Breaking the FP&A value trap: Redefining Finance for the age of AI

Hi, welcome everyone. Thank you so much for joining today's webinar on the topic of breaking the FP&A value trap, redefining finance for the age of AI. If you have any questions, feel free to drop them into the Q&A.

If we don't get to your questions today, we will follow up in an email later. And so just to get to the introductions here, I'm Tiffany Ma, Director of Product Marketing for AI and Operational Analytics here at OneStream Software.

And Faram, do you mind giving yourself an introduction? Yeah, thanks, Tiffany. I'm excited to be here. I'm Faram Madhura. I lead Forrester's Global Research on ERP and Financial Planning. Wonderful. So let's go ahead and get started.

I'm just going to give a little bit of context for today's discussion. What we're seeing here in OneStream is that finance leaders today are navigating one of the most pivotal transitions in modern business.

From our Finance 2035 return to investment research, which was a global study of over 2,000 executives and investors, we found that 88% of investors believe that CFOs will be more important in 2035 than today, with nearly half saying significantly more important.

Yet at the same time, a majority of the CFOs admit they're struggling to drive strategy because they're overwhelmed by competing demands and legacy constraints. So this tension between expectation and execution really mirrors what Faram and the Forrester team call the FP&A value trap.

This is where organizations are investing heavily in finance technology, but too often they're digitizing broken processes instead of actually transforming them. So the result is a widening gap between finance transformation and actual realized business value.

So these insights, as I mentioned, are consistent with Forrester's findings where technology alone won't solve finance challenges, but as Faram's research shows, transformation requires re-engineering processes before automation.

So you have to align architecture with business ambition and separate AI reality from hype. So Faram, let's start with that fundamental insight from your research. You call it the value trap. So tell us, why are so many finance transformations failing to deliver the strategic value that CFOs are being asked to create? Yeah, well, Tiffany, that's a critical question.

I get such questions from clients, senior executives all the time after the fact. Faram, I'm not seeing the real transformative value from this implementation, even though we spend millions of dollars and we add all the right windows to support us.

And for everyone on this call, the stakes have never been higher. It's much more than focusing on efficiency gains. It's a lot around relevancy. It's critical because based on our market study that we conducted, almost two-thirds of respondents, the leaders within the FP&A practice, are planning on making investments in planning and budgeting areas in the next 12 months.

But here's the harsh reality. The truth is that when we uncovered some of the research in this area, these once-in-a-decade investments, most of them led to a strategic failure. And it's this fall and it's this situation that falls within the FP&A value trap.

So in this very particular space that we're talking about around planning and budgeting and modeling, it's not a technology issue as much as it is a flawed premise. Just the belief that a new platform can come in and fix broken processes, it won't.

But you are simply, as Tiffany mentioned, digitizing this function. So think about it this way. For a financial analyst, it means that whenever they implement this shiny cloud tool and they bring it in, that analyst, he or she is still going to be spending 40% to 70% of their time chasing numbers, manually reconciling between the data, between ERP, the planning system, and of course wrestling with all of those spreadsheets as well.

And all of this because the tool that was deployed in the environment was not configured to handle the ad hoc analysis that the business actually needed. And then think of it from the finance planning executive side.

At least they can plan faster, but it's just managing chaos at a much faster pace. And so really over here, what everyone is trying to achieve is meaningful scenario analysis. So when the CEO asks you next time, hey, what is the situation based on the supply chain disruption? You want to ensure that all your core operational drivers are integrated to offer that critical strategic valuable insight.

And that's that space where most organizations who have tried to implement such systems kind of fail because they fall short for at least one of, if not all of these three reasons. The first one being it's a failure to re-engineer the roles, not just the reports.

And I feel this is the biggest mistake coming out of the research that we got, which is organizations are trying to automate that old rigid annual budgeting process. True transformation requires re-engineering the work that the analysts do themselves and applying it into the automated world.

It means shifting that analyst role from being a report preparer to a business partner. One that analyzes forward-looking drivers, not just historic patterns. If your project plan that you put together does not reflect a new job description for those analysts, it's a sign that you're on the wrong path.

Second, underestimating the architectural debt. And in my report, I've given a view of how some of the vendors are architecting their portfolio and how it fits into one of the four buckets when we think about it from an architectural standpoint.

But ultimately, what it means for the CIOs is that it is a very critical choice because for a CIO, it's defining that architectural situation in terms of thinking of the integration tax or the debt that they'll have to take on.

And for the CFOs, the architectural decision should dictate how agile the team can operate. So a fragile landscape for an ERP or a legacy consolidation engine and multiple planning tools, all of that, if put together, starts, tends to create that integration tax.

It's a hidden, perpetual drain on the team's resources because they are still switching to gather the right context to make it relevant so that the system can help them get those insights that they need.

So it's a combination of all of that that comes together, which is why making the right architectural decision based on the kind of ecosystems that are available out there is an uncritical factor. And third, and the last one in this situation is no conversation ever goes away without discussing AI.

And so this is that AI ambition gap that we want to talk about. Majority of the executives I speak to tend to say that our C-suite wants to have AI-driven forecasts. I mean, it's great. They saw a nice demo.

They see what the tool can do. They know what good looks like. But that falls off. The cracks start to appear when you realize that the finance team lacks the single governed, trusted data set required to produce anything credible.

And so you can't be running sophisticated ML models, AI engines on a foundation that is inconsistent. You don't have the metadata in place and you have manual spreadsheets to manage adjustments. So summing it all up, the value trap in this situation is that cycle of making significant investments that yield only marginal gains, trapping finance into its old backward looking role and not really thinking of it, how to evolve using this technology investment into the future.

Those are some really great points, Faram. Thank you for that. So my takeaway is you shouldn't be layering AI on broken disparate systems and processes. You mentioned technical debt, which I'll hit on later on in today's discussion.

And more importantly, new technology and modern technology requires new processes and roles. And so that leads me to the next point, which is the expanding mandate of the office of the CFO. Right. So we're seeing from the finance 2035 study, we are seeing a shift from for the office of the CFO from reporting on the past to be to steering the business forward with strategy.

And that requires more than just software. It demands new capabilities, as you mentioned. And again, in our 2035, finance 2035 research, CFOs told us that they must become masters of everything, balancing operational rigor with strategic vision.

Two thirds of CFOs believe their company's success or failure rests on the CFO's shoulders. A majority of business leaders expect CFOs to be masters of everything, integrating financial, operational and ESG insights.

And then CFO competence now ranks second only to market expansion in investment decisions. So Faram, as CFOs evolve from data stewards to enterprise strategists, how can they lead with confidence if their insights are still built on fragmented data? And why is a unified data model so essential to realizing that strategic vision? Yep. Before I get into that, I do agree that the role of the office of the CFO has been elevated more so now than ever.

And to touch upon that, I'm going to talk about unified data model. A unified data model, from my perspective, is one of the most critical ways out of that trap. Because our research states it very clearly.

Technology doesn't fail you in this instance. It's that architecture that you've thought about bringing it together that tends to fail. And so to put it more specifically, it's that integration tax that you need to factor in.

Because it takes three days for the analyst to put everything together, to tie all the data together, to make that cash flow statement. Because your consolidation and your planning systems have different hierarchies.

So it's that crisis meeting when the planning manager has to meet with the sales operations folks because their forecast in the planning tool doesn't match the actuals that are coming out of the ERP system.

And that's destroying or creating more friction from a business partnership standpoint where you actually need to build trust and make sure that the numbers are flowing nicely from an operational standpoint to a planning standpoint.

So it's also the situation that I come across many times where leaders enter a meeting when they're presenting their executive presentation and they need to make those caveats to say these numbers are directionally correct, subject to change and things like that.

And that's because there is no single auditable source of truth. So in this situation, while many would say it seems to be like a technology problem, no, it is truly a credibility problem that undermines the entire finance function.

Because as Tiffany, as you said, right, the expectations are evolving, expectations are going higher. It's not backward looking. It's forward looking strategic insight. So all of that is not going to be possible if you have a fragmented architecture that creates two versions of truth of the data, which means you don't really have trusted data.

And so you are then going off based on the opinions by looking at all of these disparate data sets. And we can't in today's evolving day and age think about building a multi-million, billion dollar capital allocation decision based on some of those gut checks.

All right. If you can give me a click. And this is why, thank you. And this is why our research is so prescriptive. Data foundation investments are non-negotiable. A unified platform is that single data model.

It's that single data model is not an IT convenience like how it used to be where the IT folks would go in and catalog it and link it and things like that. It is a fundamental prerequisite to establish a single version of financial and operational narrative coming together.

And the goal over here is to eliminate the low work of reconciliation and provide the governed high quality data foundation needed to deliver on that one promise of agility. And even more critically in today's day and age to make AI a reality more than a science or a pilot project.

Yeah, totally agree. And I think, you know, you mentioned technical debt. It's not just about the price for the separate pieces of software, but it's everything you mentioned of the data reconciliation transfer, metadata transfer, everything.

And so, yeah, that adds up and that prevents you from having that single source of truth to be agile and make decisions and speak from the same truth. So great points there for them. Just to continue on that point of technical debt, you know, 74% of CFOs believe unified data and data-based decision making are now the key determinants of success.

And the most successful finance organizations are the ones that invest in a unified platform with a unified data model, not a patchwork of different modules. And why is that? That's because that structure of that unified data model delivers on really three non-negotiable advantages.

The first one is what you mentioned, Faram, is that consistency and trust, that single source of truth. So you're moving the need for that data and metadata transfer and duplication, reconciliation and validation.

And you're really freeing up finance's time away from that low-value work to high-value work. And you're shifting, again, that role of finance to be more strategists as opposed to, you know, backwards-looking or backwards-reporting organizations.

The second advantage is speed and agility, which you mentioned as well, Faram, is that with a unified planning, consolidations, and reporting processes, you're able to run all these processes in parallel and gain economies of scale.

Actuals are automatically seeding your forecasts. You're moving towards continuous close and touchless planning processes. And then, finally, you have that governance and AI readiness, which, again, Faram's research reiterates here.

So OneStream's unified data, what does that mean? So OneStream's single unified platform replaces multiple spot or point solutions, which then streamlines your core processes, your data management and quality, your closing consolidations, planning and forecasting and reporting and analytics and compliance, all in one single platform, one single data model.

And what this does is it eliminates technical debt, as we mentioned before, and everything associated with it. And then with OneStream's extensibility, we're able to adapt to each business unit's unique needs while maintaining a governed enterprise-wide structure with no bolt-ons or extra provisioning or extra technical debt needed.

This means that OneStream's single platform can accommodate different business unit structures and can adapt as your organization inevitably evolves to changing business conditions. And it can do this in a single platform without additional applications, without additional administrators, and ultimately, a more agile finance function that scales without adding complexity.

We've actually seen customers with measurable benefits from reduced technical debt to efficiency, effectiveness in terms of making better decisions more quickly, and even more with improved risk mitigation.

So by avoiding costly mistakes like audit fees and legal issues or any redundancy costs. So on average, our customers in FP&A are seeing an efficiency improvements of 58% by moving from disconnected spot solutions to OneStream's unified platform.

So Faram, we've established that a unified data model gives finance one version of the truth, but AI is only as powerful as the data behind it. So how do governance and data quality determine whether AI delivers insights or just more noise? Yeah.

With that foundation of the data, we can now have a more realistic conversation about AI, Tiffany. And this year, I've attended several vendor events, SAP, Oracle, Workday. And one thing that you would see is for around their AI story, financial planning and analytics was pretty much high up there in terms of that flagship situation of showcasing the true impact of AI and how it can drive certain business decisions forward as well.

And so my role as an analyst is to separate the hype from the hard reality and the marketing claims around AI and FP&A and making sure that you folks have a clear right view and understanding what the situation is.

So let's apply some rigor over here. Talking about prescriptive forecasting, right? We've noticed that it delivers conditional value. Organizations have seen 10 to 20% accuracy improvement, but only under specific conditions and more so making sure that you have a stable business driver set up with data backing at least two years or beyond that, making sure it's clean, consistent, and the historical data is in that intact form.

So can you do forecasting about your expenses, your general selling administrative expenses? Yeah. For a mature division, sure, you'll see higher accuracy improvements and gains over there. But then if you had to do forecasting revenue for a new product launch in a volatile market, that is where the value starts appearing to be conditional and requires more surgical application to tune the system in the right way.

Natural language processing generally tends to fall short of executive expectations. And I say that because the conversational interface is nice. It's convenient. It can handle simple queries. But the system starts struggling with nuance, financial analysis that needs to be done on top of that.

So simple patterns, trends based on the graphs that you see, yes, it's going to do that. But the second you ask the system to help it understand how to handle complex multivariate interactions that an experienced analyst would want, that's where the analyst is going to recognize some inconsistencies in the way the system's handling it.

So, I mean, scenario modeling, this next one is one of the most critical AI drivers. We are seeing improvements in this space. It all depends on how all of these components around AI, ML data come together.

And there is room to play around that. Anomaly detection generally seems to be one of the most common use cases that you will see, the low-hanging fruit that you'll see across several product lines. And I'm seeing a positive pattern, mostly in terms of handling the false positives, false negatives, false positives.

And in this situation, we want to build trust. The analyst needs to know that the data that the system is throwing out, the anomaly that it's detecting, is the correct one. You don't want the analyst to be investigating, spending many days to just figure out that, oh, yeah, it had classified it the wrong way, and that's how it had to be done.

So, without sophisticated filtering, it's a distraction more than an insightful engine in this situation. So, in general, summing all of this up together, what I want to leave you with on this slide is AI is not a feature you buy.

It's a utility that you consume. And your AI strategy, especially around this, must be a series of carefully planned pilots, ROI, because with every AI deployment come all the additional costs around the tokens and the consumption and things like that.

So, the effort that's going to require to build that unified foundation and focus on all of them rather than treating it like a feature checklist on a vendor scorecard. Because that's where we've seen that folks fall into that trap and then realizing that they should have done things after the fact.

Yeah. Yeah. Thank you for that, Faram. Yeah. I agree. I think, you know, AI is important, and I know there's a lot of executives that are pushing for it, but it should be done in a very structured, planned, strategic way with proper use cases in place.

So, thank you for that. So, on that, I'm just going to go into OneStream's forecasting, machine learning time series forecasting solution. You know, Faram, you mentioned that predictive forecasting delivers conditional value and it's, you know, it's performing better than, it's providing more reliable answers than, let's say, natural language processing.

So, machine learning has been around for a long time and, you know, natural language processing is more of the more recent evolution of AI. And so, I can see it still needs to be refined a bit in terms of their output.

But for OneStream, our AI strategy centers around embedding AI and machine learning throughout the unified platform. Because AI performs best on unified data and a unified data model rather than fragmented point solutions, we've embedded AI throughout OneStream's single governed data model.

And so, that produces more accurate, trustworthy, and scalable results. And so, what we've seen, so, with our solution, one of our solutions is Sensible AI Forecast, which is a purpose-built machine learning time series forecasting solution that's purpose-built for finance.

And it's unified within OneStream's governed, trusted, and secure platform. So, the point is, it's built for finance. You don't need a data scientist to implement or use this. And it's a series of guided workflows with autonomous machine learning steps to have a finance or operations user click through each step in order to go through the entire data pipeline, from data ingestion and data quality to model building, training, to deployment, to utilization.

From the whole workflow, the end user finance or operations is using our sensible AI forecast solution. So, we're really minimizing that learning curve for training and enablement in order to get back, in order to use machine learning.

So, going back to starting from time back to Faram's point of upskilling workers in order to be able to use AI within the finance group. So, what we're seeing within our sensible AI forecast solution, on average, across all of our customers and 125 different projects, is that we're seeing an increase in accuracy of their forecasts, on average, of about 24%.

So, now they're shifting from having to wait until the month end to create a forecast that might be stale to rolling forecasts that they can be proactive about steering the business. And now they can start using machine learning scenario modeling by saying, hey, these variables and drivers, if I tweak, you know, how, if the interest rates will change this way, or if I run a promotion earlier in the year versus later with, you know, a higher percentage, how will that affect my sales? Right, you're really incorporating a lot more variables and flexibility for finance to be able to be that strategic advisor, step into that more strategic advisor role within the organization.

So, just some customer, just to walk through how our sensible AI forecast solution works. If you were an organization and you're looking at your organization in different ways, if you have products, locations, and you have your P&L accounts, let's say we're forecasting for sales, what we do is we take all of that information, historical information for those different areas, and we pull in that historical information either at quarterly, monthly, weekly, or even daily granularity.

And then we're pulling it from not necessarily within one stream, but also beyond one stream. So, it could be your data warehouses, your snowflakes, your ERPs, your CRM systems, whatever systems that hold this information at that granularity we pull in.

And then with our model intelligence that's built in, where we can pull in with a click of a button, any information from weather to macroeconomic information like GDP, inflation, unemployment rates, as well as your internal information like your promotions or whatever it may be.

And then we run it through our proprietary autonomous AI model arena. So, basically, our models compete against each other. Our machine learning models compete against each other to find the most accurate machine learning model for each one of those line items that you're forecasting for.

So, for that product, location, account combination, each one of those will have a tailored machine learning model that is the most appropriate and produces the most accurate forecast for that line item.

And so, this continuous learning loop is very beneficial in being able to have accurate forecasts and continually create accurate forecasts. Now, what this means is accelerated time to value. We're implementing these projects and creating forecasts within weeks versus months or years of what we see in the traditional machine learning forecast projects.

Again, no data scientists are required. So, finance is owning this. And so, finance knows when they go into that meeting with their business partners, with their bosses, with the board of directors, they know exactly how they got to that forecast and they can justify it.

We have, we're hyper-focused on transparency. We know finance has to trust where those numbers came from. And so, every step of the way, we have transparency dashboards telling the user exactly what variables or drivers are impacting your forecast down to the day.

So, if you're forecasting out two weeks and you see your forecast is 100 units of, let's say, demand, we can say 20 units were pulled up due to your promotion. 10 units were pulled down due to inflation.

30 units were pulled up due to GDP. So, we really have that transparency focus for every forecast and every step of the way. And just to talk through some customer examples here of our sensible AI forecast solution, what this shows is really the breadth of not only use cases, but also company sizes, as well as industry.

And so, I'm going to focus on the two on the right here, which is the multinational $500 billion revenue retailer and their use of sensible AI forecast and compare that against the infrastructure and field services organization, which is $250 million in revenue.

So, we're seeing large organizations as well as small SMB organizations leveraging and gaining benefit from our solution. And so, starting with the multinational retailer, you can see their accuracy of their forecast now with our solution, sensible AI forecast, if they're forecasting at 99% overall accuracy.

So, they're essentially predicting the future here. And when we came in, they had embarked on a two-year project with data scientists trying to create exactly what we did, which is a machine learning forecast.

And what we did was we came in with sensible AI forecast, which is autonomous machine learning purpose built for finance. And we were able to meet and beat those results within two weeks of that two-year project.

And so, very quick codification of these steps and processes to create these machine learning forecasts. And we were able to prove that value out very quickly with this multinational retailer. On the flip side, with the infrastructure and field services organization, they were forecasting more operationally.

So, they were forecasting demand. Whenever they had tickets come in, they had to deploy labor to go to on-site to resolve that ticket. So, as you imagine, the labor costs were very high within the organization.

By increasing the accuracy by 20% for this organization, we were able to save them millions of dollars because they get 10 million tickets within a year. So, if we're able to improve that accuracy, we're saving the money that they used to deploy the resources to go on-site.

We're saving money on the resources that the labor needed, including trucks, computers, and all the equipment that their labor needed in order to go fulfill these tickets. And so, huge improvements and benefits to these organizations.

So, Faram, we've talked about the evolution of the CFO, the role of unified data, and how AI can amplify finance's impact. But transformation takes focus and discipline. So, as we close, what guiding principles from your forester research would you share with CFOs to help them move forward with confidence? Well, the path forward requires mainly or mostly two things that you were alluding to, executive courage and operational discipline.

Other folks on this call, you may have expected something different, but fundamentally, this is what's needed to break the cycle to unlock real value. And our research, based on all the assessments that we've done, I want to leave you with three non-negotiable principles to take on, take forward.

The first one is the golden rule, re-engineered before you automate. Don't just digitize that annual budget, challenge its very existence. And, Tiffany, as you were talking about that goal of shifting from a rolling forecast and value-driven planning, that's really what it means.

It means you are shifting your analyst role from a data consolidator to a strategic advisor who partners with the business to model the future outcomes. Second, we all talk about returns. So, it's demand high-dollar returns, but measure the outcomes in the right way.

So, what it really means is stop measuring success based on those RCA technical metrics like reports generated, reports access, things like that. The real KPIs over your business metrics that you should question and you should force the vendors to offer is, can this system reduce forecast cycle from X weeks to X days? Can you improve my forecast actually based on all the critical revenue drivers by X percent? Those are those very practical business-focused outcomes that you would want such an implementation to drive.

The real value of such a platform is ultimately reallocating your analyst time from data validation to high-impact activities. Think about modeling M&A opportunities or even assessing competitive pricing threats.

The return that you should demand from your investment are these kind of indirect and direct benefits that your organization takes away. And third, drive quick wins to fund the broader transformation.

And don't take this out of context, but when I say the big bang rollout is dead, it's too slow and it's too risky. What I mean over here is execute such a system in phases. Delivery of certain outcomes is indispensable rather than thinking about going fast.

Start with the universal non-negotiable pain point. Think of automating your financial close and your consolidation process as an example. Use the success from that victory, the faster close, the more reliable numbers to secure, as Tiffany was talking about, to drive that political and financial capital for an ambitious phase to fund the future and also improve the future workflow and also things around sales planning and things like that.

All right. If you can give me a click, Tiffany. So ultimately, applying these principles, it's all about evolving and elevating the finance function, as we started talking about at the beginning of the session.

It's a journey. Technology is the catalyst. But your people, the C-suite's commitment to discipline and the entire organization's willingness to shift focus from how to drive the transformation is going to be critical.

The end game is to change that conversation permanently, moving from what happened last quarter to becoming a more proactive finance partner who strategically talks about advising the business on what must we do to win the next one.

And that's really when you know that the execution and the solution that you focused on so extensively has worked out well for your organization. Yeah. Yeah. Thank you so much for that. Thanks for those insights and your advice for the path forward.

Really very helpful. We have a few questions that are coming in. Yeah. The first one that looks like is directed to you is how do we prove the hard dollar ROI of process reengineering? Oh, that's a central point.

That's a really good question. The exact point is about where do most transformations fail, right? We've all been conditioned to think of software as the main project and the painful phase two being the process change that we need to go through.

In this very situation of such critical applications, it is entirely backward. The software is not the investment. The software is just the expense, quote, expense. But the transformation of your process is that actual investment.

The hard ROI that you're looking for doesn't come from the platform features that were activated. It comes from reallocating the analyst time that you freed up to do high-level, high-valued activities like modeling and things like that and stop trying to do the process work.

Start justifying the software contingent on the process work. So all said, don't boil the ocean. Our research is very clear about it. Drive quick wins to fund the transformation. So pick that biggest point that I was talking about.

It could be consolidation. It could be re-injuring other elements. But collapse that time. Then just the justification of the cost avoidance of reducing the close time and the benefits and the potential additional visibility that you can get, that in itself becomes a business case for funding your next iteration of improvements.

Yeah, I agree. Don't boil the ocean. Start with the small quick wins. I think you're, you know, going back to your third guiding principle of, you know, quick wins, I think implementation fatigue is real.

And so you need to find those accelerated time-to-value projects that can prove value and hard dollar ROIs quickly so that the rest of the company and organization leadership can get buy-in quickly and then expand out further.

So that's great, great advice. Second question is also directed towards you. So what is the practical first step to upskill an existing team? Yeah. Upskill the existing team, right? That's what I would call the capacity trap.

You cannot create a strategic analyst role when they're already drowning in manual journal entries and spreadsheet reconciliation and things like that. And it's also not helpful when the executive or the manager points them to that two-hour online course around data modeling to say, hey, this will help you fix things.

You don't find time to transform. You create it. And particularly with systems like this, the transformation itself must be the engine that creates its own bandwidth. And what I mean by that is your very first move is in training.

It's liberation. It's freeing up time. And so what you do with some of these platforms is you find those most low-valued but time-consuming activities and use that aggressively to drive some of that automation, that stewardship around.

Let us automate some of these processes that are taking time. It's repetitive. It's low-value work. It's data collection, validation, whatever that may be, basic reporting. And that's that quick win. The moment you start using technology to fix that problem, knowing how to fix that problem, that in itself will start justifying the next step, which is the automation in this case isn't just the efficiency gain that you tend to report, but it's actually a capacity creation project.

You take the time saved by that resource not having to do that low-value work, and you put that or you reinvest that time into the new R&D budget to drive some of the upskilling work that they would need to do.

So transformation funds itself in this way, and not just the dollars, but the hours that your team needs to evolve. And so it's that cycle that I'm talking about where you must automate the past to create time to build the future in some ways.

I love that. Make time for upskilling. You're right. Everyone always is dealing with fire drills and already really busy, so you have to make time for that. I love that. Okay, so this last one that we have time for could be for either of us, but I'll answer it and you can chime in if I miss anything or if you have anything to add.

So the question is, our finance organization wants to start with AI, but how do we get started? So I'm just going to go ahead and say is that the number one thing is starting off with a clear business use case.

So when we think about AI, what's really important here is value-driven adoption. So having clarity on what's the business problem, the business use case that you want to leverage AI to tackle, and how do we think about that outcome and the success factors that we want to use to measure that outcome? It's very incredibly important to really have that, you know, as you mentioned before, the hard dollar returns and really prove the ROI metrics.

And so that's what I would say. I'm not sure if you have anything to add to that, Faram. You captured all of it, mostly. I'm going to reiterate on one thing. Don't treat AI as that feature. Treat it as the utility.

So all the things that you need to make sure that is wired in the right way, the architecture, the data, all of the piping underneath, but most importantly, that process that needs to evolve. Once you get all of that right, you will then see and be ready to adapt with AI.

I think going ahead and, as Tiffany pointed out, with AI being the primary use case to activate it is that false premise and hence the value trap. So, yeah. All right. Great question. Yeah. Great questions.

All right. Well, thank you all so much again for joining and thank you, Faram, for sharing your insights and the results from your study. And I hope everyone has a great day. Thank you. Thanks, everyone.