Welcome to our webinar. Today we're going to be talking about five key strategies to unlocking Snowflake's full potential. So Snowflake started as a, a, a cloud based data warehouse. It very quickly grew into a full data platform was very exciting about the vision of what Snowflake wants to do is as it adds more and more capabilities, it also wants to roll horizontally. There's more stuff being added to the platform every quarter, every month. It seems like almost every day in terms of all the new releases they've got. So we're gonna give you five examples of things that you could be using in the Snowflake environment or something, a little asterisk number five, something that is very close, to the Snowflake environment that can give you guys significant value, something that we're really excited about that we're using internally as well with our clients. I'll do some introductions. Roger, who wouldn't mind going to the next slide? Thank you. So I'm Robert Curtis. I'm just here to say hello and introduce our speaker. I've been with Interworks for about twenty years as the managing director. I'm based out of Melbourne. It is a pleasure to meet you and for all of our returning friends, hello again. Joining me is Roger Garcia. Roger is a solutions architect. He's been with Interworks for over five years. He's based out of Sydney. Roger is originally from Spain. He's still glowing from the World Cup win. I would love to know what that feels like either as someone born in the United States or as a a newish twenty years Australian, maybe one day. The series of webinars that you are joining is part of a multi presentation, multi event effort that we're doing after Snowflake World Tour. There is the semantic white paper that you can access. You can go to interworks dot com and find that. I wrote that. So hopefully you guys find that valuable. I'm trying to write that from more of a business lens of why the semantic layer is so important. We have already done the future of analytics, the journey to BI three point zero. That was on the nineteenth of August. Again, that's on the Interworks website. We have done the preeminence of the semantic layer, the webinar version of it. Today, are doing unlocking Snowflake's full potential. Next week, we are doing AI at scale. So we're thinking about your AI COE, change management, governance, all the things you need to think about in terms of organizationally getting AI productionalized and ready to go. And then we'll take a little bit of a break, I think a week, and come back and do five essential AI use cases everyone should do. So AI is super important. There's a lot of value, but where to start? We're giving you those answers. So hopefully you can join us on all of those. You can register on interworks dot com. You're good, Roger. You register right at interworks dot com for any and all of those. Next one. So just a little bit about Interworks. Obviously, you folks that have come have heard this quite a few times. But basically what we do is data strategy solutions and support. So if you're trying to figure out the direction that you want to go, maybe mapping a course or a vision over the next three to five years, we can help. We can help you with the more tactical assessments. We can help you with more specific domain, like architecture reviews, all that kind of stuff. But the deep thinking part of this in terms of how you want to make decisions. Then from a solution standpoint, whether it's building the foundational elements like your warehouse or your pipelines or your governance, the implementation into a governance platform, we can help you do all of that. We can add value top of that. You'll see at the very next slide, we'll talk about that too, in terms of AI solutions, analytics, advanced analytics, augmented data science, machine learning, all of that kind of stuff. And then the support's perspective is we can help you support your applications or the things inside of those applications like the pipelines themselves, if you don't wanna manage those. We can help support your communities. You wanna build a data culture. We can inspire, we can lead, we can gamify. We can also help support individuals. So building skills, capabilities, capacity, that kind of thing. A little bit more detail on that, Roger, if you wouldn't mind. This is a bit of a journey, a life cycle through all the services from a data perspective. Again, starting with strategy, building the foundations, expanding and adding ROI, all the way to sustaining and growing. I'd certainly love to talk to any of you about that. A little bit more about Enderworks, and I promise I'll shut up and I'll turn it over to the man of the hour, Roger. We've been in business for more than thirty years. That's very unusual for an SI like us. We're global, but we're not the biggest in the world. We're very focused on data. We love doing what we're doing. And so we have a tremendous amount of experience, probably about three times as much as most other companies of a similar bend as us. As a result of that investment in excellence and success and continued success, seventy five, maybe seventy seven of the Fortune one hundred have been or are IntoWorks customers. I'd love to tell you all of the great stories, but obviously I can't tell you all of them because of NDAs, so we're very respectful of our clients' privacy. But there are a lot of great stories on the IntoWorks website, case studies, other things that clients have graciously allowed us to share. So it's a great place to go learn a little bit more about us. And while you're on the Interworks website, you might also go check out our blog. We write a lot of stuff, whether it's our white papers, blog articles, Roger is a prolific writer on a lot of the cool stuff that you can do in data and AI and all sorts of things. So there's a lot of great content on there. We get about four million now in terms of the number of views for our blog. And we've had the pleasure of working with a lot of customers across every vertical you can imagine. And we've collected a lot of different awards. The Forbes Small Giant being one of those. Also on our website, if you wanna do some reading. We can hop over. And here is where I'm going to tag team in my colleague, Roger, to give you five ideas on how you can best take advantage of all the cool things that you can do in Snowflake. Just before Roger gets going, we do have little chat or if you wanna throw your question to the Q and A, he'll go through his presentation without pause. And then if there are time for questions, we'll do that at the end. So if there's anything along the way that you are interested in asking questions, chuck that into the chat or the Q and A, and then we'll come back around and I'll Q and A for him. Thanks, Roger. Over to you. Hi, all. So as Rob mentioned, I'm gonna be talking about five strategies to help you unlock potential from Snowflake. Now the keyword that I want you to all remember is accelerators. Okay? I will be showing you five things that you can start doing right now, so you don't have to wait till next year to start getting ROI. We're talking about features and strategies that you could do to just accelerate the value that you get out of your implementation. I'm gonna first start easing into these. There's gonna be five themes that are separate, and to start easing into it, I'm gonna talk about one that is perhaps a low hanging fruit or might be less funky at first, but it can drive tremendous business value, And that is adaptive compute. Now, if I take a step back for a second, typically if I wanna run any queries in, whether it's Snowflake or any cloud platform for that matter, I would need the components that you see in my slide. If I go from the bottom up, I need to have that infrastructure, right? I need to provision it and doing patching. I need to make sure that I configure that compute, I size it accordingly, in and out, and up and down. I choose also what kind of compute, and then I execute the queries against that working unit. That's typically what I would need in a very, very, very simplistic way. Now, if you're running on Snowflake, Snowflake is already doing something for you right now. Snowflake is, as a SaaS application, is managing that infrastructure and doing the patching for you. And the rest, the items that integrate is actually items that are up to you as a client to just set up and configure as required. Now as of late, as of last year specifically, Snowflake has released a new kind of compute that they call it Adaptive. Now the concept is very intuitive and very easy and self explainable to understand, which is that what Snowflake does is it actually just takes more work out of your plate, and it just handles that for you. So as you see, with this Adaptive Compute, which is a new type of warehouse that you can create, Snowflake will do lots of the heavy lifting for you. So much so that all what you need to do is just simply speed up that adaptive compute, and then execute queries against it. Conceptually, it's very simple to understand. What I want to give you some intuition on is when to use it and what are some of the benefits of this kind of adaptive compute. Adaptive compute is useful for unpredictable workloads. So think that you're launching a new POC, or there's a use case where there's spikes in volume, but there is no fixed pattern, it can go up and down. Well, adaptive compute can help you there, because it can just automatically increase cluster, increase the size, scale up or down as required without you needing to do any fine tuning or configuration. So in that sense, it's doing quite a lot of the heavy lifting for you. So unpredictable workloads, good candidate for you to explore adaptive compute. Secondly, changing throughput. So if you have, for example, an ingestion job, that volume shift from week to week, then again, that alteration, that change in the volumes, is also something that is a good candidate for you to use with Adaptive Compute. A third one is if you have jobs where you have mixed BI and ETL workloads, where you're blending these queries into a single compute unit. Incidentally, if you have that, I would probably pause and urge you to actually consider, because it would not be very good practice for you to be using the same compute unit for both BI and ETL jobs. And so it's probably something that you might wanna reconsider. But if you do must have that blended approach, then Adaptive Compute can also help you with BI and ATR workloads. And lastly, if you have simply limited admin capacity, so if your team is all hands on deck, busy, you know, you don't have the capacity to just be managing that compute management, then Adaptive Compute can help you with that, because it does that heavy lifting. Now, if you only have a handful of warehouses, then adaptive compute might not necessarily move the needle a lot from you from an admin perspective. But if you have several pipelines, multiple ingestion patterns, several use cases, AI and BI, all combined, you probably have stumbled upon that scenario whereby you need to just often manage a little bit your warehouse layer within Snowflake. Myself, with some of my clients, we do monthly optimization, which means you need to adjust, see what warehouses are required, you need to adjust them, any configurations that you need to do, optimization, etcetera. So if you find, if you have a decent amount of workloads or compute that is required, I would strongly recommend for you to consider adaptive compute. I'll say one last thing. Of course, I don't know what's the measure of your the details about your Snowflake instance, but as you know, compute is the large driver of your bill in Snowflake. So it's definitely one item that if you can look into it and it helps you just sort of optimize it, reduce a little bit consumption, so that can affect as well positively your Snowflake deal. Benefits of adaptive compute, I think they're pretty self explanatory. This automatic right sizing, as I mentioned, scale up or down or in and out. So whether it's multiple cluster, it's a cluster with multiple threads, or if it's just making a bigger box, so to speak, that will be done automatically by Snowflake, and then reducing that operational burden. As I mentioned, the administration required with management and speeding up and setting up and configuring worker warehouses is always a task that you need to do, particularly if you have a big problem of pipeline, so adaptive compute can help you in that regard. So it's an easy feature to ease into. It's low hanging fruit low hanging fruit, something easy for you to just start getting value out of Snowflake today. This is in GA, so I would recommend for you to start looking into it if you haven't already. That's enough for the compute. Now I want to shift gears and talk a little bit about Cortex and Snowpack. Now, if you've been using Snowflake for a while, you probably have read or heard these two names before. So Cortex and Snowpack, you've probably heard it before. I want to tell you a little bit about some new features or aspects about these two concepts, because Snowflake, as Rob mentioned earlier, is investing heavily on AI, and Cortex is of course front and center when it comes to that dimension. So I'm going to tackle a little bit and give you some intuition about the ins and outs, the new products that have been released, and some implementations that I've done to just show you a little bit the art of the possible. Let's start with a basic definition. Cortex is in essence a repertoire of AI features. It's a family of AI features. So Cortex in itself in Snowflake is not anything, it's just actually an umbrella concept that is applied to multiple features and multiple services. Now, the common denominator of Cortex is that it's a set of features that help you make sense of structured and unstructured data, answer questions, and provide assistance to either your analyst or your developers. That is in essence what Cortex is, this set of features that have these characteristics in common. Now you'll see four boxes in this slide that explain key four categories of Cortex capabilities, but I want to tell you a little bit about it. Just give you some intuition on how to use it, some examples of implementation, so hopefully you get an understanding of the Cortex family of features. Let's start with agents. Agents, in its most fundamental definition, is a fully managed platform that it just helps you to just have agents that can do all the reasoning and orchestration for you. Now, agents, in essence, act as the backbone architecture that Snowflake uses to power some of the other AI services that I will be talking about later. So I just want to do a basic conceptual interaction to its architecture, so you understand a little bit the pieces of this Cortex Agent architecture. And as I mentioned then, because this acts as the backbone, we'll see some implementations a little bit later today in this webinar. So the overall architecture of a Cortex Agent is as follows. So there will be a prompt if we go top down on the right hand side. The first layer, there are three layers that define this architecture. The first one is the interface. So we can use the user interface, we can use an API, CLI, you name it. Depending on the use case, you might need one or the other. Depending on the appetite of your developers, and depending on the user that will be using this. That's the first interface layer. The second layer is the orchestration. This is where Snowflake brings agents into the table. This is where the LLM does its thing, so to speak. This is where the orchestration happens, the LLM is actually interpreting, taking that prompt, doing coordination, reasoning, planning, and most importantly, tool selection. Now, tool selection is important because the Cortex agent architecture, in essence, it helps translate the user prompt, or the prompt that you received, into the use of one or multiple tools that the agent has access to. That brings us to the operational layer at the bottom. You'll see lots of boxes. Some of these names might sound familiar if you've been using Snowflake for a while, so think Cortex Analyst or Cortex Search. These are all services that you can actually use, and they are, think of it like skills or tools that the agent will have in his toolkit. As I mentioned, these are tools that the agent will be using, depending on the use case and the requirement of the user prompt. Some of them are new. I included them with Asterisks, so it's in, for example, code execution, and it's got all the classics such as CortexAnalyst, search. We've got MCP connectors, web search, custom tools, you name it. The limit is the sky there. This list is increasingly getting longer and longer. So what that means is that the backbone of agentic AI that Snowflake is building is massive. There's some new additions, there's much more in the pipeline, but all of these are something you can start tapping into now. Now I will talk a little bit about implementations and how these architectures are being used by other services in Snowflake. I will talk about this a little bit later, but for now, just make a mental note of these three tier architecture and understand that from an AI perspective, that acts as the key backbone of your AI implementations with Snowflake. All right? I'll talk about this, as I said, implementations a little bit later. Before doing that, I wanna talk about another feature called AI functions. Notice the first word that I included in the paragraph, Steamtags. That's a way it's probably a bit reductionist definition, but I wanted to signal that to all of you. AI functions are, for the most part, syntax. You have your SQL scripts. You have your Python scripts. What you can do with AI SQL is introduce keywords, introduce syntax in your scripts, and these keywords allow you to, in essence, accept prompts or use AI functionality, such as prompts or receive images, and enrich your scripts with AI functionality. But I want to be clear that this is about enriching your scripts. So your developers will be injecting, if you will, some of these AI functions into their scripts to make Descript, their pipelines, more powerful, more dynamic, using more information, more data, richer, smarter, etcetera. Now, let me give you an example of an implementation that I did that really hopefully pushed this into perspective and put some color into these AI functions. So I did an implementation where I had to build a pipeline that turned video into insights. The story was traffic cameras, so cameras that are actually installed in motorways, for example, or certain spots. These cameras capture lots and lots of hours of footage. The use case was I need to be able to find specific vehicles in that footage, and I can do two things. I can buy four kilos of popcorn and watch hours and hours of footage until I find the vehicles that I'm after, or I can build a solution that is leveraging AI, which is what I did here, that helps accelerate that process. The tech stack that I used was largely Snowflake. You see it in the implementation. I combined AI SQL functions, so an example of these functions that we're talking about here. I also use a video vectorization. It's a third party library called twelve Lapse, if you've heard of it. And then lastly, I also use Rack to just enrich a little bit the pipeline. Now, the user would be having an interface, a chat interface, where he or she would be able to type any questions. So I've included at the bottom of this slide an example of the prompt that they would be passing. So a medium truck carrying sand, for example. If we pass these PROM, what the pipeline will do is go and scan all the footage that we've seen, the vectorization, and then just return those values that actually match these tags. And this is what we came up with. You see on the right hand side are frames of that camera. So that's footage from the video. These are frames. And what you see are rent, the frames that have the highest match with that answer. You'll notice that there's a red circle that's pointing the vehicle in question that was found. So what that gave the user is just within their fingertips access to be able to extract insights from video footage. That was largely leveraging AI functions, and that was, as I mentioned, for the most part, just using Snowflake. So again, that's a solution and implementation you can start doing right now. There is no blocker, there's no infrastructure, you can start doing this right now, and I would strongly, strongly recommend for you to do so. Now I'm going to talk about another AI feature from the Cortex family. It's called Cortex Code. Snowflake are actually making a lot of noise about Cortex Code. They call it Coco for short, and there's good reason for it because it's definitely a game changer in many aspects. Cortex Code or Coco, I would like for you to understand it as an agent that is built in, so there's nothing for you to set up. It's just ready. There's no customization in that sense, or there's no setup for you to do. You can customize it, but it's built in with the platform. And the main purpose is to help your engineering and analytics efforts. Now, can you ask Cortex Code what's the capital of Australia? Yes, you can. Is that the main purpose? Probably not. The main purpose is that it's built into the platform, and therefore its key power is that it's aware of the Snowflake environment. Coco can understand the context, can see the databases, can have access to the account wide settings. It can also see the SQL that you're using your workspace, or the workbook that you're using, the workspace, right? And therefore, it provides a chat interface that allows you to interact with a Snowflake instance in a natural language way. So got all the features. As I mentioned, it's chat and QA. I see, for example, lots of clients that use this for Snowflake administration purposes, so Snowflake admins that will be using this chat box to simply list, I don't know, network policies or internal stages or any configuration, any account wide configuration that they want to review. That's extremely powerful. Secondly, cogeneration. So our developers, as Rob mentioned earlier, simply just go ahead and define in natural language what they need to build. That is go ahead and build a store procedure based on Python that does this transformation, pulls data from table X, create a view, don't know, generate a CSV file and store it in an internal state, create a file format, you name it. I explain it in plain language. And then I can either do, get to just generate the code, but then I paste into a workspace and execute, or I can just get Coco to execute it for me, of course, if I give it permission to do so. And therefore it accelerates enormously the development time, because our developers are barely writing any code and just simply approving, reviewing, rejecting, and getting Coco to do the heavy lifting. COCO is largely known, and if you go to your Snowflake instance, you'll see that it's on the far right. There's a little icon with a light blue icon on the far right of your screen if you go to Snowflake website on your instance, but you can also use it from, there's also an application, so there's also a Cocoa desktop application, and you can also access it through the CLI. If you do either of these options, it has access to your local file system, which if you've been in this game for a while, you know that it means that it just amends dramatically the capabilities of Snowflake, because it means that you can read and write files on your computer and not just in Snowflake, so you can start interacting with other systems. And also, it gives you access to external connections, so you can link up other tools, connect to other sources, other instances, multiple instances, different platforms, all of that outside of Snowflake, but having that central developer assistant, so to speak. So Codex Code, extremely useful for boosting productivity among your analysts, lastly in analytics and in the engineering space. Lastly, I just want to touch very, very briefly on Snowpack. So far, we've been talking about technology that allows you to just use turnkey technology or turnkey features. You don't have to do anything. Models are there. You just select LLMs, and you configure it. It's pretty much all done and dusted. It's simple, it's intuitive, classic Snowflake style, right? Now, for some of you, or for some clients of us, machine learning still has a place in the times that we live in. So in essence, Snowpack is a framework that has been around for a while that allows you to just get your hands a bit more dirty from a machine learning perspective. So as opposed to you leaning on turnkey models and turnkey solutions that Norfolk provides, you can also get your hands dirty and do classic MLOps journey that you know, probably you're familiar with. So you can develop and iterate on your models. Snowflake provides the containment runtime if you want to use PyTorch or you want to use, I don't know, MGBoost or any of these libraries to just create your models. You can create ML jobs. You can do classic registries. So if you create models, you can put them in a registry, a feature store, which is a cornerstone of machine learning. You can have observability, and in that sense, you've got an environment that is ready for you to just do MLOps in a way that is more nitty gritty, getting your hands more dirty than just using turnkey technology. There is a place for this. There is value in this. I just simply want you to be aware of it, that sometimes you might want to get your hands more dirty and use more customization. Sometimes you will prefer to just use turnkey technology. So Snowpark is still there, still ready for you to use, and you should tap into it if your business use case demands it. Now I wanna shift gears a little bit. We'll come back to more AI stuff in a second, but just wanna tell you about the third theme, which is Snowflake Horizon. Snowflake Horizon, in essence, is a governance and discovery layer. It's designed to help people and AI, as we'll see in a second, to just find, trust, and use your data within Snowflake or outside of Snowflake. Key things that Horizon helps you with, if you cannot find the right data, if you've got, for example, block storage and you've got Iceberg tables in Snowflake that are pointing to, I don't know, S3 or block storage in any of the cloud services, and you're combining that with data that you've got as well within Snowflake, and you need to find you cannot find that, rather, you need to have that observability. You need to find sources. You wanna be able to track things. That is just not limited to Snowflake. It can also expand the scope. So if you've got silos or there's formats that log that visibility, I. E. Blob Storage, for example, Snowflake Horizon can help you. Also, you get AI correct but useless answers, That's very common in the in the era of AI. And we'll talk about this in just a second, but AI can be generating lots of information that are just useless because they lack business context. We'll talk about that. Verizon can help you there and explain how. And also if you wanna make sure that security travels across platforms. So as I mentioned, if you've got external sources, if you wanna track data across views and objects and tables and databases and schemas, Snowflake Horizon provides that layer of security and auditability. Snowflake Horizon is built in. There's nothing you need to switch on, so to speak. It's just a catalog or discovery layer that is built already in Snowflake. It's available only in the Snowflake UI experience and is designed for you to just be able to look up basic classic Snowflake style, very simple search box for you to find, locate, identify, audit data across all your workflows. Let me put a bit more color to this so we can see how we can use B using Snowflake Horizon. The problem that is fairly I'm gonna put a very simple problem that I think illuminates something that we stumble upon. Imagine you're using Snowflake, and you've got, it doesn't matter, BI tool connected to it, and then you've got multiple departments using it. You'll have your sales department that they wanna track revenue, and they have a definition for revenue, which is what I included on the right hand side. If you have another department, finance, they also want to track revenue, and they have a definition for revenue as well. And lastly, marketing and their campaigns also want to track revenue, and they have a different definition for it. Now, this is a very, very, very common scenario where you might have different departments wanting to track similar metrics, but that they each have different business definitions. This is solvable. We don't need Snowflake Horizon to solve this. That's, you know, if you've been in the game for a while, you know that there's many workarounds on this. You could start putting duct tape everywhere. We could update the initial definition and just change in the data layer, okay, I'm gonna create I mean, this is gonna sound sacrilege, but unfortunately, I've seen that. Revenue marketing, revenue underscore finance, revenue underscore sales. That could be some duct tape that will be added. I've also seen people just creating new BI definitions, of course, so you can create some sort of a local calculation or a calculated field that just creates your own logic, that lives in the analytics layer, that Snow flake isn't aware of. That's important. But that solves your business problem. You can also create a new report and say, Mate, you know what? I'm just going to create my own report, my own logic, my own definitions. Okay, that also solves that problem. And even still to this day, there are some diehats that will just say, I want to use local files and have this definition in Excel or a spreadsheet or CSV file, and just manage that definition separately. As you can see, this creates, first off, a problem because it pollutes the water. We start to see that the landscape starts to become a little bit more complex. There's more items to manage. There is not a lot of synchronization. There's not a lot of synergies, and it seems that some aspects are just simply out of sync. But there's one common denominator about every single intervention that we just described, which is that there's a human in the loop. A human can create these spreadsheets, a human can adjust the data layer, a human can create calculated fields or create new reports. There's a human in the loop that can discriminate or discern or gauge what's the right solution. Now, the problem is not just that we're complicating a lot of our tech stack. The problem is this: what happens when you start using AI? Because in the eyes of AI, revenue is the correct metric. You asked me you wanted to see revenue, didn't you? I'm serving to you revenue. But which revenue? The first point that I found. Remember that AI is probabilistic. Probabilistic in definition means that it doesn't have concrete answers. There's no ultimate truth. It's probabilistic by definition. So we'll try to find the revenue definition that most matches what you need. So how do we prepare for AI and tackling those scenarios of business complexity? Snowflake provides an answer. We can use a combination of two technologies. What we're looking at right now is just a standard definition of Snowflake, right? You've got multiple pipelines, you know, from raw, some sort of a medallion, from bronze, silver, gold, but it's box standard. How do we then solve this first problem? It's a two step process. The first one, we'll create a semantic view. Semantic views, and Rob mentioned earlier that we've prepared material on these, there's contents that we talked about these, I would love, I would personally love to tell you all about semantic views because they're fascinating and extremely versatile and useful, but in essence, semantic views are just an element that we include in the consumption layer of Snowflake. It's a first class type of object in Snowflake. Now what they have is some additional business context. So semantic views are more business context driven than standard views or than just tables or dynamic tables, etcetera. So they're designed to precisely cut off for that additional business context that will be relevant for your users, but also for your AI workloads, because your AI workloads will have additional business context with which to interpret the queries and the requests they've been given. That's step one. Step two is, let's say we know what semantic view is being used and what AI is quoting, for example. How can I track what objects are actually underpinning that view? How do I have that visibility? Well, that's where Horizon comes in. Horizon gives us that lineage capability. It allows us to track and see what lineage and dependencies, what objects underpin the semantic views or the consumption layer in general. It allows us to have that visibility for sources in Snowflake and outside of Snowflake that we integrate with. It gives us that visibility and observability across the entire data state. So this combination of horizon and semantic views can really be a game changer, and I would strongly recommend for you to start exploring it as well. Now I wanna talk, going back to AI, I wanna talk about another product or another feature that you can start using now that is called Snowflake Cowork. Snowflake Core Work is in essence a conversational interface. Remember that I talked about Coco earlier, Cortex Code? That was more geared towards analytics engineers, developers, Snowflake administrators. Cowork is actually aimed at business users. It's your business users, your analysts, but also just regular business users from any department that simply want to get access to data and do so in the form of a conversational interface right from within Snowflake. So CoWork is a product or it's a service that is built in or it's part of the Snowflake universe. Not going go through a lot into details here, but in essence, it's a conversational interface, structured and unstructured data. It's capable of both. This co work is built, and if you look at the third bullet point, is built on Cortex Agents. So remember that first architecture that I told you about Cortex agents, that underpins that it's the backbone of AI? Cowork is actually built upon that. So Cowork is a front shop, so you don't even have to spin up any agent. You can just rely on an interface that Snowflake has provisioned for you, that builds or that leans on that same architecture. What's the objective, or what's the selling proposition, if you will? It's allowing you for your data teams or your business teams to simply move beyond static and stale dashboards, become much more question and answer, but most importantly, act and interact with your data, not just simply consume it. What does that mean? Cowork is designed to answer questions about your data that resides in Snowflake. It can build visualizations, charts, and graphs, and spreadsheets, and whatnot. It's got a lot of capability built in in terms of automation, so you can ask, for example, go ahead and run this analysis. Tell me what's the difference in x y z metric, and you know what, just do these every morning at nine am so it's ready for my meeting. If it does an analysis, can get it to generate a PDF, you can get it to, I don't know, turn it to a PowerPoint for example, if you need it. You can also turn this into a recurring task so it's automated. You can also ask it to do deep research and investigate data sources in Snowflake and compare it and tell me the insights in a way that is much more mature than some other NLQs and natural language implementations that you've seen, and importantly, all from within Snowflake. Now importantly, Cobo can also allow you to talk to other systems and other platforms outside of Snowflake, and to illustrate that, I'll show you yet another implementation that I personally did. I'm going to simplify it a lot just for the purpose of this session, but I want to show it to you at a high level nonetheless. The use case was understanding invoices that are overdue, that are yet not paid, and understand why. The user was given the co work interface, so Q and A interface, and he or she would ask, So tell me what invoices are overdue and tell me why. This is a five step journey. First, there's the prompt. When co worker receives that, the co worker plans to task, so remember the Cortex Agent architecture that we talked about? So we talk onto the orchestration layer. It plans the task, it understands it, it plans it into multiple steps, and prepays an execution plan. And then it goes ahead and it starts executing them. What is the first thing that it did? It used Cortex Analyst, step number three, and quoted Snowflake and identified all the invoices that were overdue. So it went ahead, listed in you that's using structured data, so Cortex Analyst, finding all the invoices that have been paid, that have not been paid, and they're past the due date. Step number one: it stores this information in memory. Then with this information in memory, goes to the next step: Cortex Search. It leans on another tool: Cortex Search. And what it does, Cortex Search, is allows you to actually scan documents. So in this case, it would go through the internal stages in Snowflake, find the PDFs of the invoices that it had in memory, and analyze, extract the information, and understand what happened. And what happened was that there were missing approvals for those invoices, or there were charges that were disputed. Coburg then went back to the user and said, Look, this is what I found. These are the invoices, and these are the reasons why. We could end the loop here and say, Okay, the user got the information that they wanted, but guess what that user had to do after that? She had to go ahead and call the account managers to follow-up with those providers. Instead, she just went ahead and gave coworkers the last step, which was go ahead and update in the CRM these invoices and mark it or flag it for following up. So what Cowork did is, under the hood, with some plumbing done, went ahead and connected to the CRM, so think, for example, Salesforce or monday dot com. It was one of these systems, and just identified the invoice's ID with information that it had in memory, opened those tickets, updated those tickets, and tagged the account manager saying, Will you please go ahead and follow-up with those invoices? All of that is done by a business user, no code, no setup, simply using native Snowflake functionality. That is co work. Cross platform, structured, semi structured, unstructured data, cross platform interoperability. That's an example, a very basic example of co work at play, and that's something you can start doing right now, something you can start, you should be tapping into. I want to talk about one last feature. So far we've been talking about, our last strategy if you will, so far we've been talking about features that are native to Snowflake for the most part. Now historically, Snowflake has been positioned as a platform for data practitioners, people that get their hands dirty with data. Now some businesses may prefer, for various reasons, to actually not rely on Snowflake, but they wanna tap into all this functionality, but they actually want to create an interface that is external to Snowflake, that works seamlessly with Snowflake, but that it gives them a front shop for other business users. So another layer that would build on top of Snowflake to just accelerate adoption of all the technology that we've seen thus far. So Cortex and Snowpack and all the AI and all that sort of thing, but not being leveraged in Snowflake. Instead, it can be leveraged through a tool called Sigma. Now Sigma is, as I mentioned, is a tool that is separate from Snowflake, and it's part of what's been called Business Intelligence three point zero. Now BI three point zero is a new wave of BI features and possibilities, and it's where the market is going in this end. Rob did an excellent webinar. You can go to our website and watch it. I would highly recommend it because it tells you the trajectory, but also the key trends of that BI three point zero wave. Sigma is a product of that new wave, of BI three point zero, and what Sigma does is, in essence, becomes a layer, an AI app layer that is built on top of Snowflake that can act as your one stop shop for all your AI functionality for business users. So if you want to leverage all what we've covered so far outside of Snowflake in a business friendly or different interface, Sigma could be an option. If you know Sigma, that will probably sound familiar. I'm just going to give you two or three items or talking points for you to consider, and then I will urge you to also explore Sigma, because it's a very compelling proposition. Simple terms, Sigma allows you to do a degree of data modeling. So from Sigma, you can interact with Snorbly. You can do basic standard semantic layer builds, so calculated fields and whatnot, but importantly, importantly, you can do one thing that was quite sought after in the previous waves of BI, which was write back. It can write back natively into Snowflake, so your business users can consume data that they extract from Snowflake, and then they can just go ahead and write back into it. And I'm not talking memory, I'm talking persistence. So this is not cookies or caching, it's persistent. It's actually writing into Snowflake. No SQL, no code, no Python, nothing. Just from an application or just from an interface that has been business friendly. That is one of Sigma's proposition. Sigma provides governance and admin, so it supports rollable security and all the role based access control that Snowflake provides as well. It's pretty compatible with that, and also the security checks and balances, or auth, etc. So there's full governance in that sense. And also, from a NAT name perspective, it brings something that nothing of what we've seen so far supports, which is embedding capabilities. What happens if you want to use any of the functionality that I described so far from Snowflake, but you want to use it in your own app, or you want to embed it in your own website. Well, Snowflake is not quite ready for that just yet. Sigma provides that capability, provides you access, a gateway, a window to all this functionality in a way that you can embed in your own application. Sigma, and that's my sentence, app offering. Sigma allows you to create apps. Now, I've been in the tech industry for fifteen years, and when I first got into that, app development meant progressive web apps and React after that, and then Objective C if you went into the iOS world and whatnot. App offering these days means just drag and drop. Sigma is allowing you to create apps to business users, click click, drag and drop, view buttons, and you can actually have an app up and running that talks to Snowflake, that has all that power, that has all the firepower that we described, and it allows that interaction. And that interaction means reading, writing into Snowflake, or also talking to other systems. Sigma comes with an AI layer, so it comes with assistance to bootstrap and accelerate the development of ads, to ideate and build new apps, etcetera. So it's got AI at the core. Its DNA is AI based. It also provides agentic capabilities. So we spoke about Snowflake's agentic capabilities, so depending on the use case and depending on, you know, your business preference, you might also decide to leverage Sigma's capabilities, although Snowflake provides robust agenting capability as well. And lastly, just the integration extensibility. So just like we saw earlier with Cowork, Sigma can also become that gateway that allows you to talk to multiple systems. So you can engage with Snowflake, but if you have a Databricks instance or if you have a Synapse instance or if you have other platforms or systems, you can actually integrate and create that extensibility, meaning you can have one front front shop for all your business users to integrate with all systems, truly leveraging business intelligence through Pernod. This is what I wanted to cover today. I'm gonna pass it back to Rob for some final comments, and then we'll open for questions. Awesome. Thank you, Roger. So hopefully that gives you an idea. And quite frankly, we couldn't cover everything. There's a lot of stuff inside of the Snowflake platform that is out live now that is ready to use within your platform with your data using the same Snowflake credits. However you get started, we'd like to help. And one of the easiest ways we can do that is, of course, starting at the planning stage. Roger, if you would mind. Thank you. We have a product. We have a lot of different strategy products, but this is probably the best place to start. We we have what's called an SVR, strategy vision roadmap. And the goal here is to understand what your needs are, map those two specific outcomes, and then figure out how to get there. Some of that will be, let's talk about the technology and platforms, the usability and features and functions inside of those. Others will be let's solve use cases. The other one will let's make business decisions, which would be policy or whatever. We'd love to talk to you about how we can assist you in terms of maximizing your ability to use your data, to drive outcomes with analytics and other things, and hopefully ultimately build towards reusable, scalable, and predictably successful AI solutions. Next slide, please. Obviously, there's a bunch more. There's more webinars that we talked about at the beginning of this. You can check us out on our Interworks website. You can even I think if we go to the very next one, there is a QR code. Yes. You can scan there. We'll leave it here. I'd love to see if you guys have any questions. We've got about ten minutes. I'm happy to give you that time back. If you don't, otherwise, we'll give you guys a little bit of time just to type up any questions you might have for Roger. It's a pretty full conversation, so you might have some follow-up. Does not look like it. I'll stall just a second longer, and I think we're good. Awesome. Thank you, Roger, for presenting today. Thank you to all of you for joining us. And please come, I think next week, in fact, when we do our AI COE. Thanks again. Have a great day.