Name: Unlock GenAI-Powered Insights & See ThoughtSpot AI in Action
Uploaded: 2025-03-27T02:15:39.436Z
Duration: 45 min
Description: Unlock GenAI-Powered Insights & See ThoughtSpot AI in Action

Transcript for "Unlock GenAI-Powered Insights & See ThoughtSpot AI in Action": Excellent. Let's get started. Hello everybody and welcome to our generative AI week, a week of brilliant content to cut through the noise of GenAI, a few actionable frameworks and playbooks to accelerate your AI journey. This is the first webinar in a series of four. It's sort of the taste of the intro, and then we have, an additional three webinars which will be available on demand with additional contact that I'd cover, I'll cover at the, at the back end of, this session. Just before we start, a couple of housekeeping issues. As you can see on the right hand side of your screen, you've got a chat, messages, docs, Q and A. I'm sure you can work out what each of those are for, but, if you have a question that you want to ask, please put it in the q and a. If you have any issues or problems with the system, please put it in the chat, and, Linda will be online, and, be able to help you. First of all, thank you for joining us today. My name is Stuart Rees. I run Thoughtspot out of Sydney in the region. And joining me today is the very brilliant Charlie Birch. Charlie, would you like to quickly introduce yourself? Yeah. Thank you, Stuart. So hello to everyone. My name is Charlie Birch. I'm a solutions engineer here at Thoughtspot, more of a technical resource, and I'll be running through the demo today. Originally from The UK, but moved over to Australia last year. But thank you everyone for joining. Looking forward to the session. Thanks, Charlie. So a little bit about me. I I've been around in the, data and analytics world for for far more years than I care to remember, certainly more than Charlie. And and and that's been across EMEA based out of London for many years and then the last fifty or so years across the APJ region based out of Australia. And what struck me is that in that twenty five plus years, I'll use that because that's the low end of the estimate, but twenty five plus years. I've probably learned more in the last twelve months about technology. And I think that's probably due to the way that AI is essentially disrupting everything. You know, almost the world, the the the market, but the speed of change, the level of innovation, the development and the way the markets and the world is moving, it's almost as if everybody has to keep up or be left behind. And that's not just for individuals, that's the same I think for organizations as well, you know. Evolve or you'll be out of business. So thoughts about we believe, that with AI, you can truly put data in the hands of every single person in your business to make better decisions and to answer questions. And for the next five minutes or so, I'll set the scene and then Charlie will show you how through a demonstration. Next slide please, Charlie. So this age old problem of adoption has been around for many years in in the BI world. Essentially, the the the level of adoption of intelligence within within an organization has has been low. Organizations have tried many things in terms of investing data literacy programs. But essentially you see that first let's put the scale on here, maybe the first couple of years, you know, very low, incline there in terms of who was using business intelligence in organizations. It's really the IT departments, the data departments and then just tracking ever so slightly into business departments. And that all changed a couple of years ago with the advent of ChatGPT and the generative AI wave. Now everybody in your business is aware of of of what is available, what technologies could be available. And so there's this, you know, this this this desire, this need to access data across to your organization. Unfortunately, the the curve of adoption, is still low. Charlie, next slide, please. So the problems, still exist from a from a, you know, an adoption perspective. AI isn't a silver bullet. You know, from an adoption perspective, we're still trying to work out how do we how do we drive that, how do we provide a self-service environment. And the key to that adoption is really trust and the key to the trust is accuracy. And then accuracy, particularly from an LLM perspective, that's critical when you start looking at analytics specifically. So you need to drive the accuracy. That'll drive the trust. As an aside, you need to work out how to remove the backlogs that have existed for many years in a in a BI environment. Next slide please, John. And so from ThoughtSpot's perspective, our approach is to really meet those business users where they are. And I say business users, I also mean, you know, people in the data team, analysts, business analysts, everybody really whether they're in HR, sales, finance, leadership, and to build a platform of trust and accuracy of AI and B together that will ultimately allow adoption to take place. Next slide, please, Charlie. So how how does ThoughtSpot do this? Essentially, we're looking to provide AI and BI to everybody through wherever they are in your business, whether that's within a workflow, whether it's within an app, whether you need to go to a, you know, a a business intelligence environment to find it anywhere really, but also to provide this concept of a of an AI analyst, like, you know, like your own personal analyst sitting on your shoulder that you can converse with in natural language to, ask questions of your data. Next click, please, Charlie. And underpinning these these apps and agents are a set of capabilities that allow for things like change analysis, alerting, outlier detection, trends, the ability to drill anywhere within your data from any start point. And maybe having a a start point such as a a a dashboard that may be within a system or it may be within embedded within one of your apps or it may be on your mobile phone or phone or a combination of all three. Next click, please, Charlie. And and these elements are built on a, I guess, a suite of data tools and capabilities that allow you to prepare and integrate your data, so that it's trusted and governed. Prepared and integrated, not moved out of your underlying, you know, cloud data warehouse structure, but from an access perspective is trusted and governed. And Charlie is gonna show you multiple elements of this in a moment, allowing Spotter, the the name of our AI analyst, to answer questions for everybody within your business. Next slide, please, John. But just before we pass to Charlie, I think it's important to understand some of the context of why ThoughtSpot is different, why it has shorter implementation times, larger and faster uptake within businesses and lower times to drive that adoption to transform your organization. And so we're very different to a lot of capabilities on the market, which is, you know, use text to SQL conversion. And the way we're different is that we we look to guarantee, accuracy, and trust to allow that adoption. So if we consider here we've got, you know, asking a question on the left, getting an answer on the right, that's great. But what happens in the middle? So the first phase is the consulting phase. And this is really where we're trying to work out the context of what type of question you're asking, whether it's a how or why or what question. And if you imagine a a conversation between you and your your virtual analyst or your real analyst to decide how am I gonna ask this question of my data, so this preprocessing, if you like, with your thoughts on it. Next, we set the business context. Now the business context is is is exactly that. It's taken into account through technology all of the nuances, all of the knowledge graphs, the business terminology that you use specifically within your business. For example, when you say, you know, what is my revenue, that might be pretax. It might be post tax revenue. You say, you know, this week, do you mean Monday to Friday? Do you mean including Saturday, Sunday, etcetera? And and when we give all of that information to ThoughtSpot to allow it to use the context of your business to apply it to your dataset. Now the third piece is the, execution phase. Now this is the real, I I guess, the really clever piece, of of the secret source of ThoughtSpot. And it's this way we it's it's at this stage where we generate the search tokens that then form part of the optimized SQL generation that goes to do the search against the underlying data. But the the the reason that this is the real secret sauce is this is this is what ThoughtSpot has been built on for the last thirteen years since inception. You know, we we painted it, relational search. It it's our capability. It's super fast. It it it works. And so if you can find a way to go from your natural language question through all of the context that we've said and deliver an answer that is accurate and trusted, that then gives you the capability to, to scale and to drive adoption. And then the final step is visual best practice. So the so so when normally when you get an answer, you go, great, I asked a question and now I've got an answer. Whereas, really, within ThoughtSpot, the answer is really just the start point of the next question or the next point to start jumping off. So you can, you know, drill to a more granular answer. You can change the context, go to a different part of the underlying data model, and you also have the, the capability to, provide feedback with a, you know, human in the loop feedback mechanism. But the key here is that it's a conversational interface to that answer. So, you know, you you you just continue in a conversation with your analyst, and then you loop around to the start, and and and we go through the set process again. So that that's really context of, what I wanna show here. We're gonna go to a demonstration in a moment, Charlie. Charlie's gonna drive that, and he's really gonna, I guess, bring to life some of those capabilities of Spotter and some of those capabilities that, you know, until now probably haven't been seen to exist in in technology. And with that, Charlie, I'll, pass it over to you. Thank you very much, Stuart. Okay. So what we're gonna do today, as Stuart mentions, is run through ThoughtSpot, but focusing a lot on Spotter, our AI agent. We're gonna split this demo into three parts. First is what's the Spotter of today, the latest and greatest of what Spotter can do with a couple of new little skills that Spotter has learned, but also explain to you kinda how it works under the hood. Second part will go into ThoughtSpot embedded, but not how you may have seen it before. You've been able to embed ThoughtSpot and all ThoughtSpot components, within applications, whether that's internal or external really easily. Before a lot of lots of customers that are doing that, but I wanna talk about what we call bodiless embed. So if you've already invested into AI and have chatbots within your business, again, internally or externally, how do we meet you where you are and have ThoughtSpot integrate into those workflows? So that's a really new and exciting thing I wanna show you. And then the third piece is a spotter of tomorrow, some features that are not available today but coming soon and really exciting new skills that ThoughtSpot has learned. So we're gonna start. ThoughtSpot of today, natural language agent allows you to speak to your data as Stuart has already mentioned. But the first part I wanna talk about is actually improvements in data fluency. So ThoughtSpot has always been able to answer questions of your data, but a use case of a new employee or a new dataset, for example, so I'm looking at this new dataset. I might not know what my first question could be. So I might wanna ask questions like, what can I ask ThoughtSpot? ThoughtSpot can understand this as an question of intent to give me information back, relay with me like an analyst or like someone from the data team where I might ask them on, you know, Slack teams, email being like, yeah, what what's available? So we can take this. It's gonna ask ask that question for me, and then I can take, some of these suggestions and go, actually, okay. Maybe I wanna ask questions about stores. So show me sales of stores last year. Spotter now understands that this question has intent to return a visualization or return data, goes through the trust layer as Stuart has already mentioned and returns a visualization back for me. Now just before I kinda move on from here, obviously, we're focusing on the data fluency fluency part here is massive. Right? Really just lowers that barrier to entry to you, interacting with data, and you need much less training when you, have people. Now the real differentiator with Thoughtspot, though, is when we're translating this text, we're not doing text to SQL. We're doing text to these easy to understand tokens, which then executes the SQL. A lot of text to SQL, products on the market are amazing, but they're amazing at accelerating technical folk. I'm a data engineer at heart, so if I use them, I can go and validate the SQL, and I can understand if it's correct because I write SQL every day. But most people, if not the large majority of people, aren't in that space. They don't know SQL. They don't know data modeling. But what anyone can do is understand these tokens here. So when I ask a question, it translates to this so I can easily understand it. I can then go to here and ask a follow-up question and go, it only show the 10 highest performing. Got me. So pardon me. I'm saying only show the top 10 highest performing. What this does is then translates that into a keyword in Thoughtspot, which is top, limiting this to the top 10 here. So that's great. We've gone from data fluency to answering questions within here really easily. Now the next thing is about actually visual storytelling. So I can actually get to this point and go, right. I've got the answer that I want now. Actually, I wanna pin, or maybe I wanna save this to come back to later. But usually, the next step once you've got this data is the fact that, I wanna go and change the visual type. Alright. So now what I wanna do apologies. It just froze there for a second for me. Stuart, if I can just bring you back into the conversation quickly, I wanna show one of ThoughtSpot's qualities on what I wanna involve involve you. What's your, let's say, least favorite chart type? Pie chart, definitely. Yeah. Definitely pie chart. Pie chart. Yeah. I probably agree with you there, Stuart. So I can actually ask Spot now and go. Pardon me, guys. Okay. Bear with me a second. I'm becoming less of a fan of the pie chart already, Charlie. Yes. Save sure. Apologies. She's got some background noise for you as well, Stuart. With me. Okay. So apologies for that, guys. Just gonna, run that again quickly just to show this new feature. There we go. So what you can see now is that we've gone from asking multiple questions, but then ThoughtSpot has this new skill now where it can change the chart type. This is where we're really taking ThoughtSpot to become that multi agent agent. Right? There's many skills when it comes to BI. It's not just returning visuals or returning data. This is another one that's really gonna take it to that next level. What I wanna kinda segue into next is what I mentioned at the at the top of this call with the whole embedded nature. I'm gonna reset this here and just show you this ThoughtSpot search bar. We have loads of customers across the globe that are embedding all different types of components of ThoughtSpot, and one of those is ThoughtSpot. You can do visualizations. You can do live boards. In here, you can fully customize this as well. You can change the name of it so it's not ThoughtSpot. Change the icon. Do whatever you want so that people, your customers, or even internally think that you've built completely bespoke products, but you're actually using ThoughtSpot to power that. However, I don't wanna focus on that today. What I wanna focus on today is actually what I mentioned before, what we're calling bodyless embed. So here is an example, very plain web page, but this could be your internal systems or an external product that you have that you potentially monetize to your customers. Within here is a chatbot. This chatbot here has two skills. It has what we're mimicking as your skill, your internal chatbot, and then ThoughtSpot as your data bot. If I wanna ask a question like, what are some strategies to improve my sales? This is not explicitly a data question. So imagine this comes to a fork in the road, and it goes right. Is this a question for my internal skill, or is this a question for the ThoughtSpot skill? You can see that this took the first route. So when it takes this first route, it goes gives me some suggestion that goes right. I'm about to do more customer feedback, more loyalty programs, more promotions and discounts. And I go, that's great. So I wanna apply some of that, but I wanna apply it to some particular stores. So I can ask a similar question to before. Something like show me the sales of all stores last year. This is now explicitly a data question. So when it gets to that fork in the road, it's gonna initiate that spotter, the API. So this is all embedded through APIs, and when it's initiated, it then embeds ThoughtSpot spot content within here live for you, which is super, super powerful. You can then have follow-up questions within here and say things like, yeah, only show for the East Region. Meeting you in your AI workflows. If you've already invested into AI initiatives within your business, we wanna meet you where you are so you can augment that with data within here. Now this is one way to do that. Right? Using Spotter as a kind of binary on, on, or off. What I prefer to do is actually a new way of doing it, which I like to call or I say I like to call is called data augmented generation. Let's quickly there we go. So if I ask the exact same question again and I say, show me strategies to improve our sales. You could argue that this question could be improved with data. Right? You could argue that it is both a skill required from your internal system, understanding the nuance of the business, understanding the industry that you're in at a very high level, but it would be improved if we augment it with data. So what this process here is is simulating that back and forth between the business and the technical team when they're trying to come up with strategies to improve things like your sales. So you can see what the overall agent is doing is using multiple skills to go, here's some questions that I might wanna answer. And it's actually before it's kind of got to that point, it's going actually to get what you really want. Here's some extra questions. Like, what are the average sales for each product? For each item, what's the highest total sales? And it's combining all of these skills to put together, so it's gonna give me suggestions similar to before, but as the name suggests, augmented with data. Before the suggestions came in, they were very generic, very high level. And you'll see now once it's done, it's gone away, collected all the data, and returned something for me. We've got a bunch of suggestions with data pulled directly from your business, directly from your data warehouse. This is all powered by a ThoughtSpot pulling in data, pulling in visuals, and pulling in the LLM to give you summaries on these suggestions. So talking talking about top selling item types, average sales for products, sales over time, and so on. Giving me suggestions, summary, but the cherry on top gives me also a light board embedded within this experience, which again is fully interactive if I wanted it to be. This is all available today. ThoughtSpot, this is using all of ThoughtSpot skills, but just in a different way, a different front end, but meeting you in your workflows rather than having people move between from tracks. That is everything on the embedded point. So I've gone through point one, and we've gone through point two. What I wanna show you next is actually ThoughtSpot of the future. So all of that can be done today. If you've got any questions about that, I can see a couple have come in, and I'll get to them after. But if you wanna reach out sort of separate to this webinar, more than happy to help you just discussing AI strategy and how ThoughtSpot works or how we can get this implemented for you. But what we wanna do now, like I mentioned, is talk about ThoughtSpot of the future. Now what ThoughtSpot is amazing at as of today is talking about the what and the when. What we wanna take it to is the why. So I can say things like, show me monthly sales. This is something that ThoughtSpot can do today. Nice and easy. Just wanna see a trend of my sales. What you'll see at the bottom is we've actually got an extra skill we've put in called change analysis. We can actually initiate that now or in the very near future and say things like, why the recent drop, ThoughtSpot? Instead of having to go and do that analysis all myself, I can see that my sales have dropped off recently, and I wanna understand why. That is the first question that people are gonna ask when they see this insight. It's not drill down to this point. They're gonna wanna know what's made that change. This is a feature that's available in ThoughtSpot today, but you have to initiate it from somewhere else. We wanna bring all of ThoughtSpot skills into the conversational experience, and you can see that it's giving me a summary of, you know, your revenues dropped by 85%. These are some questions I think that are gonna help you get there. It then does some statistical analysis and goes, based on stage, this is what's been dropped. This area has been driving it. Based on the created date, this is what's driving it. Based on the region, this is what's driving it. So this can give you instant insight in the hands of all of your users into why things are changing so people can make data driven decisions and go and speak to the right people to make sure they can either, you know, learn scope, learn to gain growth or learn to avoid going down. And it also gives you a nice summary down here, which is everything above, but just in a really short part way, all powered by LLMs and all powered by ThoughtSpot AI. Final piece that I wanna show you is just another skill. So, again, Spotter is an agent, but it is splitting into multiple agents underneath the hood, gaining skills day by day. I can come into this LiveBoard checking out my, checking out my KPIs, but I can also initiate Spotter from any visual. You don't have to ask a net new question. We can come in here and I go, right. I'm a new salesman, just joined the team, and I wanna come and take a look at this particular visualization. Instead of looking at my sales price over a date for a KPI. Now there is a new skill that I'm gonna enable down here that we're calling deep analysis. This allows you to really converse to it like a human. It allows you to ask multilayered questions that's both relevant to the data but also relevant to the business. This is almost simulating the embedded data augmented generation, but within Thoughtspot. So if you've not built your, AI box and you don't wanna integrate in there, we'll bring that to you to sit within Thoughtspot. I'm just gonna copy and paste a long question here. But a question like, I'm a new salesman here. I I want to drive sales up for the best performing vehicles, types, and any extras that can be sold with high mileage on the odd meter. Please advise. So this is now not just a data question. Right? I'm asking for advice. I'm asking to please take a look at it more holistically, do all of that analysis in the background, and bring it into one place, and then present it back to me so that then I, one, become more knowledgeable, but make much more data driven decisions and not have to you know? Gone are the days where you have to sift through many, many dashboards, and gone are the days are gonna be a point where you don't have to sift through anything but just asking natural language questions. So you can see that the agent here has has gone right. You've asked this question. I'm gonna break this down by these questions that I think are gonna help you. Gives me monthly selling price trend, breaks things down by car types, break things down by make, break things down looking at thing the selling price of cars with higher odometers than a hundred thousand kilometers. All of these just from one question also then giving me a summary and additional questions that relate to this that I might wanna ask to then become again, get to that next point where I'm making data driven decisions. Thank you very much for listening to me going through all of the features we covered today. We covered the embedded, and we also covered a couple of new skills that will be coming. So I'm gonna pass back to Stuart just to wrap this up. Excellent. Thanks, Charlie. Could you just pop up, slide eight, I think it is, with the, the the the forward market view? So thanks, Charlie. I think Charlie covered some, some groundbreaking and market leading capabilities there. It's it's interesting. I I work at ThoughtSpot, so I I I hear this all the time. But there's there's some I think there's some basics here to consider that when we talk about you know, I talked about adoption, and and you you need an easy interface to allow everybody to adopt. Well, that's the natural language interface. But from a technology perspective, you also need to have something that can scale infinitely, and that means not having to pre aggregate or predetermine what the answer to a question that hasn't been answered yet is. And that for me is the the market leading capability. So we have through the, you know, underlying technology of the cloud data warehouses, we have the the the the speed, the scale, the governance of access to your data, whereas ThoughtSpot can now take that with our natural language interface and Spotter and allow you to access any part of it at any stage. So what was, of course, Charlie's practiced his demo today, I would have expected. He has to make sure that the data is in a format to give him an answer. But he hadn't prebuilt any of those answers. He didn't have to do any pre aggregation of the data or or, you know, additional things. He he could have gone in any direction and asked any question, literally. Okay. So with that, let's just move here. So if if you think we're at the, you know, sort of back end of '24 here, this is this is where ThoughtSpot is now. We've been there for for probably a couple of years. We're moving into this autonomous space, and that that that sort of builds on what we did in search for the for the last ten years or so. We we were the certainly the first to market of a natural language query capability, called out for that in Gartner's last BI Magic Quadrant as the only company to have, you know, multiple customers in production. But we've evolved through that side, through the spotter, and we're we're already going to this autonomous phase where you you can have individual skill or capability based spotters talking to each other, really allowing you then to ask any question within your, within your business. You set a task, and that task might be, how do I, you know, how do I take a hundred million of, cost out of my business that that, you know, that split dynamically through the multiple agents, and they'll come back with data augmented answers to your questions. So, you know, if you think of ChatGPT, you can get answers to your questions. You don't quite know where the answers come from. But with this, we're augmenting it with real data where there's a, you know, a % accuracy, a % lineage of where that data has come from to drive your your answer and therefore the decisions that you make. Next slide, please, Charlie. It's an exciting future for us, at ThoughtSpot and, you know, that capability is literally just around the call. So before we move to the, live question and answer session, I just wanted to summarize what we have in the, in the other webinars, the follow-up webinars, if you like. And and the reason we've done it this way is so that you can you know, rather than having to come on a specific day at a specific point, we've we've recorded these, and allow you to take them at your leisure and to, you know, consume them as your leisure. You you convince them if you like, like a a Netflix series, or you can watch them one at a time like the old way of watching TV. So the the the the first webinar, it's led by Cindy Howson, who is our, chief data and strategy officer with the the the the BI legend who is Donald Farmer. And, Databricks' lead evangelist, Ari Kaplan, really talking about how how to get serious about your data strategy with particular with particular emphasis on, you know, trust, accuracy, driving self-service, and therefore, adoption. So I I would urge you all to watch that webinar. It is literally brilliant. The next webinar that we have is, Tom Cronin from, he's the tech leader, EasyJet, a a recent customer of ours, out of out of London. And he's talking about how AI driven analytics have really transformed decision making, within their organization. So if you think back to that early adoption curve that I showed, you know, most organizations are maxing out of maybe 20%, but, you know, maybe actually 25%. EasyJet went from 25%, of daily less than 25% of daily users to in excess of 75% of all of their organization using data through the, ThoughtSpot interface in a matter of months. So that that one is a really great story, but you see some of the, you know, conversations to as to how they got that. And then the the the the last webinar, webinar four, is actually, some of our technical experts at ThoughtSpot unpicking what made Jetstar EasyJetstar, EasyJet so successful. You know, which frameworks did they use with method methodologies, which playbooks, what were the lessons they learned, and what are some of the key takeaways that that we might look at now and pass on to, you know, new customers and so that they can, you know, accelerate even further their adoption of the ThoughtSpot platform. So those are the, those are the, webinars that are coming up. There'll be details of those in the follow-up, email to this webinar with links. I think they're also in the doc section, also in there, you know, details of, you know, how to take out a trial, plus some other artifacts including the, 2025, ebook trends on on AI and Gen AI with regard to analytics. So before we wrap up, I'd like to go to questions. I think there's quite a lot of questions to get through. Linda, I think maybe I will simply scroll through the questions here. And, Charlie, you and I between us can, take them. Yep. So the first question, I'm definitely gonna pass this to you, Charlie. Can you create comparison charts side by side? For example, a pie chart or another chart if you don't like pie charts, showing the top 10 stores for each year over the last ten years. Is that I I'm not sure I wanted to demonstrate it, but maybe get it ready in the background. But is that something we can do? Yeah. Absolutely. And so, obviously, you will have seen through the demo that from one prompt, it can generate multiple visuals. Right? So that that's one way of doing it. But you also saw at the very end of the demo, you know, it ThoughtSpot is built on search, but it's still a BI platform. We still have a live capability where you can have all those visuals in one place, as many as you want. Alright? Well, answer your question. Yes. You can do it all in one, but you can also plot plot this on a live box. Perfect. Thanks, Charlie. The great thing about having you on the line, Charlie, is if there's any easy questions, I'll take them. If there's anything difficult, I'll obviously pass it to you, and and you can either pass or fail at that point. Alright. Next question. Now what's the format of your database? Now I I would interpret that as two ways. It could either be the format of our internal structures within our SaaS application layer, but I think it's probably meaning what's the format of the underlying database. Charlie, why don't you just give a a quick answer on that? Oh, yeah. Pardon me. Just Or or I can. I'm happy to. Very much. I tell you what, you you you carry on doing what you're doing there. I'll answer that question. So so ThoughtSpot as a tech technology essentially lives over your cloud data warehouse. So whether that's Databricks, Snowflake, GBQ, Redshift, whatever it might be, and and and connects to almost every underlying database without extracting data from that. So there's no concept of pre aggregation. We connect to to all of those, It's a long list of of underlying databases that we connect. And then there's a follow-up question. What LLM model did you use for your agents? And there's a there's a couple of, LLMs that we've used. Charlie, do you wanna go through the detail? I know we're we're we're taking new ones on board at at this moment. Yep. Absolutely. Yeah. I apologize that my iPad just took over my laptop, so I lost couldn't see anything for a second there. Yeah. So, just on the last one as well, just a final point. Under the hood today was Snowflake with the star schema, but we like like Stuart mentioned, we do all all warehouses and all analytical models we support. What LMM today? So today, we're powered by GPT, but we also support Gemini and Snowflake's Cortex. It doesn't really matter. We ask the LN to do a simple job, translate the text into the tokens, and then you apply some training as well. Most of the intelligence sits within ThoughtSpot. The LN is just a nice little, cherry on the cherry on top of the cake. And I think Yeah. From from that as well, Harsha asked, like, where is it hosted and what data is sent out of the organization. So it's hosted with you know, we have our own OpenAI server service, and the only like, we have all of the accreditations for security when it comes to using OpenAI, or Gemini or Snowflake Cortex. But the only data that is sent to them in the moment is the question and some sample values. It is then removed and not, persisted at all. Those sample values can also be turned on and turned off in terms of what you pass. So if there's PII data, you can still search on it, but we don't pass the sample values to the other. The only reason the sample values there is just to try and improve the accuracy, but you have full control over what's passed passed over. So you can actually have it, so it's just the text of the question is passed. But, again, it's not persisted. Yeah. Cool. Thanks, Charlie. And and and just, correct me if I'm wrong, but our our interaction with the LLM is essentially from the natural language, validation of what it is we want to use to build our token search based on. It's not that we're using the LLM for anything subsequent to that or to generate the SQL. We're doing that, but we're doing it with the with the the the natural language conversion, if you like. So that's that's our LLM integration partners. Of course, there's all whole host of other AI pieces within the technology, but, is that correct? Absolutely. Absolutely. How do you, as it is, I said, how do you manage PII and customer data privacy? Yep. It's just hosted within where like any other BI platform, when you bring because we don't actually store the data. We just put down all the data within your warehouse. So all the security of your warehouse, we also inherit. And, again, you can turn things off whether so we also can index your columns to improve the search speed. When they're stored encrypted, you can, again, turn that on or turn that off. So, essentially, we inherit everything from your warehouse, and we don't bring data in. So PII and customer data privacy isn't an issue. And then, obviously, RLS, all of that stuff. We have all the security measures you you'd imagine from a BI platform. We've got work with some of the biggest banks in the world when it comes to that. Yeah. Cool. And just to add to that, Charlie, of course, we're we're we're we're operating out of out of out of Sydney in the region here, but, you know, we're also a global organization, and certainly with our European clients and and actually a couple of our clients here who have European, basis as well. From a GDPR perspective, you know, that that that's a very valid standard that we comply to. And and I know recently we've just done for one customer here a detailed DPIA assessment of of of how we fit and, of course, we pass that with flying flying colors. So I'm always happy to engage in those discussions. And and maybe from a, you know, the the there's a there's an add on question there, Charlie, maybe around security, and and and maybe authentication to the platform because I know that's always a you know, from a a compliance perspective, that's always one. Can you just add a few words on that as to to exactly what you're gonna do then? Absolutely. Again, yeah, because we don't bring the data in, we can also inherit kind of the functionality of your underlying warehouse. So, for example, Snowflake has its own type of OAuth and Redshift is has its own type of OAuth, for example. We can also inherit the SSO capabilities. We also can inherit from your IDPs. So you can manage all the security kind of outside of ThoughtSpot and then have that interaction from ThoughtSpot from from our side. So, from an internal point of view, most of our customers use SSO, same way that we we log in to our ThoughtSpot internally. From an embedded point of view, we actually have something that most of our customers use called trusted authentication. So you manage the authentication from your front end, pass that token to ThoughtSpot, and then we resolve it on the back end. So you don't have to have two sets of authentication. You got a seamless login through your your login. ThoughtSpot appears completely white labeled, and it looks like, you built your own product. Cool. Thanks, Charlie. I'm just looking, over that question here. It's great to see that Spotter can directly generate insights, but can I use Spotter for SQL generation if I wish to? As of today, no. But I would say that that's not really the use case for for ThoughtSpot. ThoughtSpot is really accelerating your business on data that is already ready. We have a SQL IDE capability with its own AI where you can generate generate SQL there. But the ThoughtSpot point is you're already you're at the business end. You're asking having data ready, so it's not ready for SQL generation. But maybe in the future, but not today. Got it. Okay. Thanks, Charlie. And there's another question. I love this question. We have loads of data collected over ten years from, monitoring status of hardware on prem and cloud based services. Can ThoughtSpot tell me in advance, I. E. Notify me, of which of the services to keep an eye on that might have problems such as CPU usage going beyond 90% based on data trends built up over the last ten years? Yes. So we have monitoring, alerting, and anomaly detection through all of our, KPIs. And, essentially, what, like, you can upload that data. It understands it and then can start looking for for things like essentially, that's trends, anomaly detection, that kind of thing. So you basically bring in that KPI, and Torso will be able to tell you. Cool. Thank you, Charlie. Another question here. Can we use ThoughtSpot with rolled up data? Oh, I did see this question. Who was it that asked? It was Miley. Miley. Yeah. If you could expand on that in the chat, Miley. I just I have a way. I think I'm gonna answer it, but I wanted to make sure that I'm fully understanding your question. Okay. So whilst we're doing that, the question was, can spotted compared today versus same time last year, this last week, this last month? Absolutely. Comparison questions, is really easy for Spotter as the what have you said here, Shelley? As the underlying relation search engine as a Versus keyword. Okay. I guess the other thing to mention there, Charlie, with I'm gonna be careful if I say things that I that you know more about than I do. But, what about the dynamic function capability for, you know, for building something that maybe isn't a standard? Does it could that could that could we extend that question to do more things with those dynamic functions? Absolutely. Yeah. So Thoughtspot has 300 keywords that'll answer more than the majority of your questions. But there's always some some ones that are, slightly more complex. And with that, we have formulas, exactly the same as you'd have calculated metrics and other BI platforms. But if you don't have the results in the warehouse, you can build it in the logical layer in Thoughtspot, right, just as you'd expect. But what we can also do is we can actually leverage functions within your warehouse. So we can call SQL functions that are bespoke to Snowflake, Redshift, Google BigQuery to do much more advanced analysis and call that at runtime in Thoughtspot as well. Perfect. Thanks, Charlie. I'm just, conversant of the time. We got about two and a half minutes left. If we don't get to the questions today, we will check the chat afterwards and and send out answers to the question. I think we've covered them all. I just wanna go back to my guest my guest, Charlie, on that. Can we use Spotter with rolled up data? Only because I had this, in a in a in a recent, customer situation. And I'm I'm gonna use rolled up data as aggregated data. Oh, I see. And and I guess the the concept here from from my understanding is we would look at each of those aggregations as as as a table or a set of tables and and would process it in exactly the same way as we would, you know, instead of underlying structures within a any any cloud data warehouse. Is that is that fair? Yeah. Absolutely. Okay. I I guess then the the the limitation of the drill, though, the drill capability would be to the lowest level of granularity within that set of rolled up data. Yes. Absolutely. So, we would always recommend to use the most granular data. It's as when you roll data up, you lose potential questions that can be answered by the business. Usually, you roll data up for performance. Yeah. But we sit on top of all of the biggest cloud data warehouses and write very efficient SQL. And to get self-service, we want quick insight. And once self-service, you need to be able to answer all the questions. So we recommend the most granular data, but we understand that you can't always do that. Yeah. So it's it's it's it's almost, you know, if you if you think that you've put data into your cloud data warehouse, what whatever that might be, whatever flavor that might be, you've put it there for a reason. So to not be able to go to the lowest level of that data is is almost counterintuitive. Anyway, Charlie, I'm gonna say thank you for answering those questions. I wanna say thank you to all of you on the webinar for, attending today. I hope hope you've enjoyed the discussion. I hope you've learned something. I hope you're excited about GenAI as as as we are. It's been a real pleasure, spending time with you, and please, take a look at those additional on demand webinars. There's a a ridiculous amount of good information and insight coming from our customers and from the market in general and from our partners. So, with that, I'm gonna wrap up the session and say, thank you very much, and hopefully speak to you all soon. Thank you. Thanks very much for your time.