Alrighty. Let's get started. So I wanna say thank you to everybody for spending a little bit of your time with us today. We're gonna be talking about five essential steps to successfully introduce AI into your organization today. Well, the keyword there is successfully. There are a couple of different ways that, things can go pear shaped, so we're gonna explore those as well as give you some ideas of how you can get started. For those that have not been on these webinars before, my name is Robert Curtis. I'm the managing director for Asia Pacific for a company called InterWorks. InterWorks is a global consultancy focusing on data and all sort of the aligned things that come with data. That might be analytics, might be data science, it might be governance. And we look to try to support you on an end to end journey, which includes strategy, where we wanna go, solutions, building the things that are gonna be productive for you, and then helping support them. And that might be the platforms. It might be the things that we build. It might be your community and your users. We do all of it. We've been in business for almost thirty years. I think I need to update this slide more than twenty seven years, probably more than twenty nine. In that near thirty years, we've had the opportunity to work some work with some very large companies, including seventy five of the Fortune one hundred. We have been committed to being open and transparent and sharing our ideas, which has generated almost four million page views per year on our data blog that's on interworks dot com. I've written quite a few articles and white papers on there, so certainly recommend you go and have a look. We have thousands of clients across every industry vertical that you can imagine. So we have a tremendous amount of experience working with all sorts of use cases with the most complex datasets on the planet. A couple years ago, Forbes picked twenty five small giants, and we were one of them. So that was a great honor for us. It's one of the things that we're particularly proud of. Small consultancy. I think we're three hundred consultants, globally, very, very focused on being the very best. And some of the things that we do, these are some of the clients. Again, we have a lot, and these are some of our great partners. You can see data, ETL, analytics, governance. We work with a lot of different great technologies. So that should get us through sort of the intro bit. You know who I am. You know who Intworx is. I could see a bunch of of your names, a lot of familiar faces on there, some new ones as well. So nice to see you again. And for those folks that I'm meeting for the first time, welcome. Let's get started. Again, we are talking about AI and how to introduce it into your organization. We're gonna focus on five things, but we've got a lot of things that we're gonna cover. Before I jump in too far, I do wanna say we do have a chat. So if you wanna ask questions or comment, there's a chat there. If you wanna ask a q and a, there's a q and a button there so you can put those questions. I may not have time to get to them today just to be completely transparent because I've loaded this deck with so many cool things that I'm excited about talking about. So if I don't get around to your question, rest assured, I will circle back around, and we will, we'll answer them via email if we don't have time today. But those are the options that you have available to you. So please, if there's something interesting or you want clarity, drop me a little note in one of those two. Alright. Let's get going. So why all the AI hype? This shouldn't be a revelation, but I thought we'd put some numbers on this. So AI has been around for a very long time. Generative AI is relatively new. And and as we start to add more things that we can solve with AI, they get more and more exciting. And so some of these big numbers and you can see these numbers slightly vary, but there's a lot of consensus to just about the magnitude, if not the specific number. But by twenty thirty, there's estimations that somewhere between sixteen to twenty twenty trillion dollars is how much the global economy will grow because of AI. And then in the same study that I saw, you can see the source there, Oliver Wyman Forum, three hundred billion work hours will be saved annually through automation. That is an incredible number. And so for those organizations that are able to harness the power of AI, a significant competitive advantage. If you thought those were the only two stats that I was going to give you, you were wrong. I've got a whole bunch more. Specifically looking at work automation, fifty percent of work could be automated or something sixty nine percent of managerial tasks could be fully automated by the end of last year. AI spending was projected to be five hundred billion USD last year alone. So a lot of opportunity, a lot of investment, a lot of excitement. But what I'm trying to do a little bit here is demystify what AI is. Yes. There's a lot of opportunity. You have also probably seen a lot of catastrophic failures. And if you're like most organizations, you're sort of stuck between when to get started, what should we get started on, and what does success look like. So that's why I'm gonna ask you these three questions here. Are there really just three paths of AI? And I would say probably yes. The first path is, yay. We did it. We got an AI solution in place. It's functioning as expected, and we've got value from it. Everything is happy. On the other end of the spectrum, I get a lot of, well, we're doing stuff, it's just not happening for us, or maybe we just don't we're not getting lift off, or we've got some stuff cooking, but there's nothing that we can point to to say, wow, AI was a big game changer for us. So you're just getting a whole lot of meh. And on the other side of the spectrum, we did it. It was not at all what we were expecting. Some really bad stuff happened. Maybe we're in some headlines, and my entire team is, creatively looking for new places to work. That is a big spectrum, but what we wanna do is change the probabilities so that you guys are the happy robot. And there's some things that you can think of, as you are preparing for, AI, AI readiness sorts of conversations. There's things that you can think about that will help give you the best chance of success depending on your particular risk profile, as well as understanding costs and how to get started. So we're gonna cover those things all in this presentation. But let's dig a little bit deeper. So when I say the three paths and I've got the little mouse and I've got the big wrecking robot, let's examine what I actually mean because there's some specific things that will lead you to one side or the other of that spectrum. So little mouse of meh. We did it. Nothing really good happened. Nothing bad happened either. We just nothing happened. These are the reasons why you all most often see, a clash of symbols and nothing happening. Sometimes it's because we didn't choose an impactful use case, or we chose one that just didn't have an an obvious business value. Or maybe the executives just weren't sold, and so it was really hard to couple this together without budget or mandate. Or maybe your data just wasn't sufficient. We didn't have good data, or maybe we didn't have enough data. But all it means is we spent a lot of time and energy and maybe a little bit of money, and we just kinda spun our wheels. Now it is a distraction on that particular use case that you might have been solving, but it doesn't mean it can't be without learnings. So I wanna caveat that with maybe an asterisk. The mouse of math means we got our feet wet. Maybe we didn't get the results we wanted, but we probably are smarter for it. Conversely, the wrecking robot. This is where things have a significantly larger scale of failure. Oftentimes, that is because we have bad data. We have a lot of data enough to populate our models, but it may not be the quality that we were expecting. Or maybe we trained our models and we got, as a result, results that were not expected or predictable. Or maybe things were going well, but because we didn't have good governance or any governance at all, things started to go sour, and we weren't able to maintain and optimize the model. Or maybe we didn't have enough consideration on security so that we could protect our data, protect the, the people that were being, leveraged, whether it's your customer data or whatever. But all of these things can lead to a a pretty catastrophic result if you don't plop properly plan in advance. So the verdict for those types of things, and hopefully no one on this call ever has this nightmare scenario, is it's a disaster. So what we wanna do is we wanna certainly avoid the wrecking robot, and we wanna get more value than the mouse of meth, and I think we can do that. So let's talk about how we can move forward towards success, which brings us right to today's agenda. Five essential steps. One, let's understand what your AI risk profile is. It will be different based off of your organization and the particulars of your vertical. Understanding how ready you are to dip your toe into AI, The will does not necessarily mean that you are ready. The different types of AI solutions will be pretty brief here because, again, this is expansive. But think of these as the tools in your tool belt on what you could potentially get value from, understanding costs, and we're not talking just monetary costs, and then giving you a place to start. So there's a lot here. So let's start with your AI risk profile as a company. But before we talk about your company's risk profile, there's an interesting dynamic between users and organizations. So McKinsey released this study called GenAI's Next inflection point, and it found that ninety one percent of people that responded to their survey were using generative AI for their own work, which meant, for instance, I need to write an email. I'm using OpenAI or chat GPT or whatever to help me write it. Or maybe I'm building a dashboard, in Power BI, and I'm having Copilot assist me in creating that. But ninety one percent of people are using AI, and ninety five percent of those people have an overwhelmingly positive experience with it. So the appetite for AI is very high, and the success that people are having with AI is very high, except when you then start to look at this as groups of people, I e, organizations. Only thirteen percent of companies replied that they have multiple AI solutions in production. Another sixty percent are still stuck in sort of that experimentation quagmire. They know they wanna do something. Maybe the executives are excited, but they don't know what they wanna do. Maybe they don't know how to do it, or maybe they're worried that there's risk, and the risk is sort of this paralysis by analysis. So when we look at this from a user perspective, that's the actual breakdown. Twenty one percent are heavy users, seventy percent are light users, nine percent don't. But this is what the respondents to that survey look like. And the things that these people are using for is to improve their communication, understand data to make better decisions, the collaboration between individuals and teams, and help spark creativity. So severally, probably because you were on this call, are probably benefiting from one or more of those as either light or heavy AI users. When we flip this to an organizational perspective, we come to a very different perspective. So this is a very popular, bell curve called the technology adoption life cycle. And you can see two point five percent and this is this is a psychological profile of groups people as much as anything. But two point five percent are early adopters, meaning they are gung ho. Let's do it. Let's be the innovators, the the people that are actually taking technology that may not even be completely ready yet. So, hey. Let's just go for it. Next are the innovators. So the technology is there. Let's find new things to solve with it. Let's go for it. Then you start to go across this majority here. The early majority is okay. A path has been blazed. We're gonna follow in the we're gonna follow in the path. The late majority of those folks, okay. We've got people that are doing it. We see a lot of success. It's been productionized and it's safe. We'll follow. And then your laggards are people like, maybe it's not for us or maybe it's not for us this year or the next couple years. That's sort of your basic breakdown. Right? So when we start to think about this in terms of AI, we could start to build an AI risk profile. So let's start from the Liggers and go forward. So the laggards, those are the people that have zero appetite for risk, and I'll give you some reasons why. Generally, if you're not interested in AI, there's probably one of these four reasons here. And I would say the number one is that you probably were a laggard when it came to data. Meaning, you didn't go to the cloud. You don't have a data warehouse. You're not solving problems with data. Most organizations that are not excited about AI don't have data capabilities. That also by the way, this is not one or the other. It could be multiple. And I would say those organizations are also probably entrenched traditional businesses. Like, for instance, I've had conversations with, companies that run, shipping. They load trucks up. They drive the the trucks around. Everything is paper. And for their business, it works. Will it work forever? No. It won't. Does it work for right now? Yeah. It does. And it's worked for the last fifty years that way. So they don't they don't have disruptors yet. So there's no reason for them to go and pioneer because it's working right now. Or it could be that the business is rigidly regulated. That means it could be gaming. It could be foreign owned. It could be health care government, which means the risk, that comes from regulation or compliance means it's just not worth it for us to invest because the negative consequences so outshine any return. Or maybe it's just a lack of vision of what data and AI can actually do for the business. It could be any one of those or it could be all of those. We then move forward to those folks that have a low risk AI profile, meaning they're ready to get going, but they wanna baby step and they're just not excited to jump in and potentially inherit a wrecking robot. So this is generally where people find there's an opportunity for differentiation. We have an established business. We're talking about late majority as well as some of the early majority, but we'd like to do a value addition. So maybe we've got some data product that we are selling to wholesaler, suppliers, or direct to the market. A differentiation could be, well, we'll give you an AI added feature that then extends the service offering that we've got. For these folks, I would say, as you're thinking about your use cases, do everything you can to sort of focus more on internal solutions first, build up your pedigree so that you can build a higher risk profile because you've got the capability to do it, which means avoiding things that have significant financial exposure or brand exposure. I'd also say if you're gonna be focusing on external use cases, have very narrow, tightly defined carefully defined parameters for what success looks like. And start with simple or small data volumes. Doesn't have to be both. Simple could be we've got large data volumes, but it's very simple data. So it doesn't require a lot of data quality or modeling or all that extra stuff. We then move forward to moderate, and that's some of your early majority and some of your innovators. And so this is an opportunity for diversification. So we wanna add revenue. We wanna add value versus differentiating the stuff that we do. I'd still say be careful on external use cases, but you can step into financial and brand exposure sort of use cases. We'll talk about what specific solutions I'm thinking would be appropriate as we get further into this conversation. You really need to have great data to get into this level. Simple data is probably only gonna solve simple use cases. So the more complex your data, the better your data has to be governed, quality, performant, modeled, all that stuff. And you have to have a good pedigree of organizational data governance, meaning you have to have a culture of good data governance and security and privacy. And just keeping in mind that if you have external use cases, you have to have somebody's hands on the wheel at all time. You cannot just run it like a data pipeline, for instance, and just let it run-in the background. Somebody has to be continuously looking at it, which gets us to the high risk profile. High risk does not mean high chance of failure. It just means if something goes wrong, there's more risk for you. And this is the opportunity for disruption, meaning we want to disrupt our market and take a big step forward. And that's why the early adopters and innovators in this in technology are generally your digital natives or your first to market because they don't have a lot to lose. They don't have massive reams of business and clients and years and years and years of, tradition of the way they've done business. They're ready to change now, which is great for the market. You have to have a substantial understanding of AI and data and governance. You have to have controls in place. I think it helps to have an actual advisory board of AI. Think of this as a center of excellence for AI that has the traditional COE people a part of it. That's data, that's analytics, that's platforms. You probably need dedicated AI team members, not somebody that's in marketing that's doing AI. Your data has to be great, and your data strategy must be fully thought out and mature. Those are the things that you have to do to start doing high risk, sorts of solutions. Again, if there's any questions that you have about this, use the chat, reach out to me. We've got lots of our sales folks on the call. We're happy to help understand what your risk profile might might be. So AI readiness. This is piggybacking off of several conversations I've already done. So there's I think in twenty twenty three and twenty twenty four, I think I presented this, like, four times at different conferences all over Australia, and I've done a couple webinars on this. So I'm gonna go through this bit a bit fast. And if you do wanna understand more about this, let me know, and I'll send you the actual webinar where I spend an hour talking about this stuff. But there's four pillars. You gotta understand strategy, which is where are we going, why we wanna go there, data governance, again, controls, minimizing risk. You've gotta have unified data. Data is the lifeblood for these AI solutions. And from a people perspective, you it's really kinda two, axes here. It's one, getting a data culture built so that we can use AI responsibly and intelligently, whereas at the same time, protecting people, whether it's the people whom we're collecting data from or our people ethically in terms of how we use AI. All of these things have to be considered, and you have to be prepared to do them all excellently. I'll just go through these really quickly. So from a strategy perspective, you gotta have a center of excellence. I think it helps to have a queue of use cases based off of value and effort. I think looking at your vertical and saying who's doing AI successfully will give you a good sense of what the industry's risk profile is, which lets you assess what yours might be. From a data governance perspective, again, the controlling your training data is essential. I could give you a half a dozen examples of AI failures that's based off of data getting bad or data coming from an outside source and moving the whole model in a way that is completely unsatisfactory. I do think you have to have candid conversations about risk. That doesn't mean you have to, stop what you're doing, but you do need to understand this is what success looks like, and this is what failure looks like. And that way, you can properly plan your privacy and security policies. You can take into account regulatory compliance. And, again, that's a moving target in terms of what is expected on the state level, on the federal level, etcetera. Unified data. If you have bad garbage, you're gonna get bad models. You have to have great data to train your AI models regardless of what you're doing. Performance of the models really are important, which is why most AI solutions involve the cloud unless you've got an unlimited amount of money to have your own server form. This is this is not specific so much to this, but it is something that I wanna reiterate. There's a lot of technologies that are trying to help you automate the building of a semantic layer or, in other words, building your metadata or building your business logic. And while that seems easy and convenient in the long run, I think it is way more valuable for you to own your own your own data, own your own business logic, control your destiny. There are a lot of places that wanna lock you in, and that's exactly the idea is vendor lock in, which is one of the reasons why we like, like Snowflake and Databricks. It's basically it's all on blob storage, and then you can do whatever you want basically with SQL. So getting it out if you wanted to, not painful. Well, not as painful as some of these other things. Lastly, people, again, building those skills, building that data culture, and making sure that you're doing this in an ethical way. Let's talk about AI solutions. Again, huge topic. We're not gonna spend a ton of time here, but I'm gonna give you a a very quick overview. So we're gonna talk about five different types. Again, partial list. But, RPA or robotic process automation, this has been around forever. And what's getting exciting is when we take RPA and really start combining it with intelligent decision making that you can get from more advanced AI. So you can see some examples there, automated emails, data pop you know, data forms, that kind of stuff. But that's been around forever. It's getting more robust. It's getting more powerful. Natural language processing. This has been the flavor of the day for the last, say, twenty four months with ChatGPT. But the ability for human language to be categorized and processed and analyzed, I guess, like it would be with just normally modeled structured data. So we're talking about LLMs. We're talking about vector databases. All of this great stuff that now allows us to, you know, write all of our emails using ChatGPT or or your half of my employees, they write their annual reviews with it. I know they do, but, you know, I can't really fault them. We're an we're a technology company. Machine learning. This is where you allow computers to learn without explicit programming, meaning they can actually teach themselves, which is great. And when we add, ML into things like robotic process automation, that's when things get really exciting because then the workloads that we can take off of human beings and put to AI becomes much larger. Intelligent automation, again, this is sort of the combination now of AI, this, normal automation or robotic, the RPA stuff, adding it with machine learning, adding it with other things so that AI can not only improve workflows, but it can own them. And if the workflow changes or there's variables in that workflow, AI can account for it or or make adaptations from changes that are that are un unforeseen. Very exciting stuff. And then finally, generative AI. I think everyone has probably gotten exposed to that because it is now a dinner table conversation. Look at these funny images. Here's the pope, dressed as a rapper. Here's Putin and and Donald Trump holding hands, all sorts of funny stuff. But the AI can generate full articles, full white papers, images, videos, music, tons and tons and tons of stuff. So why that makes it popular for the for the people that aren't in this industry to understand what it is, from people that are in this industry, we could see the tremendous business opportunities that it creates. Let's understand cost. There's a couple of things here to, to unravel. I did a webinar, I think, it was it twenty twenty four, twenty twenty three, where we basically broke down how to how to calculate your ROI on data. So it was an hour long conversation. I'm summarizing here with the sort of, like, my final slide here. So your net return is the easiest way to think about this as things that you're able to put in production and how often those things get used. So if I've got three models and twenty five dashboards, and all of a sudden I've got a thousand users that are engaged, that is a massive net return. Great job. You're killing it. The other side of the equation is your cost of investment. And, yes, money is part of it. What does it cost to run this tool? What does it cost for us to have a space on Amazon or on Azure? But it's also the cost of time. Our people are spending time on this. Our people are suffering because of the opportunity cost of not doing this, so sort of that lag or delay. We might be paying consultants, folks like InterWorks to help us. So there's a lot of different ways to say, what is the cost of this? So I'm trying to simplify this cost in terms of money that you spend, the time it takes you to do it, and the delay it takes from using from getting the advantage of it. That's kind of a holistic cost. Now most of your executives are probably only interested on the money side, but when you actually say, listen. If we don't do this, then our folks over in invoicing are spending an extra twenty hours a week. That's, I don't know, four hundred hours extra a year. On average, we're paying these people fifty dollars an hour. You can start to see that just fixing that pays for the solution on its own in year one. That's a great way to think about cost, money, time, and the delay it takes from us actually getting the benefit of it. So let me give you a sample cost exercise because we go through this a lot. Sometimes people will come to us and say, hey. We want you to present a Databricks solution or a Snowflake solution. Other times, they say, we want you to help us solve this, and then we want you to give us options in terms of tools we can select. So here's a sample cost exercise bringing forward what we just talked about. So scenario one, we've got a highly managed data platform, which means the data platform is ready to go. There's not a lot of management. There's not a lot of refactoring. There are settings, but not exhaustive settings to where we can't get started if we don't play around with all the settings. So we've got money, time, and the delay. So it might be it's a little bit more expensive because it's a bit more SaaS. There's obviously gonna be some time to get it set up and to start building things and start structuring your data and building your data pipelines and blah blah blah. And then the delay is probably like, okay. It's gonna take us a month to get this off the ground, so delay is x. So when you factor all those things together, however you wanna try to equate them apples to apples, you might get a total cost of something that looks like that. On the other scenario, you might have a highly customizable data platform that has all the levers. You You can choose where you want the processing or do all the crazy things that you wanna do with it. And it might save you some licensing costs, but it might cost you a lot more time or consultants' time to actually get factored and and set up and properly engineered and everything going the way it should be and predictably, which means you're probably delaying how long it is before you can actually start taking advantage of it. So even though you are saving money on licensing, your overall cost might be higher. And that's just thinking if about what's the time frame that I'm measuring this cost. Am I measuring this cost as a proof of concept, and that's what it looks like over three months? Is it a year one cost? Is it a total cost of ownership? Do I have to hire three or four more people to get this highly customizable data platform running? Or, when I chug through six trillion records of data into the highly managed one, that's gonna be an incredible licensing cost, and it is worth it for me to go and hire a data engineer to to manage a a more customizable platform. When you're thinking about how software charges, this is an overview of of the most common ones. Most of the time, it is consumption based now. You'll see metered and other things like that, but the two big platforms from a data perspective where all your AI models can sit are consumption based, which means you have to be efficient. You have to be mindful that you're controlling your own destiny and spending those time spending the time that you're using your compute effectively. The way that you negotiate for the best pricing, these are some of the things that we we deal with all the time. So how much are we using? What level? Is it business critical? Is it, you know, a private VPN? Am I contracting it for six months, for a year, for five years? Am I growing, or am I just renewing flatly? Most people get surprised that salespeople, vendor salespeople are really incentivized on growth. They need to grow. So a lot of times people don't understand that they don't have as much leverage if they're renewing a ten million dollar deal versus adding two hundred fifty k of new spend on it? What kind of support do you want as a part of your your vendor negotiation? And are you using your hyperscaler like an AWS to help subsidize a new use case that you're bringing to their platform? Because they have money. They have funds for that to help bring your workload onto them so they'll help pay for it. There's a lot of different variables here. And what I would say is don't get too stressed about this slide because InterWorks helps people do this all the time. Whether we are the people that are helping you source your licenses or we are being a adviser to you to get the best deal, we deal with this stuff all the time, and we can really help as well as see what sort of programs are available to help you with funding, whether it's from the vendor themself or a hyperscaler or any number of different ways. Lot of cool stuff in there. Okay. Where to start? If there's nothing else that you take away from this conversation today, it is this, this one essential takeaway that I want you to understand. And I'm not gonna take credit for this. I saw it somewhere. I can't remember. Otherwise, I would give credit to it. Although it's pervasive enough that I don't know if a single person can say that was my idea. But it's this. Your data strategy is your AI strategy. They are not separate. They cannot be separated. If If you don't have a data strategy, you don't have an AI strategy. You have to start with your data. And so if you're one of those organizations that has been promising to get around to data for the last ten years and AI is now the impetus, you cannot skip data. You have got to go and do the work, which means probably building a data strategy, a vision for how you're gonna use your data, thinking about governance, thinking about security, getting your data into a centralized data warehouse where you could start to do all of these downstream use cases, analytics, self-service, data science, automation reporting, whatever. The good news is, from a getting started perspective, is that AI is already everywhere. And if you think about all the different ways that we can use AI, whether it's machine learning or RPA, those things are embedded now in the technology that we're using. Some examples. Snowflake has a whole bunch of stuff, including Copilot, which can help you actually write SQL. I don't I don't think that made that onto my list there. Tableau, Power BI, again, they are building AI automation to allow users with less technical knowledge to do more complex things. So the ability to just type in natural language and say, show me this analysis. That stuff is there today. Same thing for ThoughtSpot. ThoughtSpot is very leaning on the natural language, querying part of this. Informatica, Matillion, all of the ETL tools, they all have a form of Copilot to get you started faster. Informatica is particularly exciting because they rebuilt the entire platform, using Clair. You see Clair, the AI inside of Clair, where they can help automate with AI your data quality, your ingestion, your transformation, MDM, and all of that sort of builds off of each other. So very exciting times if you're in our space because AI is everywhere. But while that helps automate workflows inside a technology, we still can say there's a lot of easily accessible AI stuff to solve specific use cases. I love this view from Snowflake. Snowflake is not the only data provider that can help you do this, but you can get a good sense here. Use AI in seconds for this Document AI, which is a pretrained model to see images and convert them into structured data. So I get a scan. I get handwritten notes. I get an X-ray, whatever, and then the document AI can make that into something that's queryable. You can build apps in minutes, and then in just a few hours, you can make this fully customizable and off and going. I will tell you, we have experience doing this. We do this type of stuff a lot for our customers, and it is absolutely the god's honest truth. You can do this stuff very quickly and extremely powerful. This particular webinar that we're doing today is gonna be part of a series. We're gonna bring Snowflake in to talk about their solutions in a lot more detail. We're gonna bring some customers in so that they can share their experiences, particularly with Document AI. One particular customer has an amazing case study for that. So this stuff is right at your fingertips. Structured data in particular is a proven commodity. You can start doing that stuff. That would go in the low risk AI profile. Let me give you an example. We'll we'll go into, some broad examples from AI opportunities by risk profile, low, medium, and high, and then we'll go into a specific example of one of our customers who all keep anonymized at their request. So the low AI risk profile, again, unstructured data. So it's taking, images or or recorded language and changing that into data. Very predictable, very successful. Text and content generation, your your chat GPT steps up, taking parts of your workflow and augmenting it or automating it. For instance, like, this is what my data pipeline looks like. If new data starts to come in or new things start to come in, my ETL tool can use machine learning to adapt and get that data properly structured and and ready to go. Those things are low risk. And, again, remember, a lot of these are internally focused, which is what we said low risk AI profile should be focusing on. We then go to moderate. You could start to open up chatbots and conversational AI. Again, I would say keep them very narrow. So people come in and ask, where's my where's my order? Where's my shipment? Very narrow, very easy to sort of maintain. Not a lot of risk if they start asking questions that are not within the purview of what your models own to do. You can also start using AI to help you with diagnostic and predictive analysis. It can start to own the workflow versus segments of it. You can also use it specifically to do quality control. That might be quality of your invoicing, quality of, the other transactions that you might be quality of your data. And, that also then leads obviously into stuff like fraud detection. That all goes into the moderate. I would say because there's a financial, because there's an external, and maybe a brand, that's why I'd put some of these in the moderate versus the low. But a lot of these are still very predictably successful and tried and true. When we start to get into higher risk, which means you just need to have a little bit more investment in your infrastructure, that's advanced chat chat bots and conversationalize so they can answer more questions. You can actually empower AI to make decisions for you, versus owning a part of the workflow or the workflow. You can actually then say, well, based off of your workflow, you can then derive these decisions and make them for us. Again, remember, continuous oversight is required. An example of that is autonomous vehicles, as well as delivery. You know, that could be drones delivering it to people's front doors. These are the big cool stuff that you see, Amazon doing. Complex system optimization, so it might be maintaining a risk portfolio across investments. All kinds of cool stuff are out there that larger organizations that have great data and a real investment in this are are leading the way for some of these amazing, examples. So now that we've got an idea of potential opportunities based off of your risk profile, let's talk about a specific example. But we're gonna structure this in a way so that it's it's easily understandable and hopefully replicable for you. So there's six steps to building a workflow to help you automate your AI solution. And the first one is, let's assess your current workflows. What is inefficient? What's painful? You go department by department and say, where do you guys spend a lot of your time? Ugh, my goodness. Preparing our monthly reporting or preparing the invoices or or whatever it might be. You'll start to see what is creating the problem. Map the process. Literally map out the workflow. Okay. So it happens here, and what happens here? Okay. So there's a decision point. Yes, no, maybe. Okay. What happens on each of those? That will help you understand where the inefficiencies or where human beings get involved, and that's the best opportunity to automate. Depending on what it is you're trying to do, that's when you can say, okay. These are the AI tools we have available to us. Which is the right tool to help fix this particular problem or that point in the workflow that's inefficient. POC it, automate it, see how it's going. Once you've done it, test it and refine it. Get up that, success ratio or the success rate of your model. And then once you're happy and it's out in production, monitor and optimize, and do that forever. Do not get caught blindsided if your model starts to hallucinate or starts to get bad data or starts to work in ways they're unexpected. You've gotta stay hands on with this stuff, not only to derisk, but also you're looking for improvements. You're looking for more value. You're looking for ways to take that idea and apply it to other problems you have in your organization. So let's take a real example. I went ahead and just mapped all this out, but this is this is an example of toll road billing. So a car is driving on the road, a toll road. They go through a gantry. The gantry then scans them, and a whole series of events happens in the background. So the first thing is that the gantry says that car just passed. Do they have a little toll tag that sends a little signal that our gantry can record? If it does, then that means they're a customer, and they can automatically go and say, okay. That dollar fifty is automatically deducted from the customer account. Done. What if they don't, or they don't have an active account? Well, the current process was we're gonna take a picture of the car's license plate. That picture then gets sent over to a big data lake. And then a human being goes through a queue and says, okay. That's what that that's what that license plate looks like or that number plate. I'm gonna type these in, and then we can figure out where this person lives and then send them an invoice. If they have an account, then we go, okay. Great. I'll just apply it to their to their account. Boom. Automatically deducted finished. So you can probably see immediately where the inefficiency is, and it's right there where the human being has to stop and transcribe numbers, license plates. And based off of the number of people that were going along these roads, that was thousands of scans of, license plates, not not really even per week, per day. So that was a lot. And they had some basic, image recognition that would get a percentage, but not a very acceptable amount. So there was still a lot of human beings sitting there doing data entry. And nowadays, unnecessary. We don't need to have those folks spending that time doing it because we have AI automation. So for this particular example, something like Snowflake Document AI, which has computer vision, which is another way of saying it can look at images and translate that into data, can actually do a lot of automation. And one of our customers that we're gonna bring in in the future, webinar as a part of this three part series is gonna explore, not toll tags, a different industry, but getting a an amazing success almost right out of the box with Document AI. I've been really, really impressed. I'm generally kind of a skeptic, but Document AI has blown me away in just how useful it is, how quickly, you get great success. Very, very exciting. So when we talk with customers, this is exactly what we do. First thing we do is we sort of explore what's possible, very similar to what we've done leading up to here. And then we say, what's painful for you? Is it in sales? Is it in legal? Is it in risk? Is it in supply chain? And let's map everything out, and then we can say, okay. We can automate here. We can automate here. We can automate here. And then we figure out how much time you're spending there, and you can literally say, okay. We're gonna save you a thousand hours a year. Is it worth it to put the solution in place? What if it's a thousand hours every year? Is it worth it to put this? And you can start to make those value cost decisions. We can help. So if we were to think about how you go through your data journey, because remember, your data strategy is your AI strategy, it might look like something like this. There's a lot here, so I'm gonna walk through it. But ideally, we would start with what we call a DART or a data and analytics review and tactics. Where are you right now? And what we do, you'll you'll see an example of of, a little bit more on a DART here in a second, is where are you on five, essential frameworks? And you have to be mature. We we we we rate them on a scale of zero to three, and you have to be a two or higher to be eligible on that framework for AI readiness. So five frameworks, you need to get five stars. That gives us a a a stake in the ground of how how well you're doing and how ready you are. And the thing that's great about a dart is it's actually quick, pretty pretty quick. We can do it in about a week with a couple conversations. And then every year or every six months, depending on how quickly you wanna go, we can come back and revisit it and then upgrade and compare how well you're progressing, which then leads us to an SVR, which is a big strategy conversation. Why do you wanna go this way? How are we going to do it? What's what what what's your governance? What your people process? What's the vision of your organization? Of course, that then ties right into governance. We can then figure out what tools are the best tools for what you're trying to accomplish. It might be a singular platform that can do all the things that you need to do. It might be a, approach by committee. We've got real time data. We've got this. We've got this. The best solutions are gonna require this, this, and this. We'll put it together, and we'll make it work. We have what are we call rapid starts, which are fixed cost to get these tools off the ground and going. We do these very quickly. We have seen, our competitors in this space take weeks. We generally do it in days even for very complex systems. Once you have your data warehouse in place, you can then start adding use cases, adding more data, and expanding value. And and a lot of times you see that in terms of analytics or self-service and then building that community of practice and building data culture. So that's what a data journey might look like. And you might need help on step one or step two or step three or whatever. We're gonna overlay now what your AI journey might look like. And, again, your data journey is your AI journey. So we're gonna take what we've got here and just add to it. So everything that's got a green line now is AI specific stuff. So for instance, we want your leadership to understand what the opportunity and the risk is for AI. So that's why we like to lead off what we call an AI exploratory workshop for executives, which is like a two hour conversation. We talk about what is AI, what are the opportunities, not too dissimilar to this conversation, but we break it down in more business context versus a technical one. We then partner that with an AI use case discovery workshop. So remember how we mapped out what the, the toll road billing look like? We'll do that exactly for what we think are use cases that are the best opportunities or candidates for automation for AI. And then as we go downstream and we get everything into the data warehouse, that gives us opportunities to solve problems with data science, machine learning, building AI solutions and automations, and connecting all of that back in to enrich your organization and raise the culture, to build more skills and capabilities so that it is self sustaining now in terms of the ideas and things that your organization is building. Very, very exciting. We do this type of stuff all the time. And what's really exciting is we've got a, a neat little offer. As part of our investment in your journey, we're offering, the dart assessment and the AI exploratory workshop for executives for free. That is about a week's worth of our time because we believe, in sharing the risk with you. It's part of our commitment to your AI journey, your data journey, that we get you guys started off on the right foot. The last thing we want you to do is to attempt to do AI or data, not have all the answers that you need and not have a good result. Or potentially, not just the mouse of meh, you end up with the wrecking robot, and then there's a a catastrophe that we have to help, solve for you. So a little bit more information on these just so you have a little bit more information. A this is again, the dart is built off of five maturity frameworks, culture, analytics, data, governance, and platforms. And based off of how you score, we then can give you an AI readiness assessment. Because, again, you have to be good at all of these things to start AI to start AI. In terms of the AI exploratory workshop for executives, it's two hours. It's great if we can meet in person. We'll bring lunch, and that way, we can kinda get around the round table. I'll go through some key concepts and some opportunities, and then we have an open discussion in terms of what your executives get excited about or maybe they've got concerns. And so I can help guide them in terms of, yes. There are risks. Here's how you mitigate risk. And I'd talk to them exactly the same way in terms of if you've got a low risk profile, these are the things that we can start thinking about to solve problems. If you've got an appetite for a little bit more risk and maybe because you've got the investment in your data and your people and your governance, we can start to do, things that are a little bit more of a reach. These are the types of things that you might think about. So we're offering both of those things absolutely free. Just reach out. You'll get an email after this, after this conversation. So if there's anything that you've you wanna talk more about this or you wanna understand, hey. Have you done this for other customers? We'd like to talk to them. Whatever it is you need, we're happy to help. We're happy to have a conversation, however helps you guys make your decision. Last thing for me today, just as a reminder, any questions, chuck them into the chat and q and a. I'll have about five minutes. If you do wanna contact us, just scan that little QR code there, and one of our very helpful salespeople will, will reach out and see how we can help. That might be emails. It might be Zoom. Or if you happen to be in Melbourne, Sydney, or Perth, it might be us dropping by with a coffee or a lunch so we can have a good conversation about how we can help. I mentioned this at the start, but all of these webinars are recorded. So, again, there's a lot of stuff in here. If you wanna review certain parts, you're like, oh, I wanna share this with one of my colleagues. It takes us about three or so business days to get this turned around and nicely polished into a little video and and get it posted on the website, we'll shoot out the links so that you have that too. So far, there are no comments in the chat or the q and a, so I think we are probably good. I might just stall for a second or two longer to see if anybody puts anything in there at the last minute. Otherwise, I'll say thank you for everyone that attended. Certainly looking forward to talking to you again. And remember, we've got two more webinars in this series where we're gonna talk about specific features and functions with Snowflake, as well as bringing on customers to share their stories on their AI journey. Awesome. Thank you very much, everybody. Hope you liked it.