2026 FP&A trends: How AI is testing the foundations of Finance

Good morning, good afternoon, good evening, ladies and gentlemen. Let me welcome you to our FP&A Trends webinar. My name is Olga Rudakova. I'm happy to be a facilitator for the event. I'm a FP&A professional myself and FP&A Trends ambassador.

Happy to announce that today's topic for this webinar is 2026 FP&A Trends Survey where AI meets FP&A reality. We have 335 registrations for this event coming from 52 different countries. I'm very happy to see many people joining and the number growing.

What we can expect today during the session? First of all, we will introduce 2026 FP&A Trends Survey and we will dive right into key findings and the AI readiness gap coming out from the survey. Then we will move on to the technology in practice and speak about practical use cases where AI meets FP&A reality.

We will have conclusions, recommendations and Q&A session. It's time to me to introduce our speakers. So Michael Prass, please join me on camera and I will start those introductions one by one. Let me start with you, Michael.

Michael Coveney is analytics thought leader and author. He's head of research at FP&A Trends Group. So that's Michael who stood behind this survey in its interpretation. Michael has more than 40 years of experience designing and implementing software solutions that improve efficiency and effectiveness of planning, budgeting and forecasting.

Michael, thank you so much for joining us today. It's a pleasure to be with you, Holger. Our second speaker today will be Pras Chatterjee. Pras is Global Product Marketing Director at OneStream Software.

Pras is expert with over 20 years of experience and with strong finance and information systems knowledge gained in various Fortune 500 organizations. Pras, thank you so much for joining us tonight. Thanks so much for having me here once again.

Let me now introduce our technology sponsor OneStream and we are very proud to have them as a sponsor of these years survey. OneStream, take finance further. That's where OneStream helps finance leaders to move beyond the just reporting and actively steer their business by uniting financial and operational data with AI-powered planning and forecasting.

I will introduce FP&A Trends Group, even though I'm pretty sure that you know us quite well already. Just some key numbers for you. First of all, we are presenting today FP&A Trends Survey and we are proud to say that by now, since 2017, we had 3,361 participants.

We are at one of our FP&A Trends webinars and we had 193 of such events since March 2018. And I would like you to visit our FP&A Trends website where we get more than 500,000 visitors per year. All right, what we can expect tonight? First of all, that's one hour event.

We will take part in two interactive polls on the subject. So be ready to be active. You can ask questions and you can actually start right now via the chat box. Anything you are interested in and you can ask Michael Cross or just general questions for both of them.

Presentation is already available in the handout section. And you will receive the recording and the presentation after this meeting. At the very end, we will ask you to fill a very brief survey, literally a minute or so, and it will help us improve and have more relevant topics for further webinars.

I'm very impatient to pass the word to Michael and to dive into survey findings. So I will only share a couple of basic information about survey participants and demographics. First of all, it's been 10 years of survey already.

Wow. And we had 475 participants this year and altogether it's over 3000. So 3361 demographics. Well, you see that all the parts of the world are represented in our survey with majority of respondents coming from Europe, North America, but from all over the world as well.

In terms of job titles, we have a very senior participants with job titles, such as a director or senior director of finance or FP&A. We have CEOs, CFO, CIOs, CTOs, and so on. We have vice presidents and senior vice presidents of finance and FP&A.

Okay. In terms of turnover, we have different company sizes represented. So what do all these diverse people have to share with us? Who else can tell us better than Michael Coveney? Michael will cover 2026 FP&A survey key findings and the AI readiness gap.

So, Michael, let me pass this imaginative microphone to you. Michael, just to be sure that you are here because your image is still coming up. Do you hear us? I can hear you fine. Thank you. Perfectly.

So I'll pass this control to you and I'm pretty sure that image will come up very soon. Okay. Thank you. So what I want to do, I want to go through the survey and our findings that we found it. Now, the survey covers a number of topics such as the business area, the business environment, the planning approach that organizations take, what their view of the future is, the technology platforms they use.

So it's a lot of information that we collect. And what we do, we combine, for example, this year's survey results with previous years. And then we start to analyze it to find out what things have changed, what are the issues organizations face and what is the future.

Now, obviously, we don't have time to go through all the data in the survey. So what I'm going to do today is I'm going to focus on people's use of AI and to see whether or not organizations are ready for using AI into the future.

Now, what does it say about their maturity? Now, the first thing is to say is that FP&A exists to support better business decisions. But the survey shows that most teams are not yet operating at that level.

For example, only 22 percent describe themselves as being optimized or performing well, while 78 percent, that's three quarters, remain constrained by manual work or are struggling to cope. And despite FP&A's role in supporting business decisions, we find that only 41 percent of respondents describe their function as being fully endorsed or acting as trusted business partner.

Even though business partnering continues to be the most sought after skill that FP&A teams are looking for. And the issue is this, that as AI becomes more widespread, the ability to be able to interpret insights, to challenge assumptions and to influence decisions becomes even more critical for FP&A.

So when we look at where FP&A spends its time, we find that not much has changed. In fact, the recent trends indicates that it's sort of getting worse. In 2026, we find that 47 percent of time was still spent on data collection and validation, while only 32 percent was spent on insight generation driving action.

And this is at a time when the planning horizon of how far into the future teams can see. It turns out that only 44 percent can only see three months out with any high degree of confidence. So the time available to interpret results and challenge assumptions is so much more reduced.

And that's really important. So why is FP&A in this position? What holds them back? Well, when we look at the survey, we found that 96 percent of the planning and forecasting challenges for this year stem from just three areas.

And they're the ones on the screen. It's poor data quality. It's manual and inefficient processes. And it's systems that are outdated or have disconnected workflows. And while technology remains an issue, the findings suggest that data and process weaknesses are the biggest barriers to improving FP&A performance.

So let's take a look at each of these areas in a bit more detail. So first of all, let's have a look at data quality. Now, data quality is improving at the lower end. You can see on the screen on the left hand side there where organizations rate the quality of their data.

So we're seeing that low and poor data quality has reduced over the time, you know, from 2021 where it's 40 percent down to just 22 percent today. However, if we look at those organizations that say they have high quality data, we find that that is remaining stagnant.

In some ways, it's actually getting worse. And allied to that is actually data timeliness. You know, how quickly or how what is the availability of data? We find that only 11 percent of organizations have real time access to data.

And most organizations, I think it's well over 40 percent, still rely on just monthly data. And it seems that while organizations, they're increasingly say they use data for making decisions, for example, 81 percent say they use data for all or most decisions.

Yet most organizations still struggling with data quality. It's interesting that when we asked those organizations who was responsible for data, we find that only 14 percent have a dedicated data management function and most rely on FP&A to manage data.

And so that probably accounts for where much of the time is actually spent. And the thing is, without strong data foundation, FP&A will find it very difficult to realize the full potential of AI. Now, when data quality is good, it has a major impact on FP&A effectiveness.

So this is some of the analyses that we did. We looked at those organizations who report having high quality data. And what we find is that from a performance point of view, how they rate themselves, 53 percent of those with high quality data rate themselves as being optimized and ready to perform compared with just 2 percent who have low quality data.

We also find they spend more time on value added activities. Now, 53 percent of time is spent on insight generation driving decisions compared to just 13 percent with low quality data. We find not surprising, much greater use is made of data in decision making and they produce much better quality forecasts.

So that's data quality. Let's move on to the next topic. Let's have a look at the processes themselves. Let me just go back. Now, with processes, we look at really three areas in this and for this today's presentation.

So, first of all, when it comes to planning, only 10 percent of organizations fully integrate strategic financial and operational planning. It's as though that what happens locally or from a financial view has little bearing on the strategy of the organization.

And when it comes to P&L, balance sheet and cash flow, fewer than half organizations actually integrate them, which means that when they do produce these statements, they're making decisions that don't take into full account the full financial impact.

On the right hand side of the screen, we can see there we found that most organizations, we found that driver based modeling, you know, where activities are linked to corporate objectives is becoming more common.

In fact, 19 percent of organizations, that's a little green bar on the right hand side there. 19 percent of organizations now use mature models that automatically identify and apply drivers. And that's up to 10 percent from what it was three years ago.

The however, though, is in that red area. 40 percent still rely on basic or non-driver based approaches, which limits their ability to link plans and forecast directly to business performance. Now, when organizations link their plans, when they implement and use driver based models, we find that their performance improves.

So, again, on the left hand side, you can see this when you integrate strategic planning, financial and operational planning. We find those organizations have a stronger team performance. More time is spent on value added activities, greater use of data and decision making.

And again, when you get these slides, you'll be able to look at this data. The same is also true for those who use driver based models. And this suggests that driver based modeling, along with integrated planning, are key enablers of FP&A maturity.

And they form a critical foundation for realizing the benefits of AI and automation. Let's look at the systems, the third area. If we look at the systems, we find spreadsheets planning still remains the most common approach, though it is coming down gradually over time.

And we find in second position, or the most commonly used application, are actually the modern cloud-based planning platforms, which are used by 19 percent, followed by the older generation planning consolidation systems.

And what is surprising on the right-hand side of that screen is that given the increasing influence of AI, organizations are still relying on outdated systems, with nearly half of organizations still not having upgraded their systems in the past three years.

And that's going to limit their ability to support more integrated, more automated and more AI-enabled ways of working. Now, on the use of AI, we've been tracking how machine learning to enhance forecasting, we've been tracking that for quite a few years now.

And what we've found is that you'll notice that top green line at the left-hand side of the charts, you know, we reached a plateau in 2021 of 11 percent of organizations using machine learning for forecasting, and then it dropped off.

Whether that's due to, you know, hard to, like, the ability to be able to use those kind of applications. However, what we've just seen in the past year, that has now bounced back. And we're now seeing more organizations are using machine learning.

And for the very first time on the right-hand side, for the very first time, we actually asked about organizations' use of generative AI, which seems to have exploded on the scene, and certainly on the business scene, and we're finding that 21 percent of organizations, this is FP&A departments, are now using it.

And the main application is in the areas of communication, automation, and decision support. And what this tells us is that although adoption is increasing, many organizations are still in the early stages of AI adoption.

So when we look at the impact of AI, what we find, again, is unsurprising, that AI is having a big impact on FP&A. Those organizations that are using it report better team performance. Again, you can see from the green bars there.

This is comparing those who are using AI compared with those who aren't. We find they spend more time on value-added activities. There's greater use of data and decision-making, and a higher quality forecast.

And while AI alone does not guarantee success, its adoption is closely associated with high levels of FP&A maturity and effectiveness. Again, for the first time, we've started tracking AI adoption, and are organizations ready to use this technology? So when it comes to adopting AI more widely, the readiness of organizations is very mixed.

We find that 19% of organizations say they have formal governments in place, and they're ready to scale its use. But 59% still rely on informal guidance, or have no policies at all. Now, what's interesting is that amongst the high-performing teams, those that are performing well, they are more likely to be ready for AI, but it's still only 32% of them compared to 11% of other teams.

This highlights the link between overall FP&A maturity and the ability to adopt AI successfully. Now, when we look to the future, this was a bit of a surprise. 98% of organizations, that's pretty much all of them, expect AI will impact their role over the next few years.

For many of them, they expect it's going to give them capacity for value-added work, and others are saying, yes, it will do that, but also it's going to change our role. You know, we can't be the same people that we were in the past.

We are going to have to adapt if we're going to make use of AI. And the other thing that was quite a surprise is that, you know, for, I guess, the past 9, 12 months, there's been a lot of hype around AI agents.

These are, it's like, analytical tools that can actually make decisions. You give it an objective, and it will work towards it without any human intervention other than the human setting what the boundaries of what it can change.

And what's interesting is that 15% of organizations expect to be using AI agents within the next six months, and 51% expect it to be using at least sometime in the future, which is a real big surprise.

So, coming to the end of this, that was a bit of a dive in there. What should FP&A leaders do next? Well, one thing we can actually learn from the high-performing organizations. So, if we have a look at what do they do that's different from the rest? Well, and in some ways, this can be a checklist for yourself.

So, first thing is, is that with the high-performing FP&A departments, they focus on data quality, on data timeliness, and on ownership. So, this is one big area, and I'm sure Chris is going to have something to say about this later on.

So, data quality is really key, because without high-quality data, it doesn't matter how good your AI routines are, it's not going to help you that much. The second thing is to make sure you have integrated planning processes.

So, when you think about it, strategic planning is absolutely linked to financial planning, is absolutely linked to operational planning. The three really is just one planning process, but it just has different aspects.

And what we're finding is that high-performing organizations really focus on integrating those three processes. Third is to really strengthen the use of driver-based models. This will enable you to really automate and speed up forecasting and budgeting process by having, if like the system, look at what is the cause and effect of performance.

Third or fourth thing, reduce manual work. I mean, AI is there to reduce a lot of stuff. For example, we're finding a lot of organizations are using AI now to help improve the quality of data, to make sure it's in the right format and so on.

Fifth thing, investing in AI readiness. Make sure you start to have formal processes. I know for some organizations, it's not legal for them to use AI. So, make sure that when you use AI, it's used in a way that not only supports the business, but in a way that is going to really help the organization going forward.

And finally, strengthen business partnering. The whole point of these tools is to help FP&A become trusted advisors, to help organizations make the right decisions, to challenge preconceived views they may have of the business.

So, there we are. A bit of a nutshell of that, but hope you enjoyed it. There's a lot more in the report, and at the end, you'll get details of how you can download the report. Back over to you, Olga.

Thank you. Thank you so much, Michael. I really enjoyed how you eventually got to practical interpretation, practical implication of survey for business leaders. What should we do next to actually act upon the survey results? At the very end of the session, I'll share how you can download the full survey and read, obviously, all the rest that could not possibly fit into this session.

If you have any questions to Michael, don't hesitate to ask them. By the way, we are already getting some quite interesting questions for you, Michael, so get ready for them. Meanwhile, let's get to polling questions, so a question in the opposite direction.

I would like to ask our audience, where do you see the strongest potential value from AI in FP&A? So let me launch this poll, and I will share results when we have them, but meanwhile, let's see what options do we have.

Where do you see the strongest potential value from AI in FP&A? Better forecasts, simulation of scenarios, automated reporting and commentary, better insight of drivers, and none we don't plan to use AI.

Let's see how our votes will split, and very, very curious here, especially with option number five, I have to admit. So what can we get from AI in FP&A? Better forecasts, simulation of scenarios, automated reporting and commentary, better insight of drivers, none we do not plan to use AI.

The majority of you have voted already. So let's end this poll and share the results with everyone. While we don't have a definite strongest leader, but 29% goes to simulation of scenarios, and right after, 27% go to better insight of drivers, then 19%, still quite strong result, goes to better forecasts, and 21% even before that, automated reporting and commentary, and the last two is only 4%.

None we don't plan to use AI. Michael, what do you think of the results? Are you surprised or not? That's a very interesting result, actually, Olga, because most people that I speak to use it for automated reporting and commentary.

You know, that's a very easy, safe route into AI. But the real value, I believe, comes in identifying drivers and then using that information to scenarios. So I'm really quite pleased with that result.

And I think if people could have answered all four equally, they probably would have done that as well. So I think that's an interesting result. Yeah, yeah. Same here. I think that's a great answer, especially that not only generative AI, but machine learning behind it is being used.

So that's great. So let's move on to the discussion. So I would like to ask you, Pras, to join us here as well on camera, and we will move on to the question I really wanted to ask both of you. So the question is, what stands out in this year's survey compared with previous years? And you know what? I'll start with you, Pras.

So what does stand out for you this year? Thanks so much, Olga. So I think when I look at the data, three things actually stand out. So the first thing that stands out to me is data depends, I think as Michael pointed out, is at an all-time high.

I think Michael mentioned that 81% are now using data in most decisions, up almost like 22 points. But the interesting thing is that the data quality didn't move with it. Only 19% call their data advanced.

And, you know, in past surveys, the dependence and quality rolls together. So this year, it's almost like we're leaning harder on data more than ever, but the foundation isn't keeping pace, which is pretty new.

So, you know, it might be that there's just an explosion of data causing people to use the data, but they're not able to trust the data or rely on the data fully. So that's clearly a challenge. I think the other part is that often it seems like integrated planning was stuck at 10% for three straight years.

And it seems like the three-way financial integration is actually declining as well. You know, in previous years, you could tell a story of, you know, a slow forward motion on this, but this year, it doesn't seem like it's holding.

It's actually become like a regression, which is interesting because business partnering is one of our most interesting topics. And I think, you know, one of the last things is that, you know, with AI arriving, you know, some of these weaknesses are, you know, coming full flight.

Like, you know, we're basically seeing that organizations are seeing a weak foundation when it comes to the data. So that's something that, you know, is clearly evident in this, but I think there is a way forward.

Right, right. Thank you for such an elaborate answer. I'm wondering if there are anything left for you, Michael. So let's try. What standouts for you this year? And that hasn't been mentioned yet. Yeah, I agree with everything Pras has just said.

I guess there's just one other thing sort of linked to it. And that's actually data timeliness. I mean, what I was really surprised at is that I've got the numbers here. It wasn't in the presentation, but 42% of organizations still rely on monthly data.

And that is worse than what it was two years ago when only 37% had. And if you look at those organizations that are using real time or daily data, today it's 33% of organizations compared two years ago with 40%.

So it appears that data timeliness is getting worse. But of course, that could be just because of the volume of data and the different types of data. But it all adds to this. Data is an issue. I think that has to be solved at some point.

Right, right. Thank you for all these interesting observations. And we will move to the second part of today's session. And Pras will be speaking about technology in practice where AI meets FP&A reality.

Pras, let me pass this imaginative microphone over to you so the stage is yours. Thanks so much, Olga. So Michael's giving you all a really clear picture and a full picture of the survey. So I'm not going to recap on the data, but I want to do something different with my time and really talk to all of you as a practitioner, a CPA myself, about three things.

The problems the survey surfaces, how you fix them, and what it actually looks like when a real finance team actually does it. And I'm here to tell you up front the whole story, it fits on the title of this slide.

You know, we're betting more on a weaker foundation, and finance has never really leaned more hard on planning, on technology, and yet the foundations underneath in a lot of ways are getting shakier, not stronger.

So that gap is something in this area of survey that's, you know, quite important. And it's about to matter a lot more because of AI. So let's talk about what I really mean by this. So let's start with data because everything FP&A ultimately runs on it.

And here's a contradiction at the heart of this year's results. So data dependence, as I pointed out before, is at an absolute all-time high. 81% of teams now say they use data in almost nearly all of their decisions.

It's the highest that's ever been recorded in this survey, and I've been watching the survey in part spit in over 10 years. And finance has won the argument that decisions should be data-driven, which is really good.

But then look at the foundation underneath that confidence. Only 19%, 19% rate their data quality as advanced. And here's a number that should stop us. Nearly half of FP&A's time, 47%, is still consumed with just collecting and validating data before a single insight gets generated.

So think about what that means. Your most analytical, most expensive people are spending half their week on plumbing, basically. That's not an inconvenience. It's a structural tax on the entire function.

We've asked finance to be a lot more strategic, more forward-looking, more of a business partner. And then we've buried half of our capacity in doing custodial work. That's a weak foundation. And you don't fix it with another report or another tool bolted on top.

But again, just hold that thought because it actually gets a little more urgent as we see it on the next slide. So here's why this foundation suddenly matters more than it ever has. Everyone in this room is being asked about AI.

Everyone's under pressure to deploy agents or find a use case for agents to automate, to move faster. And the instinct often is to treat AI as the fix. That's going to rescue us all from this mess. But that's not what this survey shows.

Look at the numbers. Teams using AI outperform on every measure. Forecast quality, 64% versus 35%. Performance, 32% versus 18%. Data-driven decisions, 88% versus 78%. AI users and AI-driven organizations are winning across the board.

But let's be careful about why. These teams aren't winning because they just sprinkled AI on top of a broken process. They're winning because they already had trusted data and connected planning and AI compounded the advantage that was already there.

And the survey confirms it from another direction. The number one barrier to deploying AI agents cited by the survey is data quality, 35%. The exact crack we saw on the last slide is now the thing blocking the future.

So here's a reframe I want to leave you all here with. First, in this first act, AI is not the transformation of FP&A. AI is a test of whether your foundation was ever ready. Where the foundation is strong, AI makes you dramatically better.

Where it's weak, well, AI doesn't hide the cracks, it exposes them, it produces wrong answers faster and at scale, and the gap between the ready and the unready just doesn't close with AI. It actually starts to widen.

So, how do you fix this? Here's the good news and the most encouraging finding in this survey. Nobody here needs convincing. Finance leaders have actually diagnosed themselves correctly. When we ask about the top transformation priorities for 2026, this is what came back.

Create an aligned, unified data source. 20%. That was tied for first. Adopt new technology. Also 20%. Upgrade planning and forecasting systems. 18%. Strengthen business partnering. 17%. So, if you look at the shape of that list, it's not buy more AI.

It's not chasing this shiny object. Leaders are prioritizing the foundations, the data, the platform, the human partnering that make the technology and AI actually deliver something of value. The markets effectively figured out what matters.

The question was never what to fix, it's how. And specifically, whether you fix it by adding yet another point solution or by finally unifying the foundation and fixing that entire foundation itself. So, this is the heart of what I want to say today.

So, let me just slow it down a little bit. You can't fix a foundation problem by adding another tool on top of the problem. Every bolt-on you add is one more system, one more point of integration, one more place the data has to move and get rekeyed and revalidated, effectively more plumbing.

You're not fixing the foundation, you're pouring more weight onto the crack. The fix effectively is to rebuild the foundation itself. And there is a sequence to it, three steps, and the order matters enormously.

Step one, unify your data with one platform, one area, one version of this truth. The data doesn't move. It doesn't get copied between systems. It doesn't need revalidating every cycle. Remember that 47% tax? This is what makes it disappear.

This is a precondition for everything else. And don't get, you know, swindled and, you know, by data fabrics and this and that. Finance needs to own their data. Work with IT, but finance needs to own the data, trust the data, and validate the data.

The second step, connect the planning, all of your financial and operational plans and data in a single model, not three, four, five, six, seven tools attached together with interfaces, whether it's your FP&A tool, a workforce planning tool, a supply chain planning tool, a marketing planning tool, a sales planning tool, and, you know, and integration by architecture and whatnot.

And here's the thing about agents as well. Agents can only reason across a business that's connected. If your plan lives in five disconnected places where the intelligence isn't synonymous, there's nothing coherent for AI to reason over.

You hear a lot about, you know, business semantic, a semantic layer where AI is supposed, you know, you're supposed to have this layer that interprets all your intelligence. Well, if your data means different things, your actuals live in one place, your plans live somewhere else, and multiple plans of different operations live somewhere else, and you're really going to have agents that some of these mega vendors are proposing go in and really have the intelligence in terms of how you as finance interpret it, not the answer.

So, step three, you know, once you've got a foundation all set up, put agents and AI on it. Agents embedded directly into the platform where you've harmonized all your data in the context of the way that you finance consume it.

It's already, you know, put the data in it, work on the data that they can trust, you know, inside the workforce and workflows finance already uses. Not AI bolted onto chaos, but AI built on a foundation that was engineered to carry it.

So, notice a sequence. The AI step is deliberately last, and it's the dark card because it's the payoff, not the starting point. Do it in this order, and AI compounds your advantage. If you do it backwards, if you lead with agents and hope the foundation catches up, all you've done is automate your mess faster.

Foundation first, AI and agents second. That's the whole play. And that's a theory. So, let me actually show you someone who actually did it. So, Cox Enterprises, one of the largest privately held U.S.

companies, 23 billion in revenue, 50,000 plus employees across 15 countries. So, Cox's finance organization had trusted systems and sound data, but they were held back by manual data friction, ad hoc reports built by hand, answers buried in documentation, and contracts reviewed line by line.

Cox actually joined one stream, our private preview, with success defined right up front. Accuracy, efficiency, and trust by design. So, with our agents, three of them, we have our finance analyst agent, our search analyst agent, a search agent, and our deep analysis agent.

Our solution, brought intelligence directly from Cox's own financial data and documents through these three purpose-built agents that are available for all of our customers. The finance agent, it actually turned natural language into governed one stream reports.

Before, there was 10 minutes per ad hoc report. Afterwards, one minute, 90% reduction. With the search analyst, the return cited verifiable answers from their approved finance documentation. Before, it was eight minutes per lookup, afterwards, it was 30 seconds, a 94% reduction.

And deep analysis agent. That reads and structures large documents, sets end-to-end. Before, it was almost half an hour to 10 hours per contract batch, two to 30 pages each. After, all the agreements in less than 10%, anywhere from a 67% to 98% reduction.

The positive value that was realized with focused, early adoption, during our private preview and limited availability working with Cox really worked well. And Cox has plans to expand further with new releases with our MCP, finance, agentic layer, and et cetera.

So one of the questions that often comes back to me is like, well, how did they get there? We talked about the foundation being built, the foundation being solid. Well, what they focused on was high impact, low effort activities.

They were already using OneStream. They had set their foundation. They had all their actuals, their plans, everything, their consolidations in one foundation. Then they did a cross-functional discovery, identified pain points across teams.

They defined AI use case opportunities and ranked the use cases by impact, not effort. Whatever implemented had to have a far reach and impact lots of users and value to effectively validate the ROI. And it was to build outcomes and a fact-based view of use cases and also align with IT.

A six-month process. But what it did was help to build transparency and trust. And also, and what that did was with transparency and trust, they could use AI because they trusted it to link the sources, have the agents show the thought process and steps, validating the use case for them.

Thank you. Thank you. Thank you so much, Prost, for making such a clear case study and use case for AI agents. I'm already receiving some questions related to AI agents, so be ready for them. If you have more questions to Prost, don't hesitate to ask right now through the chat box.

And meanwhile, I'll move on to the polling question that we would like to ask the audience. Which AI readiness area is currently weakest in your FP&A function? And I will launch this poll and let's see where do you stand in your organization.

So which AI readiness area is currently weakest in your FP&A function? Is it people and skills, data quality, process design, or technology platform? Let's see how what will split with those four options.

So which AI readiness area is currently the weakest? People and skills, data quality, process design, or technology platform? Majority of our participants have given their votes already, so let's end this poll and share the results with everyone.

What do we see? Strongest position is process design with 40%. Then we have data quality with 30%. Then goes people and skills with 17%. And the last one is technology platform with 13%. Prost, what do you think? Are you surprised by the results? I'm not surprised.

I'm actually happy. I mean, you know, I preached about data quality and it's unfortunate that data quality is still 30%. But process design being 40%, I think that's great because, again, AI, especially with agents, they can help refine and automate some of your process.

So the key thing here is that, you know, before you get into agents, as I talked about, AI and agents is not the first step. It's a last step. Fix your foundation. Fix your plumbing. Design your process or re-evaluate your process.

Look at how you want to transform and use AI to, you know, accelerate that. So, you know, the beauty here is that process design is not rocket science. It's difficult. It's painstaking. It requires a lot of cross-business collaboration and effort, but it can be done.

And once that's done, you can move on to data quality and you can fix the foundation to deploy AI afterwards. Well, thank you so much. Just when you were speaking, the question arrived here. What should we do first? Process harmonization data foundation and you have just answered through this comment.

Thank you so much for that, Pras. And let us move to the discussion. I see that Michael has already joined us on camera and what I would like to discuss with all of you is how do you approach AI readiness across people, data processes and skills without trying to change everything at once and let's connect to this question from the audience as well.

So let's start with you, Michael. How to approach that? How not to do everything at once? Well, I'm glad it's a nice, simple question. And of course, the answer really depends on where you are. You know, the one thing about AI, it gives us an opportunity to move away from fixed time-based processes.

It enables us to move to something almost like continuous planning where the organization reacts depending on what is actually going on in the business environment. Now, having said that, AI will play a big role in that.

One, in just detecting, you know, shifts in the marketplace. It will play a role in actually identifying the drivers. It will play a role in actually then forecasting and maybe doing some simulations.

So having said all that, I think the first thing to do is to say, where could AI be used within our organization? And there's a number of use cases out there and it is changing all the time. So I think the first thing, there needs to be a recognition that AI is going to change.

But then, how ready are we? and Pras already mentioned it, we've got to make sure we've got the foundations in place as well. But with that, do people know what AI can actually do? There's a lot of hype out there.

And so it's really important. I've met with some organizations where they actually have training classes, first of all, just to explain what is AI? What can it do? What is it realistically can do? What are other people doing? So I think part of AI redness is just being aware of what it can do.

Then starts the process of saying, where could we then implement AI within our organization? And do we have the skills? Do we have the governance in place? So it is a big question, but in some ways, it's a once in a lifetime opportunity because I think the guys that get this right, the organizing that get this right, have a chance to really accelerate into how they manage the business.

So I'm not sure that's very answered very well. But I think, first of all, be aware of what it can do and how it's changing. Don't necessarily believe the hype, but find out what people are doing and then say to yourself, how could we use that? Right, right.

Thank you. Thank you for this angle. And back to you, Pras, how to approach AI readiness not trying to do everything at once. Yeah, I think I'm aligned with what Michael said. You know, I think it's all about a sequence.

You know, at the end of the day, you know, I still go back to what the survey said. 35% of teams say data quality is their number one barrier to AI. And, you know, you can't fix the people, the processes, their skills if it's built on a foundation that AI can't trust.

So I would think of it as, you know, a sequence around a single use case, you know, rather than some big bang rollout. So, you know, maybe start with the data. You know, don't try to clean the whole enterprise, but pick one process, one problem where the data already lives and make that trustworthy.

Then bring the process into that same area as well. You know, take one workflow, whether it's a variance analysis or a piece of the forecast and let AI do the first pass there. And once you've got that, the people and skills come through that use case, you know, not before it, your team learns AI, but using it on the real task and, you know, the skill building comes with that as well.

And, you know, once you have that, you know, once you've got a use case that works, one clean process, you've got a template, a proof point that you can repeat for the next process. And that's how you effectively get transformation, I believe.

Right. Right. Thank you for your answer. And you know what? We are receiving so many good questions today. So I would like to jump straight to Q&A and start with a question related to survey and to you, like addressed to you, Michael.

So which survey finding should FP&A leaders treat as the strongest warning for the next 12 months? So what's the strongest warning signal for the upcoming year? Yeah, I guess how effective is the FP&A department? How do you rate yourselves? I think that's quite, and then link that to where do you spend your time? Because that's everything.

If you're performing, like if you're being a trusted business advisor and you're finding that the decisions, you know, that are based on the data and the analysis that you're putting forward, if the organization is coming to you and saying, that was really good, I didn't know about that, that is all the reason why FP&A exists.

If you're not getting that or if you're finding that the analysis are happening far too late, that's a warning signal because it means people are making decisions not based on the data now. Instead, they're actually post-justifying why it worked or why it didn't work.

So strongest signal, how are you seen by the rest of the organization? Do they trust what you say? Are you able to deliver information in time? If you're not, that's something that's a real warning signal.

Right, right. Thank you so much, Michael. Very interesting comment. For us, I have a question to you and related to agents. Where do you see AI agents creating value first, I guess in FP&A, data validation, forecasting, reporting, business partner support, or maybe anything else? So where do you see the most valuable use case for AI agents now? Yeah, I think the, it really depends on the organization and I hate saying it depends because it's such a consultant answer.

But I think the first thing is for every organization to validate and find out where you're finding a problem, just like our customer, Cox. Like they've found problems and these are problems that they could solve readily and easily using agents.

So, you know, I would look at, for example, data, right? Maybe you can use agents to bring your data together, validate the data so you're not doing it, right? We talk about the fact that finance is doing a lot of, you know, plumbing and things of that nature, right? Let's move finance and FP&A from the planning to more of the analysis, right? So maybe it's, and then from there, it's forecasting and forecasting automated variance analysis.

So giving triggers, like doing some sort of real-time analysis on your data and providing outputs in terms of outliers and explanations to that so that you and finance can give directions. So there's quite a few use cases, many, whether you want to start with forecasts, whether you want to start with data.

I think data and validating the data is always the most value because it drives such an ROI and from there, you can move on to better and automated forecasts, bringing in more data from forecasts. So I'm not sure if that was a good answer or not, but I think that's where you would start.

Thank you. Thank you so much. And, you know, I have several young questions related to AI and becoming part of FP&A. So I'll probably pick this one and I'll ask both of you to answer, starting with you, Michael.

So the question is, as AI becomes part of FP&A, what decisions should remain within finance, with finance professionals and what should we allow AI to handle? So what to do, what to outsource? Yeah, that's a really good question.

And it's actually one that we did some research on about six months ago. But basically, I think to begin with is that AI tends to be used for if like the low, low level impact decisions to begin with, just to build up some, some assurance that it's actually doing what we want it to do.

However, I think anything that is high value, high strategic value, I would be inclined at the moment to leave that till later. Let's pick off the low, you know, the stuff where we're spending our time.

Use that, go there first of all. And, you know, let's face it, well, finance in particular are held responsible for any like decisions that come out of it, our recommendations. So it has to be, I think to begin with, it's going to be the lower value, the lower impact ones.

Well, I mean, it's lower impact in terms of if it's wrong. That's what I'm trying to say. But maybe high impact because it saves us a lot of time until we've built up the expertise and we understand. Because AI is not a, it's not a silver bullet.

You've really got to understand how it works and how it came up with the answer. You can't just blindly accept what it comes up with. And that's something called explainable AI, which is becoming more prevalent.

Maybe Pras has a view on that. All right. Thank you, Michael. Pras, what do you think, what we should keep within human professionals and what we can delegate to AI? Yeah, I mean, at the end of the day, I mean, you know, I think AI is there to automate and really accelerate our decision making and accelerate our interpretation of data.

Because, you know, we've seen, if you look at the survey this year, more and more customers are doing data-driven decisions. But again, less customers trust their data. So there's obviously a disconnect there.

So let's fix the data. And once we fix the data, we can use AI to automate so much of it. I mean, there's, you know, AI in EPM tools, but there's also AI in our daily work. We all as finance love Excel.

And, you know, I don't know if anybody's used some of the, you know, large language models, the AI tools they have with Excel, but it's actually really cool. Like, I mean, I come from a day and age where I was building these incredible V lookups and H lookups, analyzing data from here and there.

The other day, I had to do some real, strong analysis of data. And I was able to work with these, you know, AI platforms to really bring it together and generate insights that might've taken me a lot longer.

But at the end of the day, once I got the insights, I was asking more questions of those insights. And as Michael said, I had to go back and validate and trust the data. So it's a two-part because workflow, because at the end of the day, I can't go back to whoever I was bringing this data to and just say, well, this is the analysis I got from an LLM.

That's not the way to do it. Rather, I went through the data, I trusted the data, I worked with the models, and ultimately, it made my life so much easier, gave me insights that probably would've taken me a lot longer, freed me up to do a lot more work, and I think really accelerated the value that I'm presenting.

And I know that FP&A organizations will be able to do as well. Yeah, indeed. It's probably that nothing we delegate fully, but nothing we do fully by human force now. So it's always, always in mix where I can help and speed it up.

Okay, I'll have one more question, and this time I'll start with you, Pras, so you have the benefit of answering first. So please answer very briefly, what is one action FP&A leaders should take now to close the gap between AI interest and AI readiness? The foundation.

I implore every FP&A organization, do not be hoodwinked. Do not fall for, you know, you know, lights and dazzle and things like that. There are customers that have been very successful with AI. Start looking at talking to other customers to see what they've done.

Look at your own foundation of data. Look at your own data foundation. You can't, AI is not here to solve your problems. It's a second step as I pointed out. Talk to your IT department, work with them.

Again, talk to other customers. Make sure you get a trusted story of what works because last thing I want customers on this call and FP&A practitioners is to be a guinea pig for something that, you know, is a marketing message that doesn't really work or whatnot.

Talk to customers, talk to vendors, get referenceable use cases, fix your foundation and move ahead from there. Perfect. Michael, back to you. What is one action FP&A leaders should take now to close the gap between AI interest and AI readiness? It's going to be one extended action, but I think the thing to ask is, is there a better way of doing what we currently do? Is there a better way? So often we do stuff based on practices from 20, 30, 40 years ago because that's what the technology allowed us to do.

But today, is there a better way of doing that? And if there is and AI has lots of solutions, the next thing is, well, how could AI help us in that? And what would we have to change internally in order to benefit from AI? You know, one of the things I always found is that, you know, I used to be involved in technology.

If organizations had a bad and broken process, throwing technology at it just meant they ended up with a faster bad and broken process. So the first thing is, what could we do better? Could AI help us? What would we need to change in terms of our experience in order to take advantage of that? Indeed.

Indeed. Sometimes fixing a process is enough and without adding AI on top. Just AI is a really good stimulator to actually fix your foundations and fix your processes behind. Thank you so much for this session.

We have many more questions. We'll try to answer them in writing after the session. Thank you all for sending your questions in. So stay tuned for written answers. Meanwhile, let me invite everyone to the upcoming FP&A Friends webinars.

one of them will be happening already tomorrow on July 9th and will be devoted to designing the FP&A operating model for the AI era. So please join us tomorrow. Then the Digital North American FP&A Circle will, on 16th of July, will be devoted to the topic from analysis to orchestration, how FP&A is evolving in the AI era.

And the last one to announce today is FP&A Trends webinar on 21st of July from reactive to proactive FP&A real-time scenario planning. So please join all of the upcoming events. As promised, how to get this year FP&A Trends survey, there is a link at the bottom of this slide.

You'll have access to presentations. Right now, we'll be sharing this deck after the event as well. So please download it and read the research in-depth. Once again, I would like to thank our technology sponsor OneStream for making this survey and for making this event possible.

And I would like to let you know how you can connect to us to our FP&A Trends global community so you can choose between our website, our FP&A Trends Digest, LinkedIn, YouTube, PlatformX, and so on. Please don't leave just yet.

Fill in the short survey at the end so that we can improve our webinars. And I would like to thank you all who joined us live. And most of all, I would like to thank our speakers, Michael, Prost, thank you so much for being here, for sharing your wisdom and your practical knowledge with us.

Have a good day, everyone, and see you tomorrow at the upcoming FP&A Trends webinar. Goodbye. Bye.