From ideas to impact: How to build AI into your business with Domo 

Cody Irwin: Welcome to our webinar today. We're going to kick off here in about 30 seconds. We're gonna give people time to funnel in, so…

Cody Irwin: Please be patient, we'll start quickly.

Cody Irwin: Okay.

Cody Irwin: Welcome to our webinar today. Thank you so much for being here. Tom and I are incredibly excited to be with you and to talk about moving from ideas to impact, really leveraging AI

Cody Irwin: to drive impact in your business. This is a topic we're very passionate about. I'm sure you are too, or you wouldn't be here today. So, thank you so much for joining. Before we get into it, we'd love to do introductions on who you're going to hear over the next 45 minutes. My name is Cody Irwin.

Cody Irwin: I'm the AI Adoption Director at Domo. I'm based in Utah, in the USA, a little bit across the pond from you. Incredibly grateful to be here. I love our team in Europe.

Cody Irwin: And I'm, super excited to walk through these topics with you. I have with me Tom.

Tom Pugh: Thanks, Cody, and hi, everyone. I'm really looking forward to today's session. As Cody mentioned, my name is Tom, Tom Pugh, I'm a solutions engineer here at Domo. So my day-to-day role is really helping customers all the way across the data spectrum in terms of

Tom Pugh: looking at AI use cases to actually then implementing them. So yeah, really excited today to run you through a bit more about AI. So, Kirly, back over to you.

Cody Irwin: Thanks, Tom.

Cody Irwin: Yeah, so, thank you so much for being here. We realize your time is valuable. We want to ensure that we make the most of it. We're anticipating about 45 minutes for this conversation. We are going to have a Q&A session at the end.

Cody Irwin: So please feel free to put questions into the webinar. There should be an option at the bottom to add those questions, and we'll respond to those.

Cody Irwin: Our planned agenda for today is going to cover a variety of topics, with the key goal being

Cody Irwin: Moving from medial…

Cody Irwin: kind of historical task-based concepts into this AI world. We'll start by talking about the vision of AI, kind of starting with the why.

Cody Irwin: We'll go from there into Domo's philosophy, so what does it mean on Domo?

Cody Irwin: We're gonna share a series of examples. We don't want it to be all theoretical. We're gonna give you tangible examples of what that means.

Cody Irwin: And then we're gonna talk about a framework for actually making it happen. So how do you go from idea to actual implementation? And from there, we'll have a Q&A session at the end.

Cody Irwin: With that, I'll kick it off on some of the philosophical. So we're gonna… we're gonna roll this way back. This question's maybe too philosophical, but I think it's kind of worth addressing. Why do organizations exist?

Cody Irwin: When it comes down to, we're finding that AI is forcing us to go back to basics. For AI to work really well, we have to know what we want from it.

Cody Irwin: And I'm finding in many cases that people are viewing AI as magic. They just think it will solve every problem.

Cody Irwin: The reality is, we need to know what we want to solve.

Cody Irwin: Organizations exist to organize people, process, and technology to solve a problem in a way that individuals cannot.

Cody Irwin: That's why it exists. Organizations need to have a core why in how they operate, and AI is requiring that more than anything before. I'm finding, talking to a lot of companies.

Cody Irwin: That AI…

Cody Irwin: is making them do legacy things better. They have to kind of figure out what ideal looks like to a degree, what their goals really are, before they can, like, really walk this path.

Cody Irwin: I'm sure most of you have had some experience with AI. It's not new. And when I say AI, I'm not referring to the broad AI space. AI is a term that was coined in the 1950s. It's not a new term, but we've had a series of hype cycles and winters, as they call it, where people have gotten excited about AI, super excited about, like, what it could be for them.

Cody Irwin: And they've realized, oh, this is hard. It's difficult.

Cody Irwin: And that excitement has dropped off.

Cody Irwin: This latest hype cycle's been interesting because it's consumer-focused. I'm guessing many of you have been to ChatGBT, or Gemini, or Grok, or Anthropic, or somewhere else.

Cody Irwin: co-pilot, and you've asked things like, hey, write me a… write me a rap about a pony, or something, and it's done it, and it was fantastic, and you're like, this is incredible! And immediately, you're trying to connect dots between

Cody Irwin: What you saw there, and your business.

Cody Irwin: And it feels like magic. I think because of that, we're taking those dramatic leaps and thinking, oh, this can just solve our problems. It can do creative and insane things.

Cody Irwin: But we are finding that organizations need to anchor on their core whys.

Cody Irwin: AI's gonna force us all to be better, and to focus on, really, what we're trying to accomplish.

Cody Irwin: There was a report that came out back in July from MIT,

Cody Irwin: I'm sure many of you have seen that report and seen this quote. And I think for many, it was initially kind of alarming. Like, what's happening here? Like, why isn't delivering?

Cody Irwin: MIT found that despite massive investments, organizations were seeing very little return on AI.

Cody Irwin: as we've talked to companies, as we've dug into it, it goes back to that initial slide, that initial kind of conversation around the whys, and I think in a lot of ways, people viewing it as something

Cody Irwin: Overly disruptive.

Cody Irwin: Which, for some of you, that may feel a little weird for me to say overly disruptive.

Cody Irwin: We're finding that sometimes when dramatic innovation happens, companies feel like they need to dramatically innovate.

Cody Irwin: They need to kind of take this big, massive leap towards something new. Something that, in some cases, is unnatural. And they get stuck in these endless cycle… endless cycles of POCs.

Cody Irwin: Where they're just trying big, crazy, hairy things, and trying to, you know, really disrupt their business.

Cody Irwin: What we found works best, and I'm gonna pull from someone way smarter than me here, we're gonna… we're gonna pull back into Atomic Habits by James Clear, is that companies that focus on marginal gains.

Cody Irwin: seem to be the ones that are getting the actual returns. For those of you who haven't read this book or heard this story, James Clear talks about the British cycling team in the early 2000s, and how the team just was not having the success they needed.

Cody Irwin: I think they hadn't had, received, like, an Olympic medal since, like, the early 1900s.

Cody Irwin: they were not competing well in the Tour of France. Someone told me at one point that it got to the point that

Cody Irwin: the big bike brands did not want them riding their bikes. Like, they were not in a great place, and they were feeling this need for massive disruption. They saw things happening that weren't happening to them. And I imagine many of you are feeling that same thing with AI.

Cody Irwin: They had a new coach come in, and the coach really advocated for small improvements. These 1% increases compounded.

Cody Irwin: Things like changing the pills they were sleeping on, or changing the kits they were writing in, for these small improvements. And they found that those small improvements compounded, delivered massive results. That change was much easier. It wasn't…

Cody Irwin: rapid disruption, it was natural evolution. A report came out from OpenAI

Cody Irwin: in November, where they talked about the things that are driving success for companies when it comes to AI adoption and productionization of AI.

Cody Irwin: And they hit the same term. They said… they called it compounding ROI.

Cody Irwin: And they encourage companies to really focus on that. To focus on starting where it makes sense.

Cody Irwin: So what does this mean? Like, how do we apply that? Like, how do we make it a little more tangible?

Cody Irwin: We found this matrix to be very helpful for companies in thinking through where to start and how to invest.

Cody Irwin: I'm a big fan of alliteration here, so we have the three I's, increment, improve, and innovate.

Cody Irwin: Again, I think a lot of companies are focusing on innovation.

Cody Irwin: they flipped this investment ratio and said, we're going after crazy innovation, like, let's build a self-driving car, let's disrupt our business and reinvent. That's really hard to do. It's hard to change regardless of disruption. It's hard to change with rapid disruption from AI as well.

Cody Irwin: We found companies having a lot more success when it comes to starting with incrementation, or primarily focusing there. I will say, this is not intended to be, like, a start here, necessarily. It really is an investment mix.

Cody Irwin: It's focusing a majority of your efforts on incrementation, doing things you're currently doing a little bit better.

Cody Irwin: What does better mean?

Cody Irwin: Warfighting generally is better means, more efficiently.

Cody Irwin: For a lot of companies, I think AI does not equal innovation, necessarily. As the core focus area, AI equals efficiency. That's the hope. Can we get more efficient? Can we be more productive in deploying our resources to either make more money or save more money in some form or fashion?

Cody Irwin: I do recommend companies also invest part of their efforts in improvement, so not just taking what you're… what we've done historically and doing a little bit better, but reinventing some of your existing processes.

Cody Irwin: Finding optimizations that are slightly more dramatic, but have massive increases for you.

Cody Irwin: And I think every company should have someone thinking about innovation.

Cody Irwin: This Gen AI movement is going to drive innovation for companies, it's going to introduce new companies to the market. I'm sure many of you

Cody Irwin: Have fears of disruption, have fears of other companies coming in and doing what you do better, or in a more unique way.

Cody Irwin: You should have someone thinking about innovation. But this investment mix, I think, helps you conceptualize where to really go after this.

Cody Irwin: We found that when it comes to incrementation.

Cody Irwin: It can be kind of hard to find those opportunities. There's a lot of them.

Cody Irwin: I'm sure many of you, as I think through your business, are like, well, there's a lot of things we could do, where do we start? What could that mean? Later on in our presentation, Tom's gonna go deeper into frameworks for identifying and prioritizing.

Cody Irwin: initiatives.

Cody Irwin: But we have found that one question in particular seems to really help. For those of you that don't know Ken Boyer, Ken Boyer is one of our product managers here at Domo. I have personal goals to make him famous, so I've been sharing this quote everywhere.

Cody Irwin: Ken's very good at helping people conceptualize

Cody Irwin: things. He has great stories, and I've heard him ask this question of companies as we've talked about AI disruption and innovation.

Cody Irwin: What would you do today if you had a thousand interns?

Cody Irwin: And it's a really simple thought exercise. I encourage you to kind of pause for a moment and think about that.

Cody Irwin: And what that really does, that uncovers…

Cody Irwin: Those things that we find menial, those things that we find repetitive.

Cody Irwin: It really is places for incrementation. It's places that we can introduce someone to help us in some form or fashion.

Cody Irwin: Ai, in my personal opinion, is far more capable than… than some interns. I love interns, we have some great interns here at Domo.

Cody Irwin: But AI, in some cases, can be a specialist.

Cody Irwin: But I found that starting, kind of really getting momentum, it helps to start there, to kind of find those scenarios.

Cody Irwin: That feel like busy work.

Cody Irwin: and apply AI there.

Cody Irwin: Doing that, incrementing your business, delivers immediate results, immediate efficiencies. It's not these long-term, you know, big bets, which, again, we should do some of those.

Cody Irwin: It gives us things that immediately deliver a positive return to the business.

Cody Irwin: Okay, I'm gonna hop out of the theoretical, and kind of positioning at the top level there on how you get started, and I want to talk a little bit about where Dummel fits in.

Cody Irwin: I'm sure some of you on the call are customers of Domo, some of you are investigating Domo to see, can Domo help in the space.

Cody Irwin: I found that it helps to kind of, like, kind of lay the stage for where Domo fits. Like, who is Domo in this new AI world? Historically, for those of you that have worked with us, or have investigated us, we've been viewed primarily as an analytics company. We've tried hard to be full-stack in nature.

Cody Irwin: when it came to Domo's founding, like, why we exist as a company, our goal really was to unlock data for business.

Cody Irwin: That's why we were created. They happen to be proxied by dashboards.

Cody Irwin: I do want to hit that point as part of this conversation.

Cody Irwin: Entering this new evolution of technology.

Cody Irwin: I encourage each of you to look for areas where technology has become a proxy for a problem.

Cody Irwin: And I think those scenarios where the technology is more valuable than solving the problem… I'm sure all of you have situations where you've seen that, where it becomes almost sacred. We need to question those things, because AI is going to force questioning.

Cody Irwin: And again, as a company, we've been doing a lot of that. We've been trying to think through, are we overly indexed here? It has provided value.

Cody Irwin: But are there other things that are more meaningful? But this has been how Domo's been viewed historically. We've been viewed as a company that really goes from data all the way through activation with the analytics, surfacing insights through cards and dashboards.

Cody Irwin: The ideal is probably something a little different, and we're… we've been kind of coining this term of the efficiency journey. Like, how do you move up the chain to more efficient operations? Again, efficiency being that core driver of AI.

Cody Irwin: This, this slide purposely does not have anything DOMA-related on it, because I think it is a quite generic slide. I think it's going to be applied to a lot of scenarios, a lot of technologies.

Cody Irwin: Where step one there is dashboards. I'm sure some of you who have built dashboards and visuals in the past have built a really cool one and thought, we did it! We solved the problem!

Cody Irwin: Everyone go home! That's kind of tongue-in-cheek. The reality is, dashboards are inputs to bigger business problems, and they always have been.

Cody Irwin: And I think the efficiency journey is pulling us up that chain to kind of think that way. Like, what's the problem we're solving with data? Step two in this journey tends to be assistance, so we're seeing a massive push in this area for

Cody Irwin: Reactive, natural language-type experiences for getting insights, taking action to build, things like that.

Cody Irwin: We've seen a big push in the market towards apps.

Cody Irwin: And apps are viewed more as a combination of dashboards and action, meaning that you can input data, you can trigger workflows, you can do things beyond just the

Cody Irwin: Consuming of information.

Cody Irwin: Beyond that, we're seeing a lot of companies that kind of mature down the cycle push towards automation. Can you take the process and automatically do things? And automations in this context is very deterministic in nature, how we're viewing it, that journey, meaning step one, step two, step three. The ideal with AI, or where the AI market is going, is towards

Cody Irwin: Agents.

Cody Irwin: Where agents can actually help us reason through a process, through a problem, and can adapt.

Cody Irwin: And can help us, find nuance, can help us solve through complexity. Agents are kind of that

Cody Irwin: kind of crown jewel of where things are going in this current iteration. A lot of companies are very excited about those and what those can be.

Cody Irwin: I do want to call out, all those things require a strong foundation.

Cody Irwin: And we're seeing a lot of reports, a lot of findings from companies that are talking about that.

Cody Irwin: That it really requires… like, to get value from this really requires a strong, strong data foundation.

Cody Irwin: The beauty of Domo, for those that are customers, for those that are considering Domo as a possibility, is you've built a lot of that foundation already, if you've worked on the path of, like, doing analytics.

Cody Irwin: the foundation's the same. It will be adapted, like, things need to be added to it. We're seeing a massive opportunity in the market for context, and semantics, and data models, and things like that to give even more information to the AI.

Cody Irwin: But one thing Domo's been really good at historically is building that foundation, and helping you get value from that. And that foundation can be used to pivot and extend.

Cody Irwin: To take this a step further, and to start kind of layering in some of the Domo concepts here.

Cody Irwin: Domo does have functionality here. We… there's some things we built in the past that we did not build with this journey in mind, but we just saw a need.

Cody Irwin: And the AI evolution has really kind of convinced us that

Cody Irwin: We need to lean in further to that journey. We need to push people up that cycle.

Cody Irwin: So, there are tools there. There are things that we're building, things that we have built, that can help you get down that journey effectively.

Cody Irwin: If some of these tools are new to you, if you are a customer, or if you're someone that

Cody Irwin: is investigating Domo. We'd be more than happy to go deeper into them. Tom's gonna kind of touch on a few of these concepts in his demos. I found, talking to companies, that Domo has a broad enough platform that some of these things people aren't aware of. And there's some pretty cool stuff there. Some pretty unique things we can do.

Cody Irwin: I did want to highlight really quick, on the agent side, because this is… this is a newer concept. Everything else in there, I'm sure you've either seen or experienced in some form or fashion.

Cody Irwin: in your careers, agents are a little newer.

Cody Irwin: And we're finding not everyone knows what that is.

Cody Irwin: how it plays into their workflows, what it can do. And where they're different, agents are focused heavily on

Cody Irwin: Reasoning through a problem. Again, they're probabilistic in nature.

Cody Irwin: Meaning that they rely heavily on goals or instructions. Like, what do you want to… what do you want it to accomplish? This comes in the realm of, like, prompt engineering.

Cody Irwin: You'll see a lot of people kind of leaning into that concept.

Cody Irwin: Agents need knowledge, both unstructured and structured data. Structured data being columns and rows, unstructured being things like images, audio, PDFs, documents, all those things.

Cody Irwin: Now, they also need tools, things they can do, both from a research perspective, gathering information, as well as an action.

Cody Irwin: And the amazing thing is, using that LLM as a brain, and one thing to call out there, I think most of us have seen LLMs primarily in the generation and summarization place, where, again, like the write-me-a-rap scenario, they can do that, or summarizes article, they do that really well. They're also really good at orchestration.

Cody Irwin: Where they can go through, kind of reason through the problem based off of instructions, knowledge, and tools to help you solve a problem effectively.

Cody Irwin: So I wanted to call that out, because you're going to see a little bit of that later as well, in positioning. Domo has a framework and a thought around how this works. You'll see it reflected in our product.

Cody Irwin: The last thing I wanted to share before I pass it over to Tom to give some demos are some foundational patterns. So, as Tom's sharing some examples, here's things that we've seen consistently

Cody Irwin: asked for on Domo, right? People seem to kind of rally towards these patterns. And patterns are tended… are tended to be…

Cody Irwin: things that can be reused for different business scenarios. So we've seen a massive push for unstructured data pipelines, meaning I have data historically that's been really hard to use for analytics. Things like invoices, or images.

Cody Irwin: That people want to actually process and either put into structured data for deeper analysis, or use in more of a… a chat-type experience.

Cody Irwin: So we've seen a massive push for that. With that chat-type experience, we've seen a big push from a lot of companies for a talk-to-your-data experience. They want to be able to go in and

Cody Irwin: have their people actually interface with the data directly, without having to talk to an analyst. So kind of an analyst in a box, almost, in some form or fashion. And obviously, this has been made popular by the large foundational models.

Cody Irwin: And how you interface with those. We've seen a push towards automated insights, so proactive information, like, hey, this is happening in the data. This comes into effect through an alerting mechanism, as well as just within the dashboard, additional context.

Cody Irwin: What's happening there?

Cody Irwin: We've seen a big push for sentiment analysis and classification of unstructured data in structured data sets. So this is things like call center transcripts, surveys, reviews. Like, take this information that's text in nature and help me understand what's happening here.

Cody Irwin: And we've seen also a big push for content generation. Can we actually produce meaningful information, whether it's through

Cody Irwin: in product experiences like dashboards, or things like PDFs, PowerPoints, etc, that are more AI-driven in nature.

Cody Irwin: There's a strong foundation across all of these for process automation. So, process automation, I think, is getting its day in the sunlight. A lot of companies are starting to actually investigate it and think more about how they can automate things and create that efficiency that we all desire.

Cody Irwin: With that, I'm gonna hand it over to Tom, who's gonna walk us through a series of demos, kind of outlining some of those things we talked about. He'll wrap up his demo section, talking more about a framework for actually going from idea through that

Cody Irwin: implementation. So, we want this to be real. Our goal coming from this conversation is that you have tangible next steps and things you can do. So, with that, Tom, I'll pass it over to you.

Tom Pugh: Thanks a lot, Cody, and that was great. I do just love that quote about what would you do if you had a thousand interns. Made me think already, just in that… whilst Cody was speaking about some work that I could…

Tom Pugh: definitely automate. But yes, in terms of the demo, we have 3 really, really exciting examples to show you that showcase some of those things that Cody mentioned.

Tom Pugh: In terms of within a C, so, going back to that matrix, within a C couple, which are more increment, so you can already have a think about potentially where… areas where you could bring this into your work. And then we've also got an example of an innovate, you know, how we can really rethink how we, solve a problem.

Tom Pugh: So without further ado, let's see these in action.

Tom Pugh: So, this first one here, this is where we're starting to look at how we can create static dashboards, which I'm sure we all have.

Tom Pugh: and turn them into more intelligent advisors. You know, really remove that burden, where you're looking at, let's say, a pie chart or a visualization, and having to make that decision of what's the data's telling you. Why don't we have that intelligent advisor to give you that recommendations based on what it's looking at?

Tom Pugh: So, for actually this particular use case.

Tom Pugh: we worked with a major US film production company, and they came to us with the brief saying, sorry, every Monday morning, after the weekend, in terms of the box office, their executives had a 9am meeting, and they wanted to understand the performance of the latest films.

Tom Pugh: When they go into the office first thing in the morning, the last thing they want to do is be interpreting that data. So here, for example, we can start to see that intelligence advisor come to life.

Tom Pugh: So, at the top here, we can see that recommendations in terms of the key highlights, where it's getting this data from. In this case, a lot of data from, let's say, IMDB.

Tom Pugh: And this is complementing your kind of traditional BI visualizations as well. So you can see those KPIs, you know, the gross last week, the ratings, and if we scroll down even further, you can, again, see more of these visualizations. So this is aligning to that foundational pattern that Cody was mentioning in the last slide about automated insights.

Tom Pugh: But if we scroll down here, he did also mention about sentiment analysis, you know, how can we analyze unstructured data? So here, for example, we can see for certain films, we can get that recommendation or analysis straight into the, into, the dashboard.

Tom Pugh: So it creates a really intuitive experience for those executives as soon as they go to work on that 9am meeting, meaning they are prepared.

Tom Pugh: For the eagle-eyed amongst you, you may have noticed there was also an audio player at the top right. So, if we click on that.

Tom Pugh: we'll be taken to another screen. And this is where we start to innovate. We saw the increment of, actually, that automated analysis. Here, we'll be able to increment of actually redefining how we're distributing and communicating data.

Tom Pugh: So I'm sure many of you look at reports, some of you may use alerts, some of you may get scheduled reports into your email inbox, but I wonder how many of you have listened to your data via a podcast.

Tom Pugh: So this was the brief that we got from this company, was that they really wanted to, in… for that 9am meeting, a lot of them were driving into the office, didn't have time even to look at their laptop, so they wanted a podcast which they could listen to on the way to their, onto the way to their meeting.

Tom Pugh: So, just to show you how this is done. So, this is working in tandem with 11 labs, showing the flexibility of the AI service layer here, where we can, work with other, AI services. And so, if I click on generate audio here.

Tom Pugh: I, as a user, can choose how exactly I want this podcast to be. It's not just going to be a predefined one which is given to me, I have complete control here in terms of what I want to see. So, for example, I can put an executive summary, I can choose the films that matter to me.

Tom Pugh: I can then choose some analysis that I want, I can choose the time frame, I can choose the different language, and then I can even, just put my name there as well, and also share that to other users.

Tom Pugh: So I won't press submit now, because that'll just, work away its way in the background, but I have just loaded up a few examples, just to really show you how cool this is. So if I press play here… opened strong at the domestic box office on June 20th, 2025, earning over $14 million on a…

Tom Pugh: Awesome, I think that's so great. But yes, in terms of what we're doing there, so Domo is analyzing all this data using our AI Agent Toolkit, which we'll see in a bit, and working with 11 Labs to do that audio, text-to-audio, that you heard just then. So, yeah, really cool use case.

Tom Pugh: But I really challenge you as well to think about, you know, how you're communicating with your data. Is just reports enough? Or how can we embed that in people's workflow to enable them to be more productive with the work that they do?

Tom Pugh: Great, so that was one example. Let's go over to another example, just to bring this life even more.

Tom Pugh: So, completely different use case. We're now looking at fraudulent transactions. This is where every second counts. You then have thousands of transactions coming in, in real time, and you want to enable users to make decisions

Tom Pugh: That can have, ultimately, a huge impact, you know, saving companies' reputations, being, saving your customers' reputation of yourself. A really valuable, use case. The problem is, most detection systems are very slow, they're manual, they're siloed.

Tom Pugh: But by the time the fraud is spotted, the damage has already been done.

Tom Pugh: So, this is an application we've built that can help change that.

Tom Pugh: By connecting to relevant data sources, transactions, accounts, patterns, external signals, we can continually analyze that data in real time.

Tom Pugh: So here you're seeing some of the data that we're bringing in, but this is where the agent starts to come in. So this agent can work in the background, in real time, and start to look at them, and flag them for review, pass them through, or even block them in some cases.

Tom Pugh: And the key thing here is having that human in the loop to then make the decision to where to go next.

Tom Pugh: So we're empowering the human here. This is done by what the agent is doing in the background, and we all see this, is that it's going through a couple of steps. One is it's detecting the transaction, it's then generating a score in terms of how fraudulent it could be. It's then creating a case.

Tom Pugh: And Bennett's recommending in terms of what actions should be taken.

Tom Pugh: From there, the human, or the end user, can then view this. And you can see here, these are the ones which have been flagged for review. You can even see the ones which have been blocked automatically.

Tom Pugh: But this gives all the detail to the user to then make that decision, rather than having to manually annualize all those ones, and then try and think of what to do next. Really presenting that information to the human to empower them to make that decision.

Tom Pugh: ultimately, when they've made that decision, that AI agent can learn, it can update, it can then provide better recommendations going forward. So it's a really 360-degree kind of loop in terms of how we can improve fraudulent detection.

Tom Pugh: So, let's see behind the hood, and actually let's see how we can create this, and how easily those who use Domo can build an agent with the data that they have already.

Tom Pugh: So…

Tom Pugh: Cody alluded to it in the first half of the session, in terms of using our agent capitalist framework. So here is that toolkit in terms of being able to build the components for that agent to work.

Tom Pugh: Really intuitive, really simple, using Domo's low-code nature.

Tom Pugh: So, let me just walk you through some of the key things that you're seeing here.

Tom Pugh: So, firstly, we'll see the prompt. So you'll see that this is where we can put the goal of what we want this agent to do.

Tom Pugh: We can also see the instructions. So this is actually the manual of what we want, and the guardrails that we want this agent to run in. So think of it like, let's say you do have that intern joining on your first day, these are the, the principles that you'd like them to adhere to.

Tom Pugh: So here, for example, you're giving them the information of, you know, how you want, let's say, the justification to be given, and how you want the recommendation, etc, etc.

Tom Pugh: I also, at the top, you can also see Model. So, with Domo, you've got DomoGPT, which is provided by, built off of Claude and Propic. But you can also bring your own model as well.

Tom Pugh: Now, if we go to the toolkit.

Tom Pugh: you can… this is where we give the tools for the agent to run. So we have an arsenal of tools which you can choose from, and this can enable them to do exactly… the agent to do exactly what you want it to do. So that could be, let's say, sending an email, or triggering an alert to Cody if this agent spots something.

Tom Pugh: Then we have the knowledge. So this is where we're providing that information for that agent to then, to run and to do that set of instructions and goals that you set in.

Tom Pugh: So here, we've got structured data. We can also provide unstructured data as well, so similar to what we saw with that sentiment analysis with the Aurora films example.

Tom Pugh: And the great thing, and this is what I love about Domo, it's all transparent to the user. So we can actually test this, and we can see behind the scenes, when it's looking at this data, exactly what's going on, what the agent is looking at, and the answers that it's providing. From here, we can tinker with it, we can change it, we can adapt the instructions based on what we're seeing.

Tom Pugh: So if I actually… it does take a few seconds, so I've just loaded one up, which I did just before the call, before the webinar. So here, for example, you can see the information that it's processing.

Tom Pugh: And then you can see also the recommendation at the end. So from here, this is where I can start to tune it. If I'd like to see something else, this is where I'd update it, test it.

Tom Pugh: and then it's ready to scale out. To Cody's point earlier, you know, building POCs, building that momentum, it's all about doing it quickly. Back then, you can build momentum and start to scale out and build more.

Tom Pugh: But yes, you have that right in the palm of your… you have that right in the screen. I think it's a really useful tool in terms of testing these AI agents.

Tom Pugh: Great, so that was the second example. The third one is we're going to see a different example of how we can use,

Tom Pugh: AI services within Domo.

Tom Pugh: So this is actually a chat interface that we can use for users to ask questions in natural language about the data that is behind it. What we're enabling here is turning scattered content

Tom Pugh: Across your libraries, into searchable intelligence.

Tom Pugh: So we're having a searchable intelligence layer over your scattered libraries.

Tom Pugh: scattered content.

Tom Pugh: So, in terms of where you can do this, you can leverage this in your portal, you can leverage this in an application, complementing the visuals that you already have.

Tom Pugh: In this terms of a specific example, what you're seeing here is actually our support assistant. So here, behind the scenes, we have thousands of our articles, which we have online, which some of you may have already used.

Tom Pugh: We have tickets that have been raised, we have internal support websites, and this has been surfaced within this chatbot.

Tom Pugh: It's also including unstructured data. So yes, we have unstructured data, but also unstructured data. Think of, let's say, a website, so all those… that web data that you see on, a web page.

Tom Pugh: And what this does is it enables us, or as a customer, for you to be able to type in natural language and get an answer across that data. So, you can see here, for example, it's asking me, it's giving me some recommendations, and I can even ask some questions in natural language. For now, though, I'll just actually just click on that first one. How do I create my first chart in the domain?

Tom Pugh: So what it's doing here, behind the scenes.

Tom Pugh: It's taking that information I've put, it's then processing it, and then looking across the different, the content that I provided it to provide me the best results.

Tom Pugh: So here, it's providing that answer in terms of how I can create my first card, or visualization in Domo. And you can see here, for example, it's saying, you know, create the dataset. Once you have the data, let's create it using our analyzer toolkit.

Tom Pugh: From here, though, I can, see a couple of things. So, yes, I get the answer. I can also see the sources of data, where this is coming from. So, again, having that transparency, so I know exactly where this agent is looking, and I can look into it.

Tom Pugh: I can expand this even further and actually see the information behind the scenes, from a back-end perspective, exact quotes within these articles where this will be lying.

Tom Pugh: So, for example, if I click on this link at the moment, it'll direct me straight.

Tom Pugh: sorry, to a support article where that is. A great way to quickly get access to the information that matters without having to painfully search for it.

Tom Pugh: Now, the great thing about this as well is that,

Tom Pugh: Oh, sorry. The great thing about this is that I can type in more questions, and it will remember the context of what I do. So let's say, for example, we said, what about…

Tom Pugh: for connecting.

Tom Pugh: Salesforce data.

Tom Pugh: So I've asked it another question.

Tom Pugh: And it's just having to think, whoop.

Tom Pugh: And it's just having a think.

Tom Pugh: Let's just close that one there so you can see it better.

Tom Pugh: You can see it's come back with an answer. What's great is that it's remembered the context of my previous chat with it. So, it says for about chart discussions, it remembers that, so again, it just walks me through specifically how I could set it up for Salesforce, and then how can I create a visualization just specifically for this.

Tom Pugh: great use, which can be embedded, let's say, in your intranet portals, alongside your visualizations, as I mentioned earlier.

Tom Pugh: But I…

Tom Pugh: And I think this is a really good example of what we've seen across these three, is how we can use AI in different applications for different use cases. And I really challenge you to have a think about where, potentially, you can apply these.

Tom Pugh: Great, so now we've seen those three, let's see… now, I'm sure a few use cases are bubbling to mind. Let's see what you can do, then, in terms of rolling this out, at your organization. So let me share screen and move on to… back to the slides.

Tom Pugh: Awesome. Sorry, I'm just gonna…

Tom Pugh: Sorry, too many things open, and great, we can get started now on the rollout framework. So, as I mentioned, so you may have some use cases forming in your mind. Where do you go from here?

Tom Pugh: So this is what we've seen as really successful, is working with our customers, and it's creating a structured path from ideas to production-scale deployment. And the aim here is to create use cases that can be then

Tom Pugh: For a process which is repeatable, it's well-defined. And as I'm sure you'll find out, is that once you build one, the momentum will start to build, and you can start to build more, and then start more to go to life.

Tom Pugh: So let's just walk you through from left to right in terms of what we're seeing here.

Tom Pugh: So, if we start on the left-hand side with the use cases.

Tom Pugh: you may have already thought of some already, but the aim here is to really build as many as you can, across your organization. So that could be using, let's say, interviews or focus groups to really understand the pain points, the tools, what could save people time, what they don't like, what they do like.

Tom Pugh: And from here, then we can start to look at how we can map them to your… the technology that you have within the company.

Tom Pugh: And from there, it's all about prioritizing them, and working out what, for you, is the best one to start with.

Tom Pugh: For everyone, it's going to be slightly different depending on the priorities, and we can help on the next slide, just in terms of how we can help ascertain which one to go for based on your priorities.

Tom Pugh: But the aim there is to choose one which you think is high impact for your organization, but also quick to implement.

Tom Pugh: And from there, once you've defined that example use case.

Tom Pugh: we define the POC requirements, and those are really important. And the reason I say that is because when you start to build out that POC, you should always compare back to those requirements that you said at the beginning.

Tom Pugh: So, for example, that could be recording how long it already takes to do that current use case without AI, and then once you do the POC, then comparing it back and forth. So you have that means of comparison.

Tom Pugh: Then, so once we've defined them and you start to build that POC, we then really want to take in that user feedback throughout that whole process.

Tom Pugh: And in terms of that process of how long we recommend it takes, I suggest about 4 to 8 weeks. That's what we see as a really successful. You know, it's quick enough that you can, build something which is really tangible,

Tom Pugh: And… but at the same time, quick enough that you can…

Tom Pugh: move on if something doesn't work, or try something else. The aim here is that you can build that momentum.

Tom Pugh: Now, in terms of vendor deployment, so yes, you may… that may be… you've done this POC, it's been approved, you've had the business case approved, it's now then deployed at… and then at scale.

Tom Pugh: So the aim here is to then be able to help, you know, potentially, you know, the change management, countdown comms, and it's also to always keep that feedback channel open. You know, that age-identic use case may develop in a month's time, a year's time.

Tom Pugh: The aim is to always iterate, update it, and always have that feedback channel open.

Tom Pugh: That's where we start to see the real success.

Tom Pugh: And as I said before, I'm sure as you build one, and you start to deploy it.

Tom Pugh: The next ones that come down the line will be much quicker.

Tom Pugh: You'll start to learn from those ones that you've already done.

Tom Pugh: In terms of how Domo can help, so as you've seen on some of the slides and some of the things I mentioned, we have, you know, blueprints in terms of how to build a use case, what components are needed.

Tom Pugh: how we can help in terms of the change, material as well. So, do come to us in terms of how you can help across that whole, end-to-end cycle.

Tom Pugh: So that was more about the methodology. You know, how… what does that mean from a value perspective for your organization?

Tom Pugh: So, here, what we can see, if we see that kind of scale going from the bottom to the top, here we… it's going back to what Cody was mentioning about compounding the value. So, you may start seeing initially that those more potentially individual siloed tasks, like the ones where it's manually saving you some time.

Tom Pugh: When they start to compound, and you start to see that go across your team, across business units, across the whole organization, what you'll see is start to see a much larger term, sorry, a much more, kind of, enterprise or bigger transformation across the organization.

Tom Pugh: And when we see that, we see these four pillars kind of being, or four main pillars being hit in terms of what we see there. From a data perspective, we see, let's say, developer acceleration, i.e, how can we reduce the time building products?

Tom Pugh: We then see self-serve analytics, so getting people to access data to questions quicker. We then have intelligent automation, so, you know, how can we run the business more effectively? And then finally, we've got the real-time operations.

Tom Pugh: So how can we have proactive decisions instead of reactive firefighting? Similar to what we saw with that Freudius example.

Tom Pugh: But I'm a strong believer about when you have a knock-on impact from those important pillars, and from those… that kind of scale that we saw there.

Tom Pugh: Wider organisational goals will be hit in terms of increased employee satisfaction due to less, you know, manual work, doing more,

Tom Pugh: Higher value tasks, as opposed to, again, those manual tasks.

Tom Pugh: And what that'll enable is, you know, you'll start to see those higher ROI, start to see more profit, as an example.

Tom Pugh: In terms of, again, how Domo can help, so throughout, you know, working with us, you'll get a dedicated customer success manager. We also have technical consulting support who can implement those potentially more nuanced AI use cases. And we also have a great training library in terms of Domo University, where you can take certifications and really learn how to use the platform.

Tom Pugh: Great, and then one more slide, and before I hand back over to Cody, but…

Tom Pugh: You may have… you may have already, as I mentioned, just thought of some use cases.

Tom Pugh: And you may be thinking, if I have tons of use cases, where do I start?

Tom Pugh: So the aim here and what we're trying to work with our customers on, is how we can enable them to objectify, objectively compare and work out which use cases to go with.

Tom Pugh: And to do this, we really recommend on two things. One is that impact, and one is that effort in terms of implementation. So you can see this matrix on the left-hand side, where we're saying those with, you know, higher impact, but low effort, that's really the sweet spot. Those are the ones which you want to prioritize.

Tom Pugh: For the ones which I consider, you'll have those ones which may be of high impact, but may be a bit more legwork in terms of effort to implement.

Tom Pugh: And likewise, on the other side, those which are low effort, but maybe not of much impact, still worth doing, because potentially you can develop them really quickly. And again, going back to building that momentum.

Tom Pugh: Those for deprioritize, maybe have to come back to another day, potentially when your team has started to develop the skills of using, let's say, how to build AI agents, but never to take them off the list, because that will easily come back into the pipeline and still… and actually go to that prioritize section down the line.

Tom Pugh: In terms of actually those specific… well, actually, sorry, let me rephrase that, in terms of impact and effort, in terms of those specific requirements of what breaks down into that, it's really… it's up to you, and then it's going to be different across different companies.

Tom Pugh: So, for example, from an impact perspective, you know, it could be your speed to value, it could be how you work with your partners or your customers.

Tom Pugh: From an effort perspective, it's more about, you know, how long it takes to implement the build effort, who you'll need to get involved to do this.

Tom Pugh: And from there, again, you can scale that. And by having this matrix, what you enable is have a process which you can go through every time you think of a use case.

Tom Pugh: And the great thing about this, and it goes back to the other side about, you know, it's not just a one-stop shop of going left to right, it's a whole loop in terms of you're constantly feeding back, feeding back, sorry, into this… into this funnel.

Tom Pugh: So you're constantly evaluating what's a priority to you and what's not. Some use cases may go up in the order, some may come down.

Tom Pugh: The aim is to keep this fluid, and up-to-date as you develop more and more use cases.

Tom Pugh: Great. That's it from me, and what I'll do now is pass over to Cody, who's got a great offer, to share with you all.

Cody Irwin: Awesome, thanks, Tom. I really appreciate that. I want to kind of just reiterate some things Tom shared there that I think are valuable. Priority is everything, and we're in this world right now where finding the right ones to work on

Cody Irwin: Can be a little daunting.

Cody Irwin: We are finding, too, that there's evolutions happening in how we prioritize. Like, effort has always been, I think, probably the stronger driving factor for the things we go after. There's been things that we've viewed as incredibly valuable, but they're too hard. AI is also changing that conversation quite a bit.

Cody Irwin: So, yeah, priorities… priority's king. You need to figure that out. We want to be there with you. When it comes down to it, I think a lot of companies, a lot of individual… a lot of individuals are looking for… for friends in this space.

Cody Irwin: for those to collaborate with, to think through this, to bounce ideas off of, that's our offer today. Like, that's what we'd love to kind of end this with, is… is an ask of you.

Cody Irwin: To talk to us.

Cody Irwin: We'd love to come in and do an AI workshop and talk through opportunities, talk through how we prioritize

Cody Irwin: And really kind of go deeper into this. I want to share one quick story. I know we're almost at our 45-minute mark for the scheduled time, but I think this is probably relevant to a lot of you on the call. I was in Australia doing customer visits about 3 months ago.

Cody Irwin: And at the beginning of one of those conversations, we went around the room and asked everyone what they wanted to get out of that session.

Cody Irwin: To ensure that we hit the right topics. It got to one individual, and their response to me was, I want to learn how not to lose my job.

Cody Irwin: It's AI

Cody Irwin: And I'm sure us as individuals are wondering that. I think as companies, we're wondering, how do we stay in business? How do we stay competitive in this new space?

Cody Irwin: I think a big part of it is finding people you can talk to, and think through this with. This is what Tom and I do day in and day out.

Cody Irwin: is have these conversations. So please.

Cody Irwin: please, please, engage with us. Scan that QR code or click the link that was sent over the chat.

Cody Irwin: We are here. We're here to listen.

Cody Irwin: We're here to give advice, we're here to give examples, we're here to share what we've learned so far.

Cody Irwin: I want to reiterate, too, we have Q&A. We'll stay on the line for a little bit longer and ask a few questions for those that can stay. For those that have stayed this whole time.

Cody Irwin: Thank you so much. I know we've gone from philosophical to theoretical to tactical to…

Cody Irwin: deployable, all kinds of bubbles, and it's a lot of stuff to cover, so thank you so much for being here. We did have two questions come in during the actual session, as we were talking through topics, so I'd love to address those first.

Cody Irwin: The first one's actually probably one for Tom. The question was asked, Tom, what large language model do you use?

Cody Irwin: Can you go into that for us?

Tom Pugh: Yeah, great question. So, we use Domo GPT, so, which is built off Cloud Anthropic,

Tom Pugh: just to reinforce, though, what that means when we say Domo GPT is that when you type… when you use the AI service layer within Domo, all your data and your questions remains within Domo.

Tom Pugh: And we never use that data to train the model. But yes, in answer to your question, we use Cloud Anthropic.

Cody Irwin: Awesome, thank you. And one thing out there as well, we took that approach at Dome with GBT largely because a lot of companies were wondering how to keep data secure.

Cody Irwin: So we're like, hey, let's solve for that problem.

Cody Irwin: And so it's fully sandboxed, fully isolated, no retraining, like Tom mentioned.

Cody Irwin: But we've also taken an open approach. We realize a lot of companies are kind of backing into an enterprise-adopted LLM, so we have a very rich model management layer where we can swap models out. So if you say, hey, we don't want to use demo GPT, we want to use

Cody Irwin: whatever we've embraced as a company, we can use that. So, a lot of flexibility there.

Cody Irwin: The next question, that came in was, how is Domo different when it comes to AI?

Cody Irwin: That's a big question. That's a very big question. So whoever asked that, thank you.

Cody Irwin: And it's one that we've been trying, as a company, to figure out how we best position. Like, this is a new space. This is moving very quickly. We have been historically a data company. That's where we focused. So one thing that we are seeing as differentiated is the strong foundation. Back to the… that flexible foundation, the ability to acquire data, to secure data.

Cody Irwin: to imbue insights and govern that appropriately is really interesting for a lot of companies. We tend to be very fast there.

Cody Irwin: We've also found that the other side of that that's really interesting is deployability.

Cody Irwin: I, I was at a, I was at a conference.

Cody Irwin: like, 8 months ago, talking to a very large U.S. manufacturer about AI, and they're using a very technical platform.

Cody Irwin: to build some very advanced AI solutions, and they came to me like, hey, we're using this thing, we're intrigued by Domo, but, like, why should we consider even using that? We're doing all this cool stuff over here.

Cody Irwin: And I asked the individual, like, well, how do you deploy it?

Cody Irwin: Like, once you do the cool thing, how do you actually get into the hands of those that are gonna use it?

Cody Irwin: And that's one area that we tend to be really good as a company, is getting solutions into the hands of businesses, of partners, and of customers. So, very good at acquiring data and securing that, in concert with our cloud partners, as well as

Cody Irwin: Deploying that to the business user.

Cody Irwin: So yeah, and we'd love to kind of show that to you. One thing I found is really helpful in these AI working sessions is to actually work through examples.

Cody Irwin: to kind of walk through it, and in some cases, maybe, like, Domo's not the best fit for that, and we'd be happy to recommend that. We'd love to find those areas where we can help as well, directly.

Cody Irwin: Okay, with that, I don't see any other questions.

Cody Irwin: So we'll wrap this up. Again, my closing statement to you, my ask of you, our ask of you is, please talk to us. Like, we're here. We want to help.

Cody Irwin: We've learned a lot. We love to share what we've learned. We love to hear what you've learned. We feel like this next… that there's a…

Cody Irwin: Not to get overly, overly sappy, but there's an old African proverb. If you want to go fast, go alone. If you want to go far, go together.

Cody Irwin: And I will say going fast sometimes in AI is going the wrong direction. I think all of us are looking to go far. So, this is a chance for us to really partner well as organizations. Oh, and as I was waxing philosophical again, we have one last question come in.

Cody Irwin: So, I'll cover that one as well, then we'll wrap this thing up. How is the ETL process during implementation?

Cody Irwin: That's probably a question that's best suited for Tom.

Tom Pugh: Great, it would be great to understand in a bit more detail what that is, but, during the implementation, so when we look at that use case that we mentioned, or that whole process, when we define that use case, it goes back to Cody, what Cody was mentioning, we really understand the tools that are best enable to make that use case come to fruition. So, that may be tools

Tom Pugh: not Domo, maybe Domo, and then we, from there, we really work out step by step of how we can then enable each of those tools to enable that use case. And specifically, from an ETL perspective.

Tom Pugh: We do have a great ETL tool called Magic ETL within Domo to do that transformation, and we can sit on top of your, you know, cloud data warehouses, or we can even bring that data into Domo to do that.

Tom Pugh: So yes, I appreciate ETL as a broad, term, but yes, we can do that within Domo, but the main thing is really going back to those use cases and working out how best we can use the cool kit that you have to enable what you want to implement.

Tom Pugh: Cody, I don't know if there's anything to add on to that.

Cody Irwin: Yeah, that's… that's awesome. ETL's a big topic, there's a lot we can do there, we've got a lot of tech in that space.

Cody Irwin: We are seeing a shift in the market as well, and we are rapidly embracing this around

Cody Irwin: more, kind of, ELT-type models, where we can define semantic concepts on the data. So Domo's… Domo is rolling out a new feature called Models.

Cody Irwin: Which is a little… it's a slight departure from ETL. ETL is very focused on processing data and transforming data.

Cody Irwin: Models are a little more focused on helping the system understand how the data relates, how it works together.

Cody Irwin: So, that's one of those concepts that we can explore, obviously, more deeply, in a direct conversation. There's a lot of paths. One thing DOMA's very good at is data pipelines. And again, we're seeing a massive need for that in this AI era.

Cody Irwin: Again, thank you all so much for your time today, we really appreciate it, and we're looking forward to the follow-up conversations. We'd love to go deeper. Thank you.

Tom Pugh: Thank you, Cody. Thank you so much, everyone.

Tom Pugh: Thanks, Katie.

‍