Ready to Put AI to Work in Your Finance Team?

So I think most of you will recognize at least one of these, right? Let's take a look. We have our analysts that are spending hours pulling data from different systems before they can even start. We have variances that can take days to explain a budget that's out of date a few months in. And for me, it's a bunch of side spreadsheets that never really make it back into the plan. What I want to look at is this call out on the right. That's where we should focus. Someone asks a question on Tuesday and the answer comes back on Friday. And by then the decision has usually already been made just without the actual analysis behind it. So that's the real cost. It's not only the hours, it's making decisions with less information than you could have had.

So what does AI actually change here? Workday's answer to that is adaptive decision intelligence. It's the most important thing and it's built right into adaptive, not a separate tool, not a bolt-on right in your native system. It really does three things here. It's going to help you analyze, so it'll explain why a variance happened and it's going to write that up in plain language. It helps you plan with predictive forecasts and a quick what if scenario. And what it does is it helps you act because once a scenario is actually proved, it goes right into the governed plan with the assumptions, the data sources and those approvals attached. So for finance and audit teams, that last piece is a really big deal because the work isn't happening in a separate spreadsheet or from someone's laptop anymore. It's happening right where you process your data.

So this is the second idea. A lot of people were hearing AI and planning and they think it means we're just doing our monthly cycle quicker, right? But the real shift is that you're not waiting on the cycle at all. So let's take a look at the left hand of our screen. This is how most of our teams are planning today. Questions are taking days. We have variances that get explained after close. We have forecasts that are getting rebuilt monthly or quarterly. But on the right, you can ask a question and get an answer in minutes. We have variances that are flagged and summarized for you and forecasts are generated from your actuals with confidence metrics and the scenarios get built inside the plan itself. So your plan is always current in between cycles and that's really what leadership is looking for.

So let's take a look at what's actually under the hood, decision intelligence. It's the conversational piece. This is where you're asking questions about your plans, your actuals, your operational data. Planning agent really speeds up that exploration and the analysis work. And Workday says it can reduce data exploration and analysis by about 30%. Next, you'll see our variance analysis. It's going to find what moved and really lets you drill all the way down into transactions and it's going to traffic that commentary for you. Predictive forecasting using machine learning really gives you confidence metrics so you know how much to lean into it. And scenario modeling lets you test all of your what ifs just by asking. The last one is pretty new. If your team is using different tools, you have the ability to use those within your adaptive data as well while still following your access rules.

So one question we get on almost every call when we're talking about AI, who's actually in control here? And the answer should always be your team. You ask a question wherever you're already working. The AI should pull together an answer from your data. Your team is reviewing it and since everything traces back to your data, you can check the logic. And where do you do with that? That's up to you. You decide what to do. AI is going to give you the faster answer, but the decision is still yours. So this is the third idea and honestly, it's my favorite. Let's think about where your analysts are spending time today. They're pulling exports, they're reconciling versions. They have to explain variances line by line. Now let's think about where you'd actually rather be spending all that time, talking about things like pricing decisions, recommendations for leadership, where should we be investing?

What I find interesting is there's a Forrester study that Workday commissioned earlier this year. The biggest benefit actually wasn't time saved. It was $2.7 million in value that came from better decisions, things like pricing, retiring any kind of redundant software and moving the investment once the team actually had time to do the analysis. So here are the numbers. One thing to be upfront, so these are results for Adaptive as a whole. Decision intelligence is pretty new, but this same Forrester study found a 242% return over three years and payback in under six months. It's based on a composite organization of about a billion dollars in revenue and 5,000 employees, and it's built from five customer interviews. So while your results will vary, it gives you a sense of scale. That 30% is the planning agent number I mentioned a minute ago. Workday likes to say that it can help reduce data exploration and analysis by that much.

For a real example, the TX group, they use Adaptive's predictive forecasting and they cut their monthly rolling forecast from two to three days down to about four hours. Their head of cloud ERP said that in his 18 years of corporate finance, he had never seen a tool so helpful. And Workday has been named a leader in Gartner's Magic Quadrant for financial planning software four years in a row. So we're hearing all of this and you're wondering where to start. We usually like to suggest three steps. First, what we'd call a current state instance review. We take a look at your adaptive setup and figure out which of these AI features you're ready to turn on and where they'd help the most. Second, a demo built around your planning questions, not just a generic run-of-mill demo. And third, we help you turn things on, train your team, really set a baseline so you can actually measure that time that you're saving.

As for us, we're a certified Workday partner. Our team comes from a finance background of CPAs and we've built knowledge transfer into everything that we do so your team feels confident in what they own at the end.

So that's a lot of me talking about it, but I know what you're here for. You want to see it. So as you watch, keep those three ideas in the back of your mind, how fast a question turns into an answer, how the plan stays current and where your team stays in control. Feel free to keep those questions coming in the Q&A panel and we'll get to them as soon as we can. Chris, I'll hand it over to you.

Chris Campbell:
All right. Thanks, Tyler. I'm going to go ahead and share my screen. Let me just get this organized. All right. So we should all see my screen as I'm in the Workday decision intelligence basically area of adaptive. So first thing you'll see here, we have this text box at the top that's kind of giving us some suggested prompts. So really getting started here is as easy as just type in your question here and we'll kind of start off a conversation. One thing around how this is organized. So essentially each conversation that we have gets saved as a project. So we can always refer back to a specific sequence of prompts and along the way, adaptive decision intelligence will also create these artifacts. And so these are essentially dashboards that kind of pull up some analysis for the information that we're looking to learn more about.

So for my first example here, let's put our FP&A hats on and say we want to really explore our travel and entertainment expense. So first thing I did was kind of started a new prompt and all I asked for was, I said, please create an artifact, again, that's a dashboard, analyzing the GL 6450, other travel and entertainment for the working budget version for FY 2027. So that was my prompt. And then from there, decision intelligence was able to create this nice dashboard for me. So first thing I want to show here, so we have a monthly trend, this is kind of the overall organization spending trend and I can always kind of mouse over here to see the detail of the dollars over month. I get a sense of the seasonality. Scrolling down, it also prepared for us a department break.

And further down, it also gave us a heat map. So this is the spend by different departments and I'm even able right from here to drill into the different departments so I can see the full detail. So I'm looking at our sales Birmingham department right now. So I think it was a good kind of initial analysis. Now say maybe I'm specifically responsible for our Sales North America department and then I want to learn more about that one and dig in more. So from here, I kind of went on to my next prompt as part of the same conversation. I said, "Hey, let's focus on Sales North and put together a monthly comparison between the FY27 budget and FY 2026 actuals because I want to see how our budget is doing compared to actuals." So it gave me a new dashboard here. So the first thing we notice here, we have our budget to actuals here and we can kind of see the quarterly trends.

It also gave us a chart showing the variance on the right hand side here. If I scroll up a bit, it also gives us the overall fiscal year total values with the full year variance as well. So some good starter information as far as we want to look into this department. Another thing that will happen is as we're doing our prompts, Adaptive Decision Intelligence might see some things about our data that it wants to make sure we're aware of. So in this case, it's letting me know right here that this particular department is only planned with quarterly grain. So what that means is we essentially have a value for each of the quarters that was then spread evenly among months. So we didn't really have a ton of seasonality to our data. So as we kind of prompt things and have our different questions, Adaptive Decision Intelligence is going to also help us with our analysis and tell us things that it's seeing in the data that maybe don't stand out to us initially.

So I think I actually want to take this into account and I want to update our forecast with a bit more seasonality. So now we've done some analysis, now let's do some reforecasting. And say I also have some additional information, so I'm going to go up to my next prompt. And so this is the Sales North department. Oops, I went down too far. And let's say we know they have a lot of opportunities coming up this year. We kind of expect them to travel more and we really need to bump up our budgeted expense. So what I prompted here is I said, "Let's create a new scenario called Sales North, 50% T&E increase and increased travel expense in FY 2027 by 50%, keeping with the seasonality of FY 2026 actual so we actually have some more realistic monthly fluctuations." So one thing you'll note, because when we're kind of prompting a change in this, we're actually going to create some inputs in the system and really create an updated forecast.

So when I do this, I'll basically get a confirmation of the change I want to make. So the first thing we'll see here is it saying, okay, we're going to create a new scenario. So here's the name of the scenario. This is the budget version that we're going to be affecting or creating the scenario off of rather. And then here's sort of the detail of how the decision intelligence has interpreted the prompt. So in this case, it actually inferred some additional detail. Now if I scroll up a little bit, I was a little bit vague in my prompt. I didn't even say which particular GL I wanted to affect. I just said increased travel expense. But based on the context of the conversation, it was actually able to pull up. I was like, oh, well, you've been looking at GL 6450 other T&E, so probably that's the one that Chris wants to update.

So that's exactly what I wanted, so that was perfect. And then the other thing that was key here, so I didn't necessarily say did I want a 50% increase based on actuals? Did I want it based on the budget? But it's kind of being very clear here and saying, okay, it's a 50% increase above the working budget. So that's the base number that we're going to be increasing. So this is exactly what I wanted from the prompt. So I essentially accepted this and then it'll basically give us some detail about the changes that it made. And from here, obviously we could spin up a new dashboard from within Decision Intelligence, but I also want to show you how this integrates with the rest of your tools within Adaptive. So reports, dashboards, all of that. So for this purpose, I'm actually going to jump over to a dashboard I have set up here.

So right here you'll see a report that's basically comparing budget to actuals. Right now, I want you to focus on this far right section. So this is showing our original working budget and then our scenario in this yellow column here. And kind of right here we can see, okay, this is our other T&D and it's showing, okay, we had a 50% increase in the expense for the year. So looking at it at a big picture level, it did exactly the increase that we wanted, so that's great. I think the next thing I want to look at is did it factor in the seasonality? So for this purpose, I'm actually going to jump over to this expenses tab here so we can see the actual context in the sheet. The nice thing is we're in a scenario, so any changes that the decision intelligence made to the system, they're going to be highlighted in yellow here so we can clearly see what was the original budget numbers versus what are the actual changes that decision intelligence made.

And from here, we can kind of see that the seasonality has some realistic fluctuations, so that's great. And say I wanted to change any of these numbers, I'm free to type over, put in my own inputs if there's some things that I wanted to adjust at the end. And then from here, if I like the changes, I can go ahead and publish a scenario, make it live in the budget, everyone can see it and write this task as done. So that was just kind of one quick example. This is kind of a fairly simple analysis and update that we did where we're just looking at accounts and department or level, but you can also do these types of prompts with any of your dimensional data and your cube sheets. It has an understanding. It's built around all of these systems that kind of make up our components within Adaptive.

One other thing that I kind of noticed on the last report, so I'm going to jump back to this variance analysis, is over here in our actuals, we have this. So this is our March 2027 actuals, we're living in the future here. We've got this huge bump in our other T&E. And I didn't really ask anything about our 2027 actuals yet. So that didn't really come up in my prompting this far, but I kind of want to go back to decision intelligence and ask why might that be? See if it can come up with some hypotheses for us. So I did exactly that in our prompt over here.

Scroll down. So here I just said, "Something's odd with March 2027 actuals. Can you help explain why there's such a large variance in the same GL?" And for this purpose, it created another dashboard for us showing some detail around the variance. It's hugely over what the budget was and it actually ranks some different hypotheses for why we might see this sort of anomaly in our actuals activity. So the most likely one, it thinks it might be a data entry error where someone missed a decimal point. So that's totally possible. Some other possible explanations is maybe someone made an input to the wrong GL code, they posted to the wrong period, or maybe it's a legitimate one time event that we just didn't have planned in our budget, but it kind of gave a low likelihood of that one. So from here, we kind of have some good possibilities that we can explore.

So I'm actually going to go myself back to the dashboard. And then from here, we can basically, I can click here, get to the sell explorer, drill into transactions, and right here is the transaction detail. No, I see right here. Oh, so we had 2,600 set aside for company lunch, but someone input that as 2.6 million. So this is definitely a data entry error where we should probably go back to accounting, get that corrected. And yeah, I think we kind of identified that it was indeed a decimal error at the point of entry. So again, just another kind of use case. So we can use the decision intelligence for analysis. We can kind of apply some changes to our model and even come up with some explanations for the analysis. It's not just the trend, but it's like why might the trend be happening? So yeah, that's the first example.

Now I want to show you another example. So this will be part of a new conversation where I want to look at our workforce planning model. So this will be an example where we're showing you basically how decision intelligence works with your existing modeling within adaptive, because all these tools, they work together to kind of give you more accurate forecasts. So in this case, let me go back to my dashboard here. So we have a workforce planning model here. So basically we'll have our roster and then we'll have our pay rates and start dates, end dates. And in this case, what I ended up prompting is... So our pretend scenario here is say we're having trouble hiring for some of our new hire positions for Sales North America, and we actually want to bump up the starting pay rates because we want to make it more attractive to hire into those positions.

So in this case, what my prompt ended up being was for the working budget version, create a scenario where we increase starting pay rates by 30% for new hires starting in March 2027 and later, so after that start date within the level of Sales North. And just like before, I'll get the confirmation of what I'm asking. In this case, I actually wasn't even very specific about what I wanted to call the scenario even, but it came up with its own name, which I like, which is Sales North new hire pay rate plus 30%. I can understand that using the working budget and then a confirmation of the change that I wanted to make. So now we can either look here and it'll tell us exactly what positions were updated, what the pay rate change was before and after. At a high level, I can see this is a 30% change.

And if we go back to the dashboard, I showed this earlier, but we do have the highlights here of the pay rates in yellow that it actually changed. So it was actually smart enough to know, okay, let's be very targeted in how we're making the changes. So it's looking for new hires with a start date after, I said March 1st, and it only picked up on those two positions. This is exactly what I wanted. This is a pretty simple example of the change, but you can imagine you could have a much more complex prompt with a bigger data set and decision intelligence can be helping you make those inputs.

So you can basically focus on analysis because imagine we have a roster of a thousand people and we need to make all these little massaging of different changes we want to make to the model that are very targeted to specific positions. Well, decision intelligence has an understanding of all of the dimensionality and different fields in your system and is basically able to trace through that. So you don't have to go through and do that manually. So just to kind of illustrate why it's also important that we're integrating this into your normal models. So up at the top, it's not a super big example of the change, but you can kind of see the blue charts here are our scenario generated by decision intelligence. And then the line here is our benchmark budget number. So it's a little bit of an increase, but it's not just salaries that we wanted to impact, it's anything that's reliant on salaries as well.

So next I'm going to bring up a report that kind of shows us a little more detail of what's happening. So here on the far right again, we're comparing our budget to our scenario. And we'll see here how in yellow is all the different places that were basically all the different GL accounts that were impacted by our change. So in this case, our overall salaries for Sales North America went up by 5%, but it also had a downstream impact on some of these other accounts. So our retirement, protected retirement expense went up, payroll taxes of course went up. We have some recruiting fees and because this is very targeted to new hires, that went up quite substantially. And just as important, we also didn't change any numbers for anything that's not driven on salary. Health benefits we're probably driving off of headcount and that one actually doesn't change.

So that's all to say, Adaptive Decision Intelligence was able to make a very targeted change into your model. And then from there, your model works as they always do and you kind of have all of the downstream impacts of the GL that you would expect in any financial model.

So I think this is one last thing I'll talk about and then I'll pass back to Tyler. So another AI tool that exists within Adaptive is the predictive forecaster. So in this example, in green here, we have our actuals period and we have our baseline bottoms up plan for units. And this is basically for a revenue model driven off of units. And where we might use this, so essentially you can generate algorithmic plans, so using AI to basically create benchmarks for your bottoms up plan. And what this can help you do is identify where there's basically odd variances where your bottoms up plan doesn't really match with the actual trend because whenever that happens, you want to be able to explain why that may be. So built into adaptive, we're able to generate these algorithmic plans and can basically stage that against your bottoms up plan so you can kind of see what the variances are.

We can kind of correlate this with what are called lever sheets. So you might have what revenue is generally going to be correlated with your consumer price index. So again, that's all to say. We have a lot of algorithms that are kind of built into the system that were created at Ivy League institutions that you have. It's already kind of in your adaptive today. So something that's in there that you could play around with now. And all you really need is two years of actual data to get started with this. From here, I'll pass it back to Tyler.

_{Transcribed by Rev.comAI}