14 May 2026 · 51m · Sean Brown, Kevin Carmody, Andrea Tricoli, Davide Grande · 19 quoted facts
Finance chiefs are using AI to move from backward-looking reporting and bookkeeping to being the CEO's strategic partner. The guests argue the value comes from forward-looking insight, from cutting enterprise-wide costs and from freeing finance staff from manual work. They say CFOs should treat AI as a way to create value for the whole company, not just to trim the finance team.
The tension is between speed and discipline. Adoption is rising quickly, and the guests give real examples in budgeting, procurement and energy costs. Yet most failures come from habits, not technology: waiting for perfect data, running scattered pilots, skipping a clear vision and neglecting change management. Their advice is to set a vision, pick one domain, transform it end to end and demand clear returns.
Kevin Carmody says the biggest challenge is how a company has historically viewed its CFO. Davide Grande adds that technology is becoming a commodity, and the best models deliver no value if people don't change how they work.
Davide says some clients spent hundreds of millions on a perfect data platform and never got there. He advises starting with the data you have and improving it along the way.
The guests warn against the 'pilot trap' of many small experiments. They recommend transforming one area end to end, such as planning, cost or accounts payable and receivable, while working toward a clear 2030 vision.
One client cut external costs by about 9-10% across more than one hundred levers, including an AI model that flags branches using too much energy. Another used estimates of how much vendor work AI will automate to renegotiate IT help-desk rates.
Andrea Tricoli describes chained AI agents (analyst, market research, operations, scenarios, competitor stress test, visualization) that produce a first-draft budget. A person then reviews it and can ask for reruns, which saves time and improves the analysis.
Kevin Carmody says the most ambitious CFOs want to move beyond bookkeeping and reporting to become the CEO's strategic thought partner. AI helps by giving them faster, forward-looking insight and freeing finance staff from manual work.
Davide Grande cites McKinsey's CFO surveys: 24% had invested in AI in finance in 2023, 50% in 2024 and 65% in 2025. Active use, as opposed to piloting, rose from 6% in 2024 to 51% in 2025.
An agent is an AI tool that carries out a task on its own, such as analyzing past variances or researching market trends. Andrea Tricoli describes a team of them working in sequence to produce a first-draft budget that analysts and the CFO then review.
The guests give three examples. One is checking invoices against contract terms, where leakage is often 1-4% of value. Another is spotting wasteful energy use at branches. The third is renegotiating supplier rates because vendors will soon automate more of their work.
Davide Grande lists waiting for perfect data, trying to change everything at once, starting without a roadmap, neglecting change management and failing to simplify processes first. He also says to be obsessed with measurable value.
Stated by the speakers. Each quote matches the transcript, which shows what was said, not that it is true.
In McKinsey's annual CFO surveys, the share of respondents reporting AI investment in their finance function was 24% in 2023, 50% in 2024 and 65% in 2025, according to Davide Grande.
“In 2023, 24% of our participants said yes. In 2024, it was 50%, in 2025, 65%.”
Davide Grande said that the share of surveyed CFOs actively using AI in finance rather than piloting it rose from 6% in 2024 to 51% in 2025.
“In 2024, 6% were active user, in 2025, 51%. So seven times increase in finance department.”
Davide Grande advised that every euro invested in AI should return ten times the investment.
“The most important part of this, I think, is whenever you do AI, be obsessed with value creation. Every euro invested in AI must return ten times the investment.”
Davide Grande said AI can move mergers and acquisitions from manual, time-consuming research to a programmatic approach that quickly identifies acquisition targets.
“In M A, we'll move from a contingent approach with manual time-consuming research to a programmatic approach where you can basically instantly find the best targets for M A”
Davide Grande said that in McKinsey's experience more than 50% of the time spent on strategic planning and control goes into creating reports that nobody reads and crunching data.
“according to our experience, more than 50% of the time of strategic planning and control is typically spent in creating reports that nobody reads, crunching data, et cetera.”
Andrea Tricoli described a budgeting workflow in which a team of AI agents working in sequence produces the first draft of a budget.
“The example I wanted to talk about is the budgeting one, where you can have the first draft of a budget put together by a team of agents.”
Andrea Tricoli said McKinsey benchmarks show organizations often lose between one and four percent of value because contract terms such as discounts, rebates and hourly rates are not reflected in invoices.
“Our benchmarks show that often organizations leave on the table between one and four percent of value in leakage because complex terms, discounts, rebates, or specific hourly rates are not always always reflected in invoices.”
A highly efficient company combined data from more than fifty sources, including invoices, contracts, payments and operational data, to look for AI-driven cost savings, according to Davide Grande.
“why don't we leverage the data we have from more than fifty data sources, both financial data like invoices, contracts, payments, so not just numbers, but also PDFs”
Using AI on combined data, one client identified more than one hundred cost levers and reduced external costs by about 9-10%, according to Davide Grande.
“they have found more than one hundred leavers, and they they reduced external cost by 9%, nine to 10%.”
A client built an AI model that calculated the expected daily energy consumption of each branch, and the CFO's team received an alert when actual use exceeded it.
“they basically built an AI model where in each branch they would calculate every day of the year what is the statistically expected energy consumption.”
Davide Grande said the energy-monitoring AI model alone saved the client tens of millions of euros, at a time of high energy prices linked to the war in Ukraine.
“So that means only with this AI model, they saved tens of millions of euros.”
Davide Grande cited McKinsey Global Institute modelling that digital technology activities are about 11% automatable now and could reach 53% by 2030.
“For example, digital technology is already automatable at 11%, and by twenty thirty, it will be at 53%, right?”
A client renegotiated IT help-desk contracts, winning 10% savings immediately and 15-20% in future contracts by arguing that suppliers' AI-driven productivity gains should be shared.
“And so they went there, hey, we're going to share this, these profits. And so they renegotiated 10% savings right away and fifteen to 20% savings in the future contracts.”
Andrea Tricoli said an AI agent for investor relations can be loaded with a company's past market interactions, public data and peers' interactions so that it asks hard questions investors might raise.
“you can effectively load all your previous interactions with the market, public data, as well as your peers, other companies, interactions with the market”
Kevin Carmody said CFOs report that AI has cut preparation for quarterly earnings from weeks of analyst effort to days.
“So they're going from literally man hours of weeks uh to days of time to get ready for their quarterly earnings, which frees up so much time in the finance organization”
Davide Grande said some of his clients have invested hundreds of millions in building a perfect data platform and still not reached that goal.
“I've got clients that have invested hundreds of millions in building the perfect data platform, not getting there.”
Kevin Carmody said some CFOs have laid out every report they produce to find out who uses it, and in extreme cases cut most of them off to see whether anyone missed them.
“we've heard countless stories about CFOs that have literally put all the reports that they generate over time on a table to try to figure out who is using this information.”
Kevin Carmody said bold CEOs run transformations through four pillars: growth, cost optimization, the team, and digital or technology, with technology acting as an accelerant.
“they feed it through four performance pillars, which start with growth, second is cost optimization, third is your team, so the organization, the upskilling we talked about”
Andrea Tricoli said security for AI tools in finance follows each company's existing security frameworks, as it did for earlier tools.
“The way we think about it is not, it's very similar to to to the past. The appropriate security needs to be followed.”
Kevin Carmody lays out six themes showing how ambitious CFOs are redefining their role, which frames the rest of the discussion.
“the most ambitious CFOs that we've talked to have really taken the opportunity”
Changing how a company views its finance chief is described as the first building block before any technology change.
“probably the biggest challenge is culturally just how does a company historically perceive”
Complexity, report overload and financial risk keep CFOs tied to daily operations, which limits how much they can reinvent the function.
“we've heard countless stories about CFOs that have literally put all the reports”
Davide Grande walks through planning, cash, cost, investor relations, M&A and risk, and cites survey data showing fast adoption.
“So we looked at this through six domains.”
Survey figures show finance AI moving from pilots to everyday use in just a couple of years.
“In 2023, 24% of our participants said yes.”
Andrea Tricoli shows how chained agents can draft a budget and how checking invoices against contracts can recover leaked value.
“the first draft of a budget put together by a team of agents”
Two client examples show AI finding hundreds of cost levers and giving CFOs leverage in supplier talks.
“we're actually sitting on oil, but we're not deploying this oil”
AI can act as a challenger that anticipates investor questions, and it shortens preparation for earnings and board meetings from weeks to days.
“a very uh crisp and performing agent that that asks you hard-hitting questions”
Davide lists the common mistakes and says to start with a vision and one value-creating domain rather than technology or governance.
“Yeah, I mean, number one is waiting for the perfect data, right?”
Kevin closes with a sequenced, domain-by-domain plan that starts with vision and relies heavily on change management.
“We call that the CFO twenty thirty vision”
The most ambitious CFOs are saying we don't want to simply be someone that works within the finance and accounting function. And what we wanna do is help shape the way the company thinks about the business, how it makes decisions. And really, we think, especially in the era of AI, leveraging technology to get better insights faster can help position companies to pivot more quickly as their circumstances and competitive environments change.
From McKinsey and Company, I'm Sean Brown, and welcome to Inside the Strategy Room. You just heard senior partner Kevin Carmody touching on the opportunity that today's CFOs see to transform their functions and their businesses with AI. Our recent survey of CFOs around the globe reveals a shift in the function towards strategic thought partnership in the C-suite, with finance chiefs holding a unique position to leverage AI to drive performance and inform key decisions. As you'll hear, from strategic planning and cost management to risk mitigation and investor relations, AI is indeed reshaping the function and creating opportunities for value creation that stretch across the organization.
In today's episode, we'll explore what the finance function of the future looks like and discuss practical examples of AI in action today. And we'll also unpack what makes for a successful AI implementation for CFOs and finance teams, including common barriers and how to overcome them. And for those wondering where to start, I can tell you right now that it is not going to be trying to implement the perfect technology or the perfect data. And now I'd like to introduce our guests.
Kevin Carmody is a senior partner in our Phoenix office. He leads our CFO practice globally and was a founder and leader of our transformation practice as well. Kevin advises senior executive teams, boards, and other key stakeholders through periods of significant change, including CEO and CFO transitions. Kevin has also served in interim chief transformation officer and chief restructuring officer roles at a number of global companies. Kevin, it's great to have you here today.
Thanks, Sean, it's a pleasure to be here.
And Davide Grande is a partner in our Milan office. He leads our data and artificial intelligence work in Europe, the Middle East and Africa, as well as our strategy and corporate finance work for banks across the region. Davide, welcome to the podcast.
Thank you, Sean. Great to be here today.
And Andrea Tricoli is an associate partner in our London office and a leader in our strategy and corporate finance practice. Andrea, welcome.
Thank you, Sean. Great to join you today.
Great to join you today.
Excited to have all of you here with us today. And Kevin, I thought we could start with you to set the context. You speak with many CFOs and their finance teams about how they help drive value creation in their organizations. What are you hearing from them on their priorities, on how they use technology and AI to help drive performance, especially in a more disruptive environment?
Sure, well, well, as we talk to CFOs as they think about how they drive performance across their entire organization, speed has always been at the top of their list. But with the disruption that you're seeing in the marketplace, industry, geography, et cetera, and the rapid pace of technological change, what we're seeing is that CFOs are embedding how they leverage technology, how they leverage AI to get better insights and drive performance in a way that hasn't been seen before.
So it's an exciting time for CFOs to harness this change.
And Andrea, anything you would add, given your focus on AI and finance?
I would say it's not just CFOs. We are talking also to a lot of finance teams, and the opportunity from AI is also coming. Uh, uh, very live in the daily lives of of people that are working in finance organizations, uh, through processes for sure, but also in terms of personal productivity.
And one of the things that you all mentioned in your article is the real shift in the role of the finance team, and in particular, the CFO's role from that of traditional financial reporting or even operational and risk processes toward more of a value creation focus. Kevin, how much are you seeing this now? And and how has this surfaced in your survey and in the conversations you're having with finance chiefs and their teams?
Sure, and there really are six points that routinely come up in conversations with CFOs and and folks throughout the finance organization. And the headline really is the most ambitious CFOs that we've talked to have really taken the opportunity to redefine the role of the CFO in the organization, but also what their finance team does. CFOs that are the most ambitious will say, I'm a member of the executive team. I want to be a partner to the business, even more so than maybe we've done before, or even more so relative to the way the organization or the markets perceive the role of the CFO.
And what in my mind, what that really says is that CFOs are looking to elevate that role. In many cases, becoming the strategic thought partner of the CEO. I like to say the first among equals in the C-suite because they have a unique view of the entire business with the technical capabilities that can help drive performance. So their insights, their natural skills, how they see the business, and also how they can interact with business units and different functional leaders is unique, which we think gives them an opportunity to really step up how they think about driving performance across the organization.
Just second point, just to contrast, would be that the most ambitious CFOs are saying we don't want to simply be someone that works within the finance and accounting function. That's important, but we don't want to be a classic accountant or a classic uh treasury management uh person. What we want to do is help shape the way the company thinks about the business, how it makes decisions. And really, we think, especially in the era of AI, leveraging technology to get better insights faster can help position companies to pivot more quickly as their circumstances and competitive environments change.
So a real opportunity to move from kind of a bookkeeping role into more of a strategic partner that helps uh the company make better decisions. The third point that we talk a lot about is how to manage risk without compromising how you manage risk in the classic sense across the organization. We we see CFOs that take a step back and say, how do we actually enhance that? So we we look at enterprise risk management issues in a different way, more forward-looking uh to tackle some of the challenges that companies face.
Again, just another opportunity to step up the role of the CFO to help advise and counsel the CEO, but also the broader executive team, all the way through line management that execute on a daily basis across the company. And what we see is that CFOs have historically been the leadership of the finance function, treasury function, the accounting function, but we see CFOs that want to lean into their role to help make better strategic investment decisions that drive long-term value creation.
And we've heard them say, we have excellent functional leaders, they want to spend more of their time thinking about how it impacts the business. And in many cases, changing the perception across the organization as to how these CFOs are perceived in non-finance roles. So helping operators that are on the assembly line, for example, think about driving change, giving them better insights, better tools, better information, different types of reporting, and wrapping that in technological change and AI to ensure that they're on the cutting edge to deliver the best possible information insights to the companies.
And the final two points quickly there's there's something about how to think differently about allocating capital. It goes into the strategic investment decisions that we talked about, but really taking the investor lens to think about how you're spending uh the company's capital to drive ROI across the enterprise. And the last point is continuing to upskill your teams. So we we've seen CFOs that have taken finance folks in the finance function, embedding them across different areas of the business, could be functional areas, could be supply chain, procurement, or business units, so they understand the business differently, but they bring the expertise that is natural to them in the finance organization to make the collective team stronger.
And net net across each of these six dimensions we see real opportunities and real progress for c f o s and finance organizations to really change the way they partner with their management teams to drive performance
And Kevin if I were a c f o or a member of the finance team the notion of contributing to value creation sounds a lot more exciting than the traditional transactional expertise that uh c f o s are known for focusing on so how do you make that transition a reality as a c f o and what are some of the barriers that you're seeing uh in c f o s and their finance organizations in making this shift
probably the biggest challenge is culturally just how does a company historically perceive the role of the c f o and the finance organization I think the best CFOs and I'm thinking of one in particular that was in retail uh he aligned with with his CEO also with the board but he leaned into it so simple things like for example when they're the company was putting together its strategic operating plan he would lean into the business it wouldn't just come from a finance perspective but he would ask um some difficult and challenging questions about key assumptions does this actually make sense are we looking at the market right can we run our stores harder those types of questions typically would not
come from the CFO and as he did that because he was very well researched and very smart he could distance himself from the finance organization and become a better partner so I think changing the cultural dynamic at a company is really the first building block that needs to be addressed
Davide it looks like you have something to add here too
No I mean I I would say that uh and there's also AI would play a very critical role because uh ultimately AI will enable you know uh reduce the effort required to manual work automate most of the transactional activity and really free up time free up capacity within the finance function and deploy this capacity to really play you know a strategic role in driving uh business and and value creation
super thank you both and Kevin it's clear that CFOs and finance teams are increasingly expected to and even seek to create value and reinvent the function yet they still need to deal with the day to day challenges of helping run the business from a financial perspective can you speak to how they balance those two different expectations effectively
Sure the reality is as bold as CFOs want to be to change the way the finance organization operates and therefore how it's perceived across the organization the reality is that there are a lot of challenges a lot of complexity in this day and age that simply don't go away as a CFO is reinventing his or her finance organization and it really in the first instance starts with the complexity that they deal with on a daily basis although this is changing with technological advancements we do see decisions that are often made on inconsistent information or historical data so one of the disconnects as we compare what happens in the finance organization with other parts of the business is that decisions are
based on historic data as opposed to forward-looking data well they still have to run the business as a CFO but the question is can CFOs leverage AI to get better information so it is forward-looking so that they can react to changes in their business model much faster without compromising the day-to-day expectations around performance and that's difficult um the second one is really identifying those opportunities both cost and I'd even say growth opportunities in a way that's consistent that doesn't put the enterprise at risk and how do you embed uh technology to get better information faster I think the the final point I'll say in this first carry area on complexity is also taking a step back and saying what is
the discussion that the CFO wants to have with his CEO his or her s uh executive team the broader organization and really changing the performance dialogue that is happening on a day-to-day basis, monthly basis or quarterly basis, focusing on not just the what, but also why are we doing certain things and should we should we rethink that? And then bringing the discipline that you would expect from a finance organization that is fact-based, hypothesis-driven to help the broader team get to better answers.
The final two points that we see around common CFO pain points is the time and effort spent on regular finance operations. So we've heard countless stories about CFOs that have literally put all the reports that they generate over time on a table to try to figure out who is using this information. In the most extreme case they've cut off most of those reports to see whether they're missed across the broader organization and what that has done is really helped the CFOs take a step back and say what is the information our company and our management team needs to run the business and it may very well not be what has been produced in the ordinary course but it's tough to
do that again because of the volume of the reports and the pace at which technology is impacting the finance organization and the broader business over time and the final point is the financial risk that CFOs are facing on a daily basis and how do you manage that liquidity could be covenant issues could be capital allocation issues could be disruptions seeing a lot of that today and how does that impact the business and how quickly can CFOs adapt to those changes and how quickly can they adopt and utilize technological advancements to get better insights over time
and Kevin if you consider that the CFO usually directs the capex investments across an organization how are the CFOs you work with perhaps justifying their decisions in investing in AI in the finance function itself because you can't make all of the capex investments that you might want to make how do they make the investments in the finance function not come across as self-serving
So maybe I'll just have a quick answer and then it would be great to hear from from the team as well. But I think one of the best stories I've heard from a high-performing CFO is that there's something around educating the broader organization not assuming that folks that are not finance that don't have finance backgrounds understand how value is created. So there's a bit about this is actually how we we create value across the organization and then leading to the next step, which is really the transparency.
If you understand how value is created, then capital is scarce. How do we think about looking at different capital projects and then stacking and racking them so we make the best decisions for the business, as opposed to assuming that there's an endless supply of capital? It's that discipline and rigor that's required to get to the best answer, which is a combination of what happens in finance, but also what happens with the with the other business partners that are looking to uh to allocate capital across their business.
Davide, uh, Andrea, anything you'd add?
No, maybe maybe I would just add that the CFOs have typically the privilege to play a double role when it comes to FI to AI because they can either deploy AI in their own finance function but also they can sponsor and deploy AI to generate value for the full enterprise. CFOs are taking the lead to use AI to decrease non-personnel cost, right? Because or they had of procurement or cost control, and they generated a lot of value, not for their own function, but for the full enterprise.
Or think about also a cash optimization. Again, it's not for the CFO themselves, but it's for the for for the organization.
You know, I'm also intrigued about what this means for the arc of the finance function over time. Many finance organizations may have already outsourced, for example, aspects of their back office or even their front office work to lower cost locations. And are you seeing situations now where the implementation of AI may be changed? The calculation and and the model is now changing in terms of how you staff and where you staff and lead the function as AI and agents are brought in along with other technologies.
Andrea,
I'll maybe start offering that in a number of conversations, this is an active topic. The question from being able to delegate to a certain extent to agents or teams of agents, the execution of specific processes. It's opening the door to a question of what is the right shoring decision for the enterprise for every specific process. Now, I cannot say that I've seen organizations that have yet to go the full way, but at the same time, it's definitely a very active conversation.
Davide, can you talk a little more extensively about what the finance function of the future looks like, and how AI is informing that transformation?
Yeah, thank you. So we looked at this through six domains. So the first one is strategic planning and control, which basically will be transformed for backward looking. So a lot of effort in trying to understand the past, describe the past, lots of reports, data crunching, according to our experience, more than 50% of the time of strategic planning and control is typically spent in creating reports that nobody reads, crunching data, et cetera. So, two forward looking, exactly what Kevin said earlier.
So I've got a bunch of clients that are really building very sophisticated AI models, a chatbot to get real-time answers and automating reports. The second domain would be cash optimization and working capital. So moving away from fragmented processes with a lot of manual efforts in reconciling, for example, invoices with contracts and unreliable cash flow forecasting, to having agents that actually automated the full process of or order to cash or procurement to pay. On cost, cost management and optimization, moving away from cost control, so really trying to understand and doing cost allocation, et cetera, to cost to steering.
So I got a lot of clients, a lot of CFOs that are actually leveraging AI to really steer productivity for the full enterprise.
And Kevin, if you consider that the CFO usually directs the capex investments across an organization, how are the CFOs you work with perhaps justifying their decisions in investing in AI in the finance function itself? Because you can't make all of the capex investments that you might want to make. How do they make the investments in the finance function not come across as self-serving?
So maybe I'll just have a quick answer and then it would be great to hear from from the team as well. But I think one of the best stories I've heard from a high-performing CFO is that there's something around educating the broader organization, not assuming that folks that are not finance that don't have finance backgrounds understand how value is created. So there's a bit about this is actually how we we create value across the organization and then leading to the next step, which is really the transparency.
If you understand how value is created, then capital is scarce. How do we think about looking at different capital projects and then stacking and racking them so we make the best decisions for the business, as opposed to assuming that there's an endless supply of capital? It's that discipline and rigor that's required to get to the best answer, which is a combination of what happens in finance, but also what happens with the with the other business partners that are looking to uh to allocate capital across their business.
Davide, uh, Andrea, anything you'd add?
No, maybe maybe I would just add that the CFOs have typically the privilege to play a double role when it comes to FI to AI because they can either deploy AI in their own finance function but also they can sponsor and deploy AI to generate value for the full enterprise. CFOs are taking the lead to use AI to decrease non-personnel cost, right? Because or they had of procurement or cost control, and they generated a lot of value, not for their own function, but for the full enterprise.
Or think about also a cash optimization. Again, it's not for the CFO themselves, but it's for the for for the organization.
You know, I'm also intrigued about what this means for the arc of the finance function over time. Many finance organizations may have already outsourced, for example, aspects of their back office or even their front office work to lower cost locations. And are you seeing situations now where the implementation of AI may be changed? The calculation and and the model is now changing in terms of how you staff and where you staff and lead the function as AI and agents are brought in along with other technologies.
Andrea,
I'll maybe start offering that in a number of conversations, this is an active topic. The question from being able to delegate to a certain extent to agents or teams of agents, the execution of specific processes. It's opening the door to a question of what is the right shoring decision for the enterprise for every specific process. Now, I cannot say that I've seen organizations that have yet to go the full way, but at the same time, it's definitely a very active conversation.
Davide, can you talk a little more extensively about what the finance function of the future looks like, and how AI is informing that transformation?
Yeah, thank you. So we looked at this through six domains. So the first one is strategic planning and control, which basically will be transformed for backward looking. So a lot of effort in trying to understand the past, describe the past, lots of reports, data crunching, according to our experience, more than 50% of the time of strategic planning and control is typically spent in creating reports that nobody reads, crunching data, et cetera. So, two forward looking, exactly what Kevin said earlier.
So I've got a bunch of clients that are really building very sophisticated AI models, a chatbot to get real-time answers and automating reports. The second domain would be cash optimization and working capital. So moving away from fragmented processes with a lot of manual efforts in reconciling, for example, invoices with contracts and unreliable cash flow forecasting, to having agents that actually automated the full process of or order to cash or procurement to pay. On cost, cost management and optimization, moving away from cost control, so really trying to understand and doing cost allocation, et cetera, to cost to steering.
So I got a lot of clients, a lot of CFOs that are actually leveraging AI to really steer productivity for the full enterprise.
Investor relations from having an experience-based equity story with a lot of efforts to produce and to prepare the meetings for with investors, for example. Now you can have agents and AI to really draft equity story for you, synthesizing investor calls, et cetera. In M A, we'll move from a contingent approach with manual time-consuming research to a programmatic approach where you can basically instantly find the best targets for M A, automate synergy calculation, et cetera. And risk management, moving away from reactive compliance checks to proactive real-time risk detection.
Think about geopolitics, for example, now, which is a big source of risk. So these are the major trends. Now, the reality is that this is happening already now, right? Every year, we actually do a bunch of surveys with hundreds of CFOs. We ask many questions, right? Among this question, one of the questions would be, have you already invested in AI in your finance function? In 2023, 24% of our participants said yes. In 2024, it was 50%, in 2025, 65%.
So it's really growing exponentially. And the second question would be, are you actually using AI? Are you an active user? Or you're just piloting or researching? In 2024, 6% were active user, in 2025, 51%. So seven times increase in finance department. So we're really moving away from piloting thinking to really using AI. And, you know, very quickly on this, what is actually AI in the finance function? It's a combination of analytical AI, so basically crunching data and building models for forecasting, combined with generative AI, right?
So creating new text, synthesizing a lot of material, audio code, think about invoice to contract compliance or conversational QA, and now agentic AI. So really having agents do the work for you in a report generation or automation of invoice to pay or order to cash. The most important part of this, I think, is whenever you do AI, be obsessed with value creation. Every euro invested in AI must return ten times the investment. And Andrea, I'd like to dive a little bit deeper into how these finance domains are being transformed by AI.
Perhaps we could start with the strategic financial planning and control domain.
Now, of course, there's very many ways that companies deal with reporting and have a focus on reporting in their FP A journeys. And, of course, one of the key elements and aspect of that is ensuring performance management towards supporting value creation, as we discussed earlier. And organizations we are seeing are implementing AI differently in FP A. In certain cases, this is taking the form of conversational interfaces. You can imagine the popular chat bots out there, where the focus is in enabling a number of questions of a different degrees of complexity, from the simple what, which is simple but still very valuable because it basically puts in the hands of the finance organization as well as the stakeholders within the business,
a tool and a level of access to the understanding the company financials and KPIs. That is now mediated only through natural language, which creates a meaningful opportunity for teams to save time to be on top of what is going on, whether they are in finance and are more skilled in accessing information or they are not in finance. This is the first level, and this type of tooling is enhanced with analyses that are accessible on giving teams and the stakeholders an understanding of what is the variance, wherever that variance is coming from.
I'm thinking variance, of course, of actuals versus budget or forecast results as they progress through the year versus versus what is the expected landing, as well as, you know, more in-depth root cause analysis of why certain phenomenas are happening, following the appropriate, you know, value trees that every company has. And then getting into the world of what could I do? What are the potential actions? What are the scenarios that we could look at, given given the certain conditions in the data?
And this type of this type of analysis, this type of assistance to decisions have a number of effects. I mentioned the opportunity to democratize access, but it's also about skipping the kind of back and forth of can you run that additional scenario? What does that scenario mean for me? And really being focused when there are the high-power interactions that are review meetings or resource allocation meetings, as we were mentioning earlier, being focused on what are the decisions to be made, what are the scenarios to look at, and can we look at more scenarios?
So this type of tooling and this type of implementation of AI is connecting to a change in the way decisions are made within the companies that are doing that. And the other aspect of that is the other side, but similar, is that these same capabilities, where AI is able to both combine data, but also commentary, as well as tools like root cause analysis or or or forecasting is is at the core also of rethinking reporting, where agents can effectively walk through workflows.
You could imagine them as QA, questions and answers, but they are chained together to build a first draft of a report that then the appropriate person, either in the business or in the FP A team, can review and build on, and most likely also connect with those with those agents and ask for a rerun if there's either something not correct or something that is needs deeper or better analysis. So just to summarize, this only leads to saving time, but also more effectiveness.
And it's it's really two sides of the same coin, different organizations implementing differently, QA on one side or reporting on the other side, but in a way they are related on the way they they actually perform.
And it would be really good to understand how finance organizations are actually implementing this. And are most large finance organizations using internally developed AI, or are they using off-the-shelf solutions? Because finance organizations typically deal in very sensitive data and wouldn't necessarily want to be putting this into an open LLM, for example.
Yeah, let me start. I mean, there's a number of elements that you mentioned there. Um, let me take the easy one first. And security, I mean, these tools are not different from other tools that have been available in the past in terms of security. Every company has their own security frameworks. We have as well. The way we think about it is not, it's very similar to to to the past. The appropriate security needs to be followed.
So, as you say, if the company doesn't allow for data to go public, it should probably stay private, right? And that's one aspect. The more important and interesting aspect is how to implement and think through these tools. I mean, you know, more more um, um, more famous people than me have said that the future, often the future is here, but it's not evenly distributed, right? And and indeed, as you say, many companies are, and we have seen in the data, many companies are implementing this, but not everyone is at the same either speed or the same place.
It depends on their priorities, but also on how they are thinking about this, you know, being at the forefront versus waiting for the tool. What we're seeing for for leaders is that they are combining uh their system and architectural landscape with the capability to build their applications on top. And so combining the the security as well as reliability they get from backbone systems with the ability to build, for example, an agentic reporting layer on top of that in combination of that so that the data comes from the right place and but at the same time is handled and uh um uh managed by the AI uh through through the workflows that suit that specific organization.
Maybe let me also add Andrea that these tools typically need to be trained, right? So so one of my clients, for example, they they built such a conversational chatbot on sitting on their data and they tested it with thousands of questions. And these questions were asked to the business guys. What did you ask, right? In this what, why, now what fashion? Initially, the percentage of questions that were correctly answered was around fifty sixty percent, and then gradually, in a matter of weeks, they reached above 95%, right?
So you need to train the algorithm. In the beginning, you will receive sometimes hallucinations, but after training, it will get better. And typical clients do a parallel running, so they keep their old approach with a parallel running using AI and gradually shifting towards AI when they feel confident. And then a lot of change management. We're going to talk about that. Change management absolutely crucial. How do we think about the way you do strategic planning, budgeting, performance reviews, forecasting, using AI?
Andrea, can you share any specific examples that demonstrate how organizations are embedding agentic AI in the finance function?
The example I wanted to talk about is the budgeting one, where you can have the first draft of a budget put together by a team of agents. Their work orchestrated and in sequence, where you can have the analyst agent that looks at the past budget, what were the variances put together the starting point, of course, getting the data from the right systems and with the right analysis in accordance to how the company thinks about it. Then a market research agent puts together the latest trends, the right understanding of the different products, depending, of course, on the industry of the company, what is the most pertinent.
Then internal operation agent that connects these external the past view with the external view and with the current setup of the the way the company fulfills its mission in terms of either manufacturing or supply chain and so on, and and builds an integrated view. An agent that is in charge of planning scenarios, creating both scenarios that are, um, sensitivities, you know, the high and the low scenario, if you want, as well as more specific scenarios, depending on the instructions that it receives.
A agent that pressure tests that first draft against a competitive landscape, looking to gather information on competitors, uh, or peers, or other market, and thinking through what could be their reactions. And then finally, an agent that puts together these as a in terms of data visualization and creates the opportunity for the analyst as well as the CFO, as well as the other stakeholders to interact with with this flow.
And so, Andrea, is the process that you've just taken us through one that can be standardized for other companies to follow, or does the application need to be bespoke?
Every company will is uh organizing uh this slightly different, so the agent architecture and workflow needs to be adapted. But uh but the opportunity of getting agents to work through a workflow and a sequence of tasks to support a process is definitely there, and it's uh it's coming to different processes uh as we speak.
And what about other finance domains? Can you apply this logic, for example, to cash optimization or working capital as well?
What I just described on processes is very true here as well, and that is why earlier we were discussing about how these processes, which often are distributed across central and shared service centers, are are also being affected. An example of a specific agent, maybe a narrower agent, but it is particularly valuable, and that is focused on thinking through beyond the typical three-way matching in in the procure-to-pay process and really looking at invoice-to-contract compliance. Our benchmarks show that often organizations leave on the table between one and four percent of value in leakage because complex terms, discounts, rebates, or specific hourly rates are not always always reflected in invoices.
And so either they spend a lot of time monitoring this, or they leave they leave value on the table. And one of the real opportunities from AI in this area, in this domain, is really the opportunity to do this at scale, to do this monitoring at scale.
Thank you, Andrea, and Davide. Um, let's now turn to how this could work in a domain like cost control.
Okay, so how as a CFO you can really leverage AI to decrease, for example, non-personnel cost for the full enterprise. So this is a case of a company that was already super efficient, right? It was one of the best efficient in their sector. And the question was, how can I actually getting even more efficient, right? Improve my cost to revenue ratio? And the CFO basically realized that, look, we're actually sitting on oil, but we're not deploying this oil.
The oil is the data, right? And so they basically said, why don't we leverage the data we have from more than fifty data sources, both financial data like invoices, contracts, payments, so not just numbers, but also PDFs and operational data like energy consumption, daily football trips, combine them in one place. And by the way, at McKinsey, we have one capability, which is called Spendscape, that really brings this together. And once the data is in one place, really leveraging AI to very surgically find opportunities to reduce costs.
And they have found more than one hundred leavers, and they they reduced external cost by 9%, nine to 10%. And one example out of this one hundred leaver energy costs, right? And we were in a period where energy costs were to the sky because of the war in Ukraine. And basically, they said, are there ways to use AI to really reduce the wastage of energy? And they basically built an AI model where in each branch they would calculate every day of the year what is the statistically expected energy consumption.
And this is calculated based on climate areas, type of system, it is gas, heat pump, system age, square meters, et cetera. And whenever in a branch the actual energy consumption would go above the black line, the CFO team would get an alert and say, hey, this branch in this city today consumed more energy than what it should be. And maybe there was a very good reason, or maybe there was a bad behavior, like stupid things like, you know, keeping the the the windows open or when the air conditioning is on or whatever.
Only with this lever, out of the one hundred, they saved eleven of energy cost. That were in the matter of hundreds of millions. So that means only with this AI model, they saved tens of millions of euros. So this is one example out of more than one hundred. Now, another case, you can still use AI to support renegotiation with suppliers. So our McKinsey Global Institute has basically built a very sophisticated model that assessed the potential automation of more than two thousand two hundred one hundred activities from vendors.
So basically, the model works that we have 850 occupations, networking, computer system, lawyers, HR external specialists, et cetera. Each occupation is splitted into activities. We have 2010 activities and then skills. And so we can say that, for example, you know, writing code can be automated by agentic AI up to whatever 52%, right? So what is the result? Is that actually there are services that you're currently buying from your suppliers that will be automated to a large extent by agentic AI?
For example, digital technology is already automatable at 11%, and by twenty thirty, it will be at 53%, right? This means that your vendors will have an extra margin in the coming years. And you can go there and say, hey, why don't we share this extra margin and renegotiate our tariffs? Because you're going to do the same thing at a lower cost for you. And this is what most of my clients are doing. For example, client that had renegotiated the tariffs with their IT service providers in help desk.
And basically, they figured out that leveraging automation, assistive generative AI and agentic AI, that suppliers would increase productivity by 60%. And so they went there, hey, we're going to share this, these profits. And so they renegotiated 10% savings right away and fifteen to 20% savings in the future contracts. So this is another example on how you can really leverage AI to bring a lot of value.
Thank you, Davide. These examples are really bringing these opportunities to life. And Andrea, I was wondering if you had anything else you'd like to add. I imagine there are a lot of other applications for this technology. And I know you've done a lot of work applying AI in investor relations, for example, to help remove decision-making bias. Maybe you could tell us a little bit more about that.
Look, the only thing I wanted to add, and you know, thank you, Sean, for for um for mentioning that that you know, we're happy to share and of course, for those interested to connect. But you're very right that one of the meaningful opportunities from AI, for example, in the area of investor relations, but of course in in other areas as well, is to act as um as a debiasing um agent, if you want, or support in preparing for what the market could be asking, it could be challenging to a certain extent.
That is, uh, you know, something that uh, leaders have been doing all the time in normal preparations. But one of the advantages of of AI is that you can effectively load all your previous interactions with the market, public data, as well as your peers, other companies, interactions with the market, and get a very uh crisp and performing agent that that asks you hard-hitting questions in a way, right? What could investors ask? Now? They don't need to be questions that have actually been asked to appear, but rather an evolution that takes into account what you are going through as a company and what others are going through, and really tries to help prepare and offer a debiasing at this biasing angle.
And not only that, this becomes also a um competitive opportunity, competitive intelligence opportunities to hear what are the themes in the market. Again, uh, this is something that can be accessible to people that read and go through, but the AI can be a meaningful accelerator and also the biasing mechanism.
Maybe one other question about board meetings and preparing for board meetings. When you talk about the equity story, what are you seeing in your work with clients in terms of the effect and the benefits that AI can have on the time that executives need to spend preparing for board meetings as well as investor roadshows?
Yeah, look, one one thing I'll I'll say, and o of course, my colleagues may have more to uh to add is that not all it's it's it's two aspects. One is meaningful time saving. Uh, you can get uh uh um a lot of uh data processed. That's one of the major advantages of language models, the ability to to deal with unstructured data, but as well, you can also get an opportunity similar to the one we were discussing when looking at performance management, to create a common base of understanding that is accessible through natural language, which for for many board members that are coming from different, let's call them walks of life, in terms of professional exposure, can become a way
for them to get more a lot more into the details in a way that is productive for the conversation.
Thank you, and Kevin, anything you'd like to add here?
The the only thing I'd say is what we've heard from CFOs is that by leveraging AI, you're seeing what has typically been a team of analysts spending weeks of man hour time, researching for board meetings, understanding what they need to say in an earnings release, looking at what their peer groups uh are doing, really basically doing the manual work around quarterly earnings releases, analyst reports, et cetera. And a lot of that can be automated right now.
So they're going from literally man hours of weeks uh to days of time to get ready for their quarterly earnings, which frees up so much time in the finance organization to think about the value they can create within the four walls of the company, and that has changed rapidly.
That is really impressive. And now let's turn to the other side of this, the hard part. Davide, what are some of the barriers to scaling AI?
Yeah, I mean, number one is waiting for the perfect data, right? I I've got a bunch of clients saying, Okay, I will not do AI until I have the perfect data platform, the perfect data quality. When, in my experience, you'll never get there. I've got clients that have invested hundreds of millions in building the perfect data platform, not getting there. So start with the data you have and then build your data foundations along the way. Second, trying to transform all at once, right?
You've got a bunch of clients saying, Okay, I'm going to do a lot of use case, a lot of pilots, falling into what we call the pilot trap. What we suggest pick one domain. You've seen some examples, strategic planning or cost or accounts payables and receivable, invoice. Pick one and transform it end to end. The third barrier is jumping in without a clear roadmap. What we'd recommend is start by saying, okay, how will the finance function of the future look like?
Have a clear north star, align your your, your stakeholders, both internally and the rest of the organization, the business leaders, the CEOs, on how the finance function of the future look like, and then reversely reverse engineering. What are the initiatives, what are the use cases to get there, as opposed to moving randomly? The fourth would be neglecting change management. This is super important. I mean, technology is becoming more and more a sort of commodity, it's not the biggest barrier.
The bit the most important thing is change management. You can have the best AI models, you can have the best technologies, but if people don't use it, if people don't change the way they work, you're not going to get the value, right? So invest in change management. And the fourth is not fixing fixing the process first, right? There are a bunch of things you can do in the processes, and we described the order to cash, the procurement to pay before embedding AI.
So simplify the process. I'm working right now for a client to really rethink the budgeting process end to end. There's a bunch of things you can do to simplify it before before adopting a technology. And last, again, I repeat, be obsessed with value creation. Eventually, you need to create value in your AI transformation. Think about where the value is, and then start from there.
Indeed, it is all about the value. And maybe Davide, now you can comment on which of these that you just brought us through are the biggest or most common barriers? Is it, is it waiting for that perfect data? And what's the barrier that you would recommend that organizations tackle first?
Well, I I I think that the first thing I would do is to to have a road, a clear roadmap and a vision, right? So build a vision, uh, share with your your stakeholders, understand where the value is, and then, uh, la launch implementation starting from the most impactful domains. So don't think about data, don't think about technology, don't think about governance. Governance is another typical barrier. Or gonna I need to understand, you know, uh, the the how to govern the process?
That's uh, you're going into politics, right? And you're gonna get stuck. Start with a clear vision. Well, not to do is don't start from technology, don't start from governance, don't start from random pilots, right? So start with a with a clear strategy and vision, uh uh uh, and one domain where to create value.
Thank you, Davide, and Kevin, uh, let's wrap this up. What are the core themes CFOs in finance should be thinking about to be successful in their approach to AI implementation in their function?
First one, Davide, you just mentioned uh, setting a clear vision from the perspective of the CFO. We call that the CFO twenty thirty vision, but not beginning with technology, but rather, where do you want to take the business in a very clear and thoughtful way that has some detail behind it? And then, second, what does it take to align the executive team around that vision? Once you get that figured out, the next question is how do you harness technology to make that happen at scale and at the speed that uh, fully capitalizes on what technology provides to the company.
The second point is really translating that vision into a concrete domain approach. So really thinking about the use cases that generate value across the company from a PLL perspective, but also from an enterprise value perspective, and how technology can can help uh your company get to that um that landing spot. The third point is a bit around adopting a domain by domain driven approach to build the data and the technological foundations in a thoughtful way. And that really relates to page uh point four, which is you can't do everything at once.
So thinking about how you sequence this work over time to harness technological change, executing it in waves so that you're generating real value throughout the entire journey without losing time or making mistakes is really important. So investing that time and really putting forth the roadmap that you can execute against that is really robust and and gets the buy-in from the broader company is important. And the fifth one is something around change management. We've talked a little bit about this throughout the conversation, but what we think is for every dollar that you're spending uh as you adopt or implement new technology.
And another five really deals with the upskilling that is required throughout your organization, the change management that needs to happen to ensure that not only executives, but all the way through line management, that folks understand how to utilize technology. So you're getting the most uh that you can out of the technological offerings to really drive value to help your company reach full potential. So that change management aspect is one of the most important elements that we see as a key driver of success, Sean.
These are super useful, and I think provide a great set of principles and starting points for CFOs right now. I have just a couple more quick questions before we conclude. Kevin, I mentioned earlier that you helped found our transformation practice. If you're a CFO looking at implementing AI, but you're also about to embark on a organization-wide transformation, is there anything you'd do differently?
I I think the CEOs that are really bold in thinking about how they want to reinvent their their companies through transformation really start with the strategy. And and I think they feed it through four performance pillars, which start with growth, second is cost optimization, third is your team, so the organization, the upskilling we talked about, and the fourth one has always been digital or technology. But I think technology, in the context of an enterprise-wide transformation, is an accelerant.
So I think it feeds directly into how a company will think about their path to reach full potential, how they leverage data better, and really the speed at which you can get there. So I think you can find that transformations that are thoughtful about how they harness the technology and AI in particular, can shorten the amount of time it takes to really drive performance and change the business in a fundamental fashion. That I think is exciting and relatively new.
Indeed, it is, Kevin, thank you. And for my last question, finance teams can often be viewed as cost centers, even by the finance function itself. So what is the finance team's value creation story that builds excitement around implementing AI in the finance function for value creation, aside from just removing cost? Kevin, maybe you'd like to take this one to bring us home.
I'll start I'll start quickly. I think it goes back to where we started the conversation. The quicker the finance organization, beginning with the CFO and then a team, can be accepted as a true partner of the business, the better it becomes. Because I think what happens then is the finance organization is seen as part of the solution, more forward-looking than historic. And we talked about challenging sessions, things like, for example, are the assumptions right? It's not just a finance question in general, it's how do they partner more effectively with the business?
Thank you, Kevin, Davide, Andrea. Really appreciate you sharing these insights with us today. Thank you again.
Thank you, thank you.
And thank you to all of our listeners for joining us today. We hope you enjoyed the conversation and we welcome your feedback and ideas for future podcasts. Just email us at itsr at mckinsey dot com, which stands for inside the strategy room. You can also share your ratings and reviews on any podcast player, with many thanks to all who've already done so. We really appreciate the comments and feedback we receive every week and encourage you to keep them coming.
If you enjoyed the episode and you'd like to subscribe, just follow our weekly series on any podcast player, where you can also access our entire library of more than 300 episodes. We also offer an inside the strategy room podcast collection page, available at mckinsey dot com slash itsr, where you can easily browse our prior podcasts across nine major themes and access written transcripts of all those conversations. And finally, if you'd like to automatically receive our latest publications and insights, we encourage you to sign up for email alerts on our insights page at mckinsey dot com slash scf, or to connect with us on LinkedIn at the McKinsey Strategy and Corporate Finance Practice page.
Thanks again for listening. We look forward to having you join us again next week inside the strategy room.