically front end developers looking to build front end pages you by calling Coveo. So it's sort of a layer on top of our APIs to make things simpler. But, in our case, we'll just use Atomic for the simplicity. So I'll go into it. So this is just what my page looks like. Right? It's a it's a dummy page. If you're familiar with HTML, it's it's just doing it's the pages in English, UTF-eight. I have a name and author. The atomic part is these. I'm loading atomic through our CDNs. These are public CDNs. You're you're more than welcome to use those for your public pages, your public website. Right? We are making sure that these are production ready. A lot of our customers are using those, so you're more than welcome to use them. But, so I'm doing the same thing. I'm just calling that, I'm just calling the the the CDN to get atomic. V two means I'm using the latest version of v two, because I'm not worried about versioning. But if you were, if you you can always specify exactly which version you wanted. In my case, I want the latest all the time. Another important part is the initialization script. I'll make sure to zoom oh, wait. How do I zoom with that? I should have looked at that a little bit earlier. Oh, I don't know how to zoom inside of that. I I'm hoping that you can that you guys can see a little bit. Here is I'll just I'll just zoom inside of notepad. There we go. So what this script does, it's very basic. Right? It's, this is all in our documentation, but it's awaiting for a custom element, called atomic search interface to be rendered. Once it is, it is going to get that search interface inside of a you know, it's it's going to get it using a query selector, and then it's going to initialize it using an access token and an organization ID. We'll we'll get to that in a second. This is definitely all that's missing. And then after that, we we just call the search interface execute first, first search. And then after that, I have a very, very basic search interface. Again, I'll just I'll put it inside of Notepad so you can really see it. Oh, I forgot to change that name here. I wanted to call it I I sorry. I was reusing one of my old one, wikiHowSearch. So I'm calling it wikiHowSearch, making sure that the search hub, is going through the right pipeline. So let's just make sure. WikiHow search. So that, it's going through our pipeline that I'd created earlier. Then there is a just add a little bit of styling just to make it not boring. Just, you know, the minimal amount, your max width, other margin right, other position relative, so nothing crazy. Again, this is all documented. And then I have a very basic layout called search. Again, this is just for styling purposes with a search box and then any and this is the magic here. Atomic generated answer, that is RGA. That is what Atomic uses to create to generate that full RGA box. It's a so it has those filtering options. It has the thumbs up, thumbs down that you can leverage. It has the the the full rendering. That that's just this one HTML line. No no properties needed. That's just what it no attributes, rather, I should say, needed, and that's all. And then I just added a quick layout section with a result list, just so that I can see those results coming back from Coveo after. Technically, it wouldn't necessarily be required for a quick demo, but, you know, I I think it still makes sense to have it. So that's all I have here. And so organization ID, that's easy to get. It's it's in the URL, but it's in a lot of places. I'll just I tend to just take it from the URL because it's easy. So it it's right there. But, you know, it in my case, you know, it's called gen AI workshop one. So I'll just copy here as well that does the same thing. I go in here, and I add it to my organization ID that I had. For my access token, there's multiple ways of going about doing this. But because my content is public, I don't really care to I don't wanna complicate things by adding permissions and everything, and creating search tokens. So I I'm just gonna quickly and simply, create a a an API key that it has very limited permissions, and is able to make queries to RGA as well as to the search API. So I'll go down into the organization section in API keys. And right now, I have nothing, which is fine. I'll click on add key. I'll name it wikiHowSearch. Right? Calling it the same thing is always useful. And then I'm gonna go under privileges. I can actually already say I want a search, search privileges. So what this did is it added, push analytics, which is a very basic, what it's telling Coveo was that this API key is allowed to push analytics content. It cannot view. You don't want push and edit or view. In fact, it's grayed out because I have I have an another privilege. But you do not want view because view would allow the API key to to see what everyone's searching. That's not what what you want. We want just to be able to push those events. And so I'm I'm I'm adding that. There's nothing else. And then if I go to the search section, what it added as well is oh, and see now it's even hang on. It well, I'm gonna disallow impersonate, in fact. But I am going to give it, execute query. Impersonate will allow it to, would allow it to do, search token retrieval so that you you could do permissions. But in my case, I just want execute query for and execute query for that kind of content would be only public content. I'm gonna allow it, and now it's telling me, hey. That's too broad. Please make sure you are enforcing a search of value in your so that the API key is enforcing a search of value. And so I already have one, thankfully. Well, in fact, I I have one. It's called wikiHow because I was testing earlier, but I'm gonna call it wikiHowSearch. And now then now that's fine and it's working. So I just need these two things. And now see, it's even telling me it's anonymous search that I have as a preset. So I'll add the key. Now it's going to show me this. It it's important. This is the only time that the key is ever gonna be shown to you. So don't lose it. Make sure you copy it when you create the key. We can always go back and change the permissions after change the privileges after, but we cannot go back and look at what the API key is. If you lose it, you need to create a new one. So and I'm just gonna copy it to clipboard, and that is what I'm I'm gonna be using inside of my access token. So I'll put it right there. Now what I'll do is I'll just quickly run a a quick, Python server on local host, and I'll access local host right here. Local host eight thousand. My page was called index, dot HTML, so that's why the that's why it's it's just loading by default when I access local host. And then if I do the same well, I have a lot of queries that I was testing earlier. But if I make a request, like, how to make a budget, how to oops. How to make a budget, It should start answering me. There we go. So, again, we get we're getting answers. It's based off of the, it's based off of the content that I have in my index, which is, again, the wikiHow. Again, another example of of of, like, a good, a good query for the content that I have is I I entered how to make a budget, and it's it's getting information from, in this case, four different pages. Things like how to move out your parents' house, how to move out at sixteen and at eighteen, and how to live under with under twenty k a year. Or and so the you know, normally, that's those are the times you'll need to make a budget. And so it's figuring out if you need to make a budget, it's like you likely need a tight budget. You wanna make sure that you're that that you're frugal with your money. And so that's why it's it's telling me, you know, calculate your monthly income and and and expenses, your spending habits, prioritize your expenses. So make all these stuff. So this is, again, very, interesting, and and answer that there's no single page that tells me how to make a budget, but there are a lot of pages that include content about budget budgets. And so that's how we're getting an answer here. One thing, you know, now I have my search pages working, but I I do wanna take a minute, to talk about this is one thing that's really important is that when you are, when you're implementing RGA, you know, we have a fully working RGA model. I it didn't take a long time, and it's already fully working. Of course, I didn't style it, but I and I didn't do a crazy amount of stuff in it. But still everything is is, you know, working. What we we find is that when you're implementing RGA for the first time, or not even for the first time, but any time for a customer, you're going to spend a a a large amount of your time, not doing the implementation, but rather doing quality assurance. And and there's multiple ways of doing quality assurance, but the main one is you need to get people who are subject matter experts in the content that you are indexing. So that normally means you'll need to go to your customer and ask them, hey. Do you have a or rather provide me with a list of people that know your content very well, a list of people that can do QA testing for us, A list of people that can really dig deep and and understand whether the answer that's being provided is accurate. If you're working with organizations that that already keep, you know, their their content quite, accurate, and you when you made the scoping earlier, you made sure you were scoping it to only those specific pages that have content that's super accurate. That is going to be a lot much smoother process. But let's be honest, those customers are are are are rare. You tend to fall under, you know, those enterprises that that they just need a solution to to fix their their their, knowledge problem. And and that means, you know, you might end up with a lot of inaccurate answers. You might end up with, you know, inaccurate answers because you're you have inaccurate content. And so you need to confirm that what you're doing is is what needs to be done. Once you have those SMEs working, you know, there you are more than welcome to to have them use these, these thumbs up and thumbs down. Thumbs up, nothing happens if I thumbs it up. But if I thumbs down, I have this little survey that comes up. And it's it's this information is going to be sent directly inside of the reports in Coveo so that you can then go and dig down. Okay. After it, say, give them two weeks to do those QA testing, you know, the typically, those people who are subject matter expert, they have other job they have a job to do. They have something else to to to do. But, but, you know, once you give them enough time and they take, say, thirty minutes out of their day to to look and and make a lot of query that they can think of and then see if that makes sense, then they can say, yeah. Why did I say it was bad? Is the answers is the answer about the right topic? Yes. Is it the answer free of hallucinating content? That might be the problem. In our case, you know, yes. It's free of a list. So there's no hallucination. Can the answer be doc be answered by the documentation? That is a tricky one. Right? If they answer yes and and let's actually go through every question and tell you and and and talk about, how we you can go about fixing that problem. Is the answer about the right topic? If they say no, that probably means that you have the same nomenclature for multiple, for multiple different types of products inside of your content. When that's the case, the way to to fix that is, well, there's multiple ways, but either you change the scope of what you you are, answering or, or you go about, you know, adding certain filters. One thing to keep in mind is that if you have your facets on the left and you select a specific facet, the answer that RGA provides you is scoped to that facet. So you could you can make sure, you know, make sure you add a facet so that the user can select which product they're talking about and make sure we're on the right topic. So it's typically what why we see that. Is the answer for your hallucinated content? That's that's not something that happens often. If that's the case, that's when you need to open a support ticket with Coveo. If you get a lot of people saying, hey. That's hallucinated content. That's not normal. Then Coveo needs to step in and understand why are you getting so much hallucinated content. That is not normal. I don't see it often, honestly, but this is a good, a good use case for, for opening a support ticket about that. The third one, can he question be answered If they answer, yes. It can be answered, but it was not, well, you know, it didn't return an answer, then you need to go and have a look at the scoping of your RGA model. Did you include the right documentation pages? Maybe you you were a little bit too restrictive when you made your filter earlier, or maybe there's missing metadata on the documentation that they're talking about, or maybe there's additional things. So that's when you need to go back to the indexing stage and and and figure out, okay, where which page in the in the documentation was and and that's why we have a link to correct answer right there. Which page in the documentation gives me the right answer? Why is it is it in my index? Maybe you're not even indexing that content, and so make sure that you have that. And then is the ease answer readable? That is similar to hallucinated content in the sense that this should not happen. But if it does, this is a good, example of of opening a support ticket with Coveo for to to to go and look at that. So that's how you would go about, troubleshooting and and you were doing QA testing. What are your next steps for that? So it's a lot about making sure the scope that you have is right and making sure that that you are indexing exactly the right amount, not too restrictive, not too broad. So so that's really what you get. But this is how you would go about, sending the feedback regarding that. They can always skip if they don't want to, but that's always bad, because then you get no answer as to why the SME said that was a bad answer. Ideally, you know, especially if they're working for that company, you'll hope that they give you a a bit more information. But that's just one way of doing things. Another way of doing things that I've seen is, sometimes you'll have, companies making a spreadsheet with a list of queries that they wanna make sure Coveo answers, and then, they have their SMEs go through. And this is actually something that I've done, when Coveo went live with RGA, Coveo a year ago now. It's been a while. When Coveo went live for with RGA for the first time, I was I was, selected as an as an SME, as a subject matter expert in Coveo. And so I went through we had a list of of, of of spreadsheet, a list of queries that we wanted to make sure Coveo could answer. And so I went through every single one of those and and and typed them in, looked at the answer. Was the answer appropriate? Was it and, you know, those queries, what were they? They were the top queries that we were seeing in our in our in our Coveo search. They were things that we were expecting our GA to answer a a good answer, and they were things that we were just thinking about on the spot. Like, I came up with some questions that I added to the spreadsheet afterwards. But going through those questions, making sure is the answer accurate, is it the right thing, why is it the right thing, or rather right, why is it the wrong thing if it isn't. And, and then, you know, we we we sent that to the web team once, once that was implemented, and they were able to to to make those changes, based on that. That being said, you know, it was over a year ago, and it was still quite good. And we've made a lot of changes since then. But, you know, this is one way that I've seen people do things. But you can also go very freeform if you prefer. But the so that's why I'm I'm saying, you know, there's a lot there's a lot of tweaking that normally happens with RGA. So even, you know, just getting the right amount of content is crucial. This is mostly what I wanted to talk about today. As I as we said, you know, as I said the last week, you know, we're talking about a ninety minute to to implementation. I might have taken a little bit less than that, for a fully working page, that that has these things. So now what I'll do is I'll open it up to I'm gonna put my account here. I'll open it up open the floor for q and a, and, in fact, I'll go back directly to, to my slide here. So I'll go back to the q and a. So if you have any questions, please let me know. Again, as you're going through this, right, let's say that you are trying to implement RGA. You have a customer. You have a demo for a customer. You wanna make sure that you have RGA, working for them, feel free to reach out to us. And, you know, as a partner, you have a partner, a partner manager assigned to you. As a customer, you have a customer success manager. You should always have a point of contact with Coveo, someone you can reach out to. And if you're looking to to build RGA, if you're looking to, pitch RGA to a customer, we'll make sure that we can provide you with more tools. But I'm hoping that we're we were able to do this week gives you at least a basis from which you can build a a basic POC that's already working. And then after that, you know, if you need help with further scoping, we can come in and try to help you out with that. If you need help with any other, other things, you know, maybe you you were struggling with a different aspect that's not directly RGA related. We can come in and help you, guys, with that. You know, at the end of the day, our goal, at Coveo is to make sure that you understand how to build with Caveo, that you understand the value of Caveo, and that you're not wasting time, troubleshooting things that that we can just help you with. And so, so I'm opening the floor to any questions we may have. I know why we had a few, last week, and we had a few on, earlier this week. So, you know, you know, I understand if there's if if there are no questions. Reminder again that next week, we have a, seller enablement that, that would be useful if you are looking to build a practice for more business oriented folks. And also a reminder that next week and not next week, but next month, rather early September, there is the, the the partner exclusive, Relevance three sixty, with our VP of partner alliances, Anne Marie, who's going to talk about, how to build a practice in general with Coveo, not just RGA, but the whole the whole deal. How do you make that work? How do you make sure that you have the right resources? How do you make sure you have the right expertise? Who do you reach out to? What to build, etcetera? So, yeah, I I don't see any questions in the in the chat or in the q and a. So if, if I don't get any any questions, I guess I'll let everyone go. I hope this was insightful. I hope that you guys had a lot of fun playing with it. The organizations that you guys were added to is going to be staying up for a month. In a month from now, if you need that org to still be up and running, please reach out to us. In fact, reach out to us before that, ideally, and and we can extend it or we can, you know, create a new org for you, depending on what you have, and we will make sure to to to help you, migrate the content from one org to the other if that's the case. But please let me know. But in the meantime, you know, you have a full month to to go and play with that model and play with that organization, index new content, whatever you want, and have fun with it. So I hope that everyone has a good day. And, yeah, I hope that I speak to you everyone to everyone soon. Thank you.