Making AI Work

Hello. Good afternoon, or maybe good morning, depending on where you're joining us from. Welcome to this webinar. Thank you all for being here.

We're going to be looking today at how you can make AI work in your organization. By way of introduction, my name is Jamie Morrison. I'm the field CTO for Domo here in EMEA, and I'm joined today by my colleague, Sasha. Hey, Sasha.

Hi, Jamie. Thank you. My name is Sasha Autenrieth, and I'm a senior solution engineer and SC team lead for the EMEA region. So today, together, we're going to look at some of the challenges around why organizations so often struggle with AI use cases, and that will include looking at what we see in the market.

So in the research, as well as what we hear firsthand from our customers. We will then see how we work with AI use case ourselves. So with some practical field-tested advice and frameworks that we will then share and that you will be able to use yourself. So let's really get into it.

And Jamie, could you please get us started by setting the scene, and maybe illustrating why we feel today's session actually matters? Yeah, absolutely. So I want to open up with some numbers, some stats. Yeah, to kind of paint the picture and set the scene a little bit.

So yeah, Sasha said we're going to look at some research that we've used that's out in the market. And part of that is this. The numbers you'll see here are from a recent KPMG report. They very recently, the last few weeks, released their global AI pulse report for Q2.

And there's some really interesting findings that we picked out of this. So they surveyed as part of this report, over 2,000 senior leaders across 20 countries, many of them EMEA markets. What was interesting is they found that 76% of organizations that they've surveyed now say AI is delivering meaningful value. And this was something that was up 12% in a single quarter, versus that Q1 report.

So it's showing us that AI adoption is racing ahead. It's growing. However, what was particularly interesting was the share of respondents who could point to established or proven return on investment was actually only 7%. And that number barely moved between their Q1 and their more recent Q2 report, even as we see adoption doubling.

So it isn't necessarily the case, let's say, that AI doesn't work. Value seemingly is everywhere, but proving that return is still really rare. So that gap between something that feels useful and something that is then making P&L impact, financial impact, that's the really difficult part for most organizations. So maybe begs the question, why is that gap so persistent?

Why didn't that 7% increase quarter over quarter? And we often find from the work that we do at Domo with our customers, it comes down to a few things. And in most cases, it's not a technology problem, right? It's not the choice of the model.

It's not your LLM that you've selected. It comes down often to things like the data not being ready. Maybe the correct use case isn't being selected. Often we see organizations chasing something that's really exciting, rather than actually picking something that's really valuable for the business.

And what we're seeing is, yeah, organizations that are pulling ahead, getting more value, they're not the ones with the cleverest AI, the best model. They're the ones being most practical about it. And that's what we want to give you today, is this practical, repeatable way to find those AI use cases that are actually worth doing, how to prioritize them, and then how to get them live. Okay.

Thank you, Jamie, for providing us that overview of what the research is currently saying. In addition of what we see in the market, and over the past couple of months, Domo has taken part in hosting a series of roundtable events. So where we brought together senior data and technology leaders in London. So CDO, CIOs, chief data and analytics officers.

So across a lot of different industries or sectors, such as retail, housing, FMCG, research, recruitment, or travel, for instance. But that maybe shared the same challenges or where we were able to discuss the same challenges around AI, AI use case, and really what it takes to move an initiative that is at the pilot or maybe at the proof of concept stage into actual production. And I think, Jamie, that would be great if there were some insights or challenges that you can share from those sessions with us. Yeah, definitely.

It was great to be a part of these sessions that we ran, these roundtable discussions. Lots of senior leaders in one room, kind of discussing their pain points, their challenges, with getting AI use cases from pilot into production. The thing that I would mention, first of all, off the back of these discussions and roundtables, is that no one truly has this all figured out, right? Everyone's still kind of figuring out the best case or the best ways, rather, to get the most from AI.

Perhaps if we look at those stats from earlier on that you can still see on the screen, maybe we can say 6 or 7% of organizations have it figured out, and they're getting AI to deliver tangible ROI. But the vast majority of us are not quite there. And what was especially interesting across the sessions that we had, was how many common themes there were across all of the discussions. So the same three themes reoccurred in practically every session, and I want to spend a little bit of time now talking about them because If you recognize yourself in these challenges, apparently you're in very good company.

I'd also say as well, these discussions, roundtables, these weren't rooms of AI skeptics. These were leaders running real AI initiatives, many with results to show for it. Which is what made these consistencies, these kind of themes that we saw again and again, so interesting. And so the first one is data foundations.

I'm sure you've heard this term, if you've been trying to implement any kind of AI use case in the past 12 months, but every discussion came back to it. And I think one way to look at it is, I've worked in data and analytics for over 10 years. I think anyone that's worked in data and analytics for any period of time, we've known this importance of a strong data foundation. We have this age-old concept of garbage in, garbage out.

It's always been relevant. AI hasn't created this data problem for us, it's just made it impossible to ignore. And so, one attendee we had in the roundtable, they put it really well, put it really aptly. She'd been making the case for cleaner data in her organization for 18 months and was getting nowhere with it.

Then AI arrived and, suddenly this garbage in, garbage out problem, it became a board-level problem. It became a board-level discussion, because AI forced it to be. So it's really started shining a light on the importance of having a strong data foundation in place. If you're rolling out a chatbot to your organization and someone asks a question, they get a poor response back.

Many cases we can point at having an immature data foundation set up, that could be the root cause of that. Now, typically, nobody's data foundation is perfect, and that's okay. The trap I feel is thinking you have to fix everything before you start using AI. We typically say that's not the case.

The teams that we see doing this really well, they're running the foundational work and the experimentation with AI in parallel. You don't need to do one then the other. And I would also say start small. You don't need to, like we say, build out your entire data foundation before you can start an AI initiative.

Map out the data you need for one use case, start small, build the context, the requirements, connect those sources together for your use case, and get that one off the ground to begin with. And I think this is where Domo really shines as well. We've been doing this since day one, building these data foundations. We have our library of connectors.

We can connect to cloud applications, on-premise databases, structured data, unstructured data. We can transform that. We can put our consumption layer on top and build agents on top of it, which is really powerful. Can let users take action on those signals.

Everything's governed, everything's trusted. So, it's something that I think we're really strong with. But again, this data foundation came up again and again. The second key theme that reoccurred was culture.

And this is one that, from our discussions, people probably underestimated the most of these roundtable discussions. And it's maybe the hardest part. One director gave us the example in their teams, in her organization, they pull their own data extracts for their reporting from various source systems. They build their own metrics.

They build their own reports. But when it all goes wrong and everything is incorrect, they go and complain to the data team, that the numbers are incorrect. The technology to give everyone that single source of truth had existed for years. They invested in a data warehouse.

The behavior, however, to actually have teams use it was the hard part. In this case, it's pulling a team out of self-serving with Excel into a governed BI platform, but it applies to AI in the same way. So often, the point, I guess is, the blocker is almost never the technology. What it comes down to is trust, I would say.

It's changing how people actually work. And especially with AI, it's getting someone to rely on an output that they didn't actually produce themselves. It was generated by a model, but do I trust what it's generating if I'm going to attach my name to it? And something we're seeing about cultural change as well is that it has to be led and not delegated.

And that KPMG report that we mentioned at the beginning, it actually assigns numbers to this. So three-quarters of organizations now say that their CEO actively owns AI as a strategic priority, which I think is a great thing. Top-level executive buy-in is critical for success. However, only a quarter can point to the CEO as ultimately accountable for the decisions that AI is informing.

So we're seeing ownership and accountability as two different things, and that accountability matters more than the other when it comes to seeing ROI, return on investment, from AI. And so organizations with clear accountability, at the top level, at C-level, they're reporting return on investment at more than three times the rate of those without. So they're three times more likely to say that AI is delivering meaningful business value and that they're confident that their AI strategy will hold up over time. So something that's really important, having a named leader on the hook for outcomes, AI outcomes.

Who can override an AI output? Who steps in when something goes wrong? Who owns the running costs? Accountability at the top gives everyone beneath it permission to actually change how they work.

And then the third theme that I want to touch on is value. So culture was the hard one to discuss at the roundtables. Value was the one that made everyone go quiet. So everyone was able to talk about hours saved.

Okay, we've automated all these things. But when you actually push on Anything beyond an efficiency gain, like we saved a few hours. When you push on actual profit and loss impact, real commercial value answers and discussion kind of thinned out a bit. And it's not a criticism, I would say, of anyone.

It's just where organizations are at across the board in their journey with AI. We saw that number earlier on, right? 6, 7% of organizations only have a measurable ROI impact. So proving that commercial value is the really hard part. And something that came up in our discussions, the example I would give on this one, was we had one leader, director of a private equity-backed travel business.

What they did is they tied every AI initiative back to a defined work stream, and those work streams had monetary value or a dollar amount that laddered up to a strategic or commercial target. So in their particular business, they had around 90 of these kind of commercial targets, each one pre-validated against a business outcome, a strategic outcome. All of this was done before the data team even gets involved. And the way he put it was that the board or the CEO, they don't mandate AI.

They're not saying, "Go and use AI." What they do is they mandate an outcome. We've got these strategic goals and these outcomes we want to achieve. If AI helps deliver them, we're going to fund AI. But if not, that's also fine, right?

So I love this point, at the end where we see a lot of organizations trying to, I would say, shoehorn in or squeeze AI into a particular work stream or process. It doesn't actually require it. So I love that mindset that he had of it's just outcomes first, technology second, is a good way to think about it. So, just to kind of summarize there, we've got foundations, culture, we've talked about value.

None of these are actually technology problems when you think about it, right? The operational, the cultural, and from our discussions, the leaders we've spoken with, those that are treating those as actual disciplines from day one are having a lot of success. Perfect. Thank you so much, Jamie, for walking us through those actual concrete customer use cases.

That really brings it a lot more to life. What I noticed through all that you just showed us is that really those three fundamentals or pillar that we have now on the screen, so foundation, culture, and value, seem to kind of come back whenever we talk about making sure that a pilot or a proof of concept is successful. So my question to you is then, what does that tell us about AI readiness? So when do we know we're ready, and what do we need to know prior to actually starting a pilot or proof of concept to make sure that it would actually work?

Yeah, it's a great point. I think, we come to this readiness formula next. So again, what is readiness? What does that look like?

The way we piece it together is that governed data foundation. We've already seen how important that is. We need business context. What is the problem we're trying to solve?

The operational workflow that has got the business side, that technical side that come together, and then we have the people alignment, so that cultural side of things. How do we make it stick in the organization, right? If we miss any of those, typically POCs tend to stall or tend to fail. I think it's really easy for organizations to start up, spin up a POC.

It's very, very little barrier to starting a pilot. But the challenge is moving that into a production use case. If you have all of these four things in place and you select the right use cases, which is what we're going to look at later on in the session, is how we get there and how we pick the right use cases. But once you have all of that included, we see much higher levels of success moving from that kind of pilot stage into production.

And then there's the side of things, I would say that the piece that turns, let's say, a pile of pilots into actual return, and that's orchestration. So let's take an example of a dozen disconnected agents. They solve their individual use cases perhaps, but they never add up to much all together. And what we're finding is the value really shows up when your people, your data, your agents, your workflows, they're all sitting on one governed data foundation, and they can all work together.

They're all using the same information, the same data. And we find that AI is 20% algorithms and maybe 80% is that organizational rewiring. And that's really the gap I think that we're trying to close at Domo. Bringing data together, applying governance, applying tooling to build agents, and then applying the capabilities to have human oversight, human in the loop.

With all of that together, it becomes something you can run and kind of scale into production. Okay. No, and that makes a lot of sense. Thank you, Jamie.

Maybe then my question is, if we think a bit more practically, how do we actually get started? How do we know where we actually start in terms of the use cases that are worth doing? Yeah, I think this is where we come into our frameworks. This is kind of the really good question.

This is the kind of the first of a few frameworks we can share with you, right? This first one is kind of where you place your bets. This is where we always start, is how do you spread your efforts that you have. If I've got 100% of resource allocation to place my bets, where am I placing them?

And we think at roughly of the split of 70/20/10. Where 70% of your effort, you want to be putting that into things that you already do, but to do them better. So small improvements to workflows that already exist in your business, in your teams. Very often we see that as being the bread and butter.

It's where most of your early wins will come from. If you're just starting out, your early wins are in that 70% category. 20% is about properly reinventing a workflow, so redesigning it around what's now possible with agents and automation, rather than bolting AI onto an old way of doing things. And then 10% is the new stuff. These are your moonshots, the things that you just couldn't do before.

Are there new product offerings that we can incorporate? Are there better ways to serve our customers that just previously were not possible? And the reason I like this mix is it gives you permission to start now with the data you've got, with the processes you're already doing today. It's great that we've got that 10%.

We can take a moonshot or apply some work there, apply some effort there. But, most of that value that you'll achieve is in that 70%. And what we see is those small wins begin to compound. Each one builds a bit of trust.

If it's efficiency gains, it frees up a bit of time. And if we can free up more and more time across the team, those gains begin to compound. Over time, the next one becomes a little bit easier to get over the line. And then the second framework I want to share here is what we call the anti to-do list.

And this is really how we find use cases that are valuable for the team. Often, we go into customer meetings, and we sit down, we say, "Where do you want to use AI?" And you get one of two things. You get either silence, or you get these science fiction answers that are really difficult and these huge projects. So the anti to-do list is a good way to get past that silence part, I would say, and get really good grounded answers.

So instead of asking what are the new things you'd like to do with AI, we ask what work that individuals, teams would rather not do at all. So, that manual report that takes four hours to build, the month-end spreadsheets that we reconcile by hand every month, end of week status updates where we've got to go and get information from different source systems. These tasks that I would rather just not be doing. I want to be spending my time on higher value work.

And so this is a far better starting point because it's grounded in, I would say, it's grounded in real pain rather than this kind of hypothetical future that we could have. And just from our experience with our customers, some of the best use cases we have, this started life as something on this list. And if you go and try this in your teams, you'll be surprised how quickly that list fills up once your team realize you're asking about their actual week, their actual pain points, and not some kind of hypothetical AI wish list. So we've got this kind of mix that we've talked about.

We've got a good way to start getting some ideas of use cases. That's great. We've got a list of things we can start to tackle. So once we've got our set of ideas, we can then begin to figure out, okay, which ones are actually worth it?

Where do we want to spend our time? And so we use this prioritization matrix with these four quadrants. So we score every idea, every use case that we go into. We've done this internally with customers.

We score every idea on two axes. So business impact on one side. So that's things like the people affected, the time saved, the revenue it's going to generate, the cost, the decision speed. And then on the other, we map implementation effort, and this is the data ready, how complex is the build, how much change management's required, what does it cost to maintain?

And we measure all of this on a scale of 1 to 10. So we plot everything out on those kind of criteria, on our quadrants, and we end up with a set of use cases dotted around the plot here. And what we're looking for is those quick wins. So anything that's low effort but is going to give us high impact, we're starting there.

Those are the things we want to tackle straight away. Secondly, we've got big bets. So things that are high impact but a little bit harder, we want to plan for those. Everything else, we're going to park.

We're going to focus on those two quadrants to begin with. So this is a great framework. We use this with our customers. We've used this internally.

What I would say, yeah, is the next question is, where does this all live? I've shown it to you on a slide. So, we want to show you now how actually you can go and use this, how you can input your ideas. So let me hand over to Sasha, who's going to show exactly how you do that with our AI planner.

Perfect. Thank you, Jamie. Let me just share that with everyone. Okay.

Perfect. Okay, so thank you, Jamie, for showing us all a little bit more around the methodology and how you really get from those raw ideas to, I would say, like a governed roadmap and actual prototypes that will then turn into productionized solutions. And as Jamie mentioned, we've gone through the theory, and where now we want to see is where does that live, and how can you actually leverage that framework yourself. So the good news is what I will be showing to you right now is something that you'll have access to as well.

And when we look at it here, what's really important is we're not looking just as a whiteboard exercise. We're really looking at something that's live, that is continually improved on, and that allows you to have a concrete, governed, prioritized, and ever-changing, in some ways, roadmap to be able to have a great AI use case, AI transformation strategy. And we see here that those four stages all live within one single app. So those four stages are the process.

So the ideation process what are our pain points? What do we need to fix or improve? Or again, those 10% where we innovate, but it's also then turning those into concrete and smart plans. So like a structural blueprint of what we will need to start to build and look at, but also prioritizing them.

We might have great ideas, but again, we looked a little bit at the metrics, and we want to make sure that we know what are our quick wins, and what does require a little bit more investment, but will be big wins in the future. And ultimately, the last bit is a little bit more around turning this into a concrete app, a concrete solution, so leveraging as well AppCatalyst within Domo to be able to prompt this blueprint into an actual live prototype. One more thing to notice when we're on this screen is sometimes even with what Jamie just highlighted, it might be a little bit more difficult to know where to start or to have the inspiration. So we have a library of over 200 different use cases, and they are classified or categorized by different departments or different industry.

So just really helping you get started if you needed it as well. Another way, if we look here at the process that you can start to find the answers or start to get a template for those idea faster is as well to use this AI assistant here. So I've asked about a bit of a pain point that we see a lot of sales team have with the pipeline and the opportunities, and then with a little bit around the steps that they are currently doing manually. What we see here, asking a little bit more questions.

So we are looking at sales rep mainly, and what I will do is creating here kind of this overall process with a lot of different steps that we include for that process. Again, we're looking at a template so you can edit everything, but more importantly as well is for all of those steps that you don't like doing or that are a little bit always the same, you also can start to add a bit of duration. And as Jamie said, value is in a lot more than just looking at, okay, what's the cost of actually doing those tasks? But that's also part of the process.

So you do have the ability here for the different steps to add how much it would, here maybe one hour for that, a couple of hours here for the individual deals, and have all of those steps attached to your plan. Once you're happy with that and you have a series of different processes, so this is what we see here. We've done the exercise internally as a team as well within the region. So we can see we have lots of different ideas for those different teams.

But then we turn a selection of them and as well, the one I just showed you, just going to use a pre-filled one so we don't have to fill everything, but with a lot more of the detail around those use cases. And this is where we really think about it again as making that plan smart. It's not about, oh, I need a button here. I need this to be written that specific way, or like the detail of the actual app.

It's really about the outcome and the value you'd get from it with the understanding of what are the consideration around that use case. So you'll have a little bit of a description, use case name. Then we'll look at that tier or investment tier that Jamie introduced earlier. So what is it that we are doing?

Are we doing something better? Are we changing the way we're doing thing, or are we really innovating and doing something that technology allows us to do that we weren't able to do prior to that? We then have a little bit of information around the current state and the decision context. So again, who owns this decision and what does it matter for the business?

So really for any of those AI initiative, you need to know, concretely speaking, who do we need to talk to? Who will be backing up that initiative, and why do we need to be looking at that right now? The current state is a little bit more around like, what's the pain? Why are we even considering that use case as being one we want to focus on?

Another point, and maybe not the most glamorous one, but realistically one of the most critical one is around what is the data that is needed, but also where does it live? Is it ready in terms of the connection, the integration? Is it clean? Do we have the right level of granularity?

All of that kind of hygiene of data that needs to be done prior. So the foundation here for the data side of things. As well as the understanding of potentially some limitation that you could encounter based on the data sensitivity. You might have a great use case that's actually great for the business, not too difficult to implement in terms of effort or things like that.

But that is actually one that will take quite a long time to be implemented because of the sensitivity of the data and all of the approval or steps that need to be taken to make sure that this is a use case where we can leverage AI. Another important thing will be around what is this agent actual role? So is it here to give me recommendations, suggest what we could do? So in this instance, it probably will be suggesting next step.

Is it more of a copilot approach, where here, instead of just suggesting next step, it could produce an output that's a plan as to what to do next for that specific account. It could be drafting an email to send to the customer. So it's just defining as well, or the last one, just execute with that human in the loop. So it's defining to what extent does this agent support that use case?

Then a little bit more around the MVP. What do we need to prove to actually know that this is a solid use case, that we have trust in it? And again, for any project, any initiative, what does success actually look like? What do we measure it against?

And it could be things, again, that are more related to the time we spend or save by leveraging an agent, as well as the actual financial return. So in this example, if we're looking a little bit more at the sale pipeline and the portfolio, it'd be the increase in close one deals. It'd be the size of the deal. It'd be the speed at which we're able to close those deals.

But it's just quantifying it, giving it good period of time that's short enough that we're not doing a pilot that lasts three years, but also long enough that we can actually see results, and we know they are directly related to this initiative that we implemented. Again, a little bit ROI here, an estimate on the time saved. But the last part is also what I think is really important, and this is what will allow us to use that prioritization matrix that Jamie introduced before. And this is where we look at those two core pillar, so business impact and implementation effort.

If you're like me, sometimes scoring can be a little bit overwhelming in a way where, what is an eight, what is a ten? What do I even know where to get started? And that is why for each of the specific questions that we've given you, you'll see that when you scroll, you'll have the ability here to have a little bit of a description. So not only does it help if you yourself are not sure what a seven or a ten is, but also it means that if Jamie and I are both working, for instance, on AI use cases, we have a similar view on what the scoring actually means, and we have therefore, a more impartial scoring of the AI use cases.

To give you an example here, one that's really easy to understand and see is here, okay, we're looking at the impact. Are we looking at small impact, just a small amount of people in the company, large impact, or in this case, we're looking at all of the sales ecosystem, so probably quite extensive, and then it's a critical workflow for the company. And you'll have the same thing for the implementation effort. So in terms of the data, in terms of the culture, so important point as well that Jamie mentioned previously, so like change management, training, the risk that we see associated to the adoption, and then again, a little bit more around security, compliance.

As I said, not the most glamorous, but a very important consideration as well. When we are happy with that, we can save the plan, and we'll have a list of all of those different plans that we've filled with the actual score and just a high-level view of what it does. The next step is, as a team, we'll have put all of our ideas down, written plan for the best ones, but now we need to decide what do we start with next Monday. And so this is where this matrix that Jamie mentioned earlier really comes in.

We have those quick wins. Again, this is where we'll probably select a couple, two, three to start with. Do one, success, do a second, and then when we're happy with that, we know how those AI use cases, processes are going. We see how people are reacting to them.

Maybe we start to invest on the big bets. And we see here, for instance, the one we filled together on more of the sales opportunity and this deal intelligence AI agent. So this is really important in terms of making sure that you have your effort set on the right agent. Again, one thing I would mention is, as I said, it's not just like a whiteboarding exercise, one-off and done, is that this can change.

Something that maybe right now is a nice-to-have. Because we've made other changes before, because technology has changed, or because the market is changing, we might need to shift our perspective on different use cases. So anything that is here, I could go back, go to my plans, and review and reassess in the future. So you have all of that that you can track over time because it's all within the platform as well.

You can also see that investment mix that Jamie mentioned. So we can see for us, we have maybe a little bit too much in innovate or improve, and there are a couple of things where we should think about what can we do better, not just what can we do differently or what are cool new things we could do, but what can we do better here? And then lastly, you have the prototype. So here for the prototype, and let me scroll down to the last one because this will be the one that we've looked at.

But it just gives you a prompt that you could use directly into Domo App Catalyst. So App Catalyst will allow you to vibe code a prototype of this app directly in your Domo instance. So you'll have written here a prompt that can be then used in App Catalyst. And then once we've gone through all of that, and we'll go into this one really briefly, but it's like this element of, I have gone from the idea, I've built my prototype, and we have here an example of something that is very similar to what we use internally.

Still with that use case of deal opportunity, reviewing our pipeline, where here as a rep, I could log into that specific app. I can start to filter, so different categories that are interesting. I can sort by what makes more sense here, and then I can start to see, okay, this first here, Atlas Logistic, for instance, is the one I need to focus on first. And I can go into that one and have a little bit more here of an overview of where does that information come from, a summary of where we stand, the important dates that I have.

As well as all the milestones, the risk, action items, timeline. I can add some comments as well for my forecast, so we can have bidirectional integration, for instance. And we even have here more of that AI assistant where when I ask it, "Okay, take action." Here we see we're preparing the competitive positioning, and it's giving me already that kind of templated answer and even an email draft. So remember when we're talking about what an agent can be, this will be a typical example where it becomes a copilot.

So we're really working together. I want to be able to review everything. It's important for me to review everything, but we are almost a team tagging that together. So really going from that raw idea to going and having a productionized solution that can help you increase your processes, speed up the processes, get more return, and really have concrete value.

And before we go back actually to the questions, one last thing I wanted to say is, because we've covered quite a bit today, in the actual app as well, that will be shared with you. You will see at the very top you have a best practice tab. So here you have a little bit of a quick overview of what Jamie presented. So kind of your cheat sheet, if you like, of what we saw today, just to help you with that process as well.

You will have seen that here and there, those indication are already built in those different steps, but it can be a really nice addition as well to make sure that you have all the tools needed to attack your transformation as well. Okay. I think that's all that we wanted to show you here in terms of the actual demonstration and content. But it'd be great to start to look at questions if we have any.

Okay. I see one around teams getting stuck. Maybe I can take that one, Jamie, if you're happy with that. Yeah, go for it.

I think, yes, Jamie's given us a lot of tools already as to how you can approach it. What I would say is maybe think about questions that you could ask your team, and it could be like one of the ones Jamie mentioned is, what do you not want to do on Friday afternoon? Or what do you postpone all week long and end up doing on Friday afternoon? Another one could be, what's something I do every week?

Or what do customer ask us that we could do faster? Or maybe if you hired someone new tomorrow, what task would you like them to target first? So that's maybe one approach is more around those questions. One other thing that can be helpful is maybe start to think about foundational pattern.

So not just about what are your use case, but start to think, "Okay, where does AI shine?" And maybe that will give me ideas. So that could be sentiment analysis and classification. Do you have contact center call analysis, survey analysis, review of analysis you want to do? Is it an unstructured data pipeline, maybe invoice processing?

Again, call transcript, competitive analysis. So thinking about as an option, like what is AI good at, and go from there. And then ultimately, you do have those example that I showed at the very beginning. That can also be a good way to start.

Perfect. I can take the next question here. So the question is, everyone in our business can point to hours saved. A CFO wants peanut impact.

So I guess fundamentally the question there is, how do we quantify AI value? I think that's a good question. It's probably a hard one to get right. I think there's probably a few things to consider.

It's a bit of a multifaceted approach to do this. The first thing I would say is that app towards the end that Sasha just showed, there was that side of it where we can identify efficiency gains for a use case. So it'll say, okay, for our use case, I'm saving X amount of hours at a cost per hour. The cost per hour of automating this task is Y, and that gives me a number.

So, if something normally takes me two hours a day and I can reduce that to one, and I do it every day, I can figure out quite quickly the value of that automation. And if that's used by lots of people in the team, we can extrapolate that out to try and assign a dollar value to the use case there. Lots of efficiency gains. That's in that kind of 70% if we think of that mix.

I think for other use cases, it becomes quite difficult. If it's something new, we maybe don't know what we're saving. Some things might be a longer time horizon as well. For example, we might have a use case that's to reduce customer churn.

If our customer contracts are one to three years, it might take us a few months before we actually see the effects. And so we have to factor in these long time horizons to the evaluation that we make on whether we're getting value. But once we do start extending that time horizon, are we reducing customer churn? We can look at the percentage drop versus our book of value, our customer book of business, sorry, to figure out what that saving would be.

And I think something else to think about is costs probably in the last few months have shot up the priority list. If you're paying, token costs are going up as models get more advanced. So cost is always something to factor in. I'm sure you've seen stories as well.

Organizations are reducing budget for AI spending because they're not seeing the value. So it's something we're really transparent with at Domo, tracking token costs, input and output costs. So we want to measure, are we getting value from the money we're putting into the process as well? So measuring all of these aspects, but also setting your organization up correctly.

We heard earlier, organizations that have C-level buy-in and accountability are more likely to get higher levels of ROI. So at the use case level, there's things we can be measuring, but also we want to set our organization up to make it as successful as possible, I would say. I think we've run over maybe by a couple of seconds here, but if there's any other questions, we're happy to take them. I'll just wait a second to see if there's any other questions.

Perfect. If there are no other question, the one last thing I would say is that you will see a small blue button above our heads that says Domo AI Planner. So this is the link to the planner that we showed you. We'll also share it via email for the attendees.

And thank you so much for joining us today.