AI in Finance: Driving Insights, FP&A & Financial Governance

Speaker A: M welcome to the Financial Executives Edge, a production of the Financial Executives Journal. Here, finance meets bold leadership. Join us for sharp insights and practical strategies to elevate your thinking, drive change, grow your impact and empower your career. This isn't just insight. It's your edge. The Financial Executive's Edge this episode is sponsored by AI Dentified. The nation's top RIAs and wirehouses use AI Dentified to turn relationship data and money in motion events into organic growth without adding headcount. It's AI powered intelligence built for modern finance leaders. Learn more@aidentified.com welcome to the Financial Executive's Edge.

Speaker B: Today's podcast centers around AI enabled finance, driving insights, FPA, uh, and financial governance. I'm, um, Lynn Gargano, your host and moderator with the Financial Executives Networking Group and editor for the Financial Executives Journal. In this episode we explore how AI is redefining both FP&A and the controllership, where FP&A seeks faster insight and strategic impact while accounting safeguards accuracy, compliance and trust. The AI enabled CFO must bring these two worlds together, leveraging AI to accelerate insight while maintaining governance and financial integrity. Ultimately, this conversation brings both perspectives together to explore how the future Ready Finance team is not simply about automation, but intelligent, governed and strategically aligned. Today we are joined by Devin Coombs, founder of Devin Coombs llc, Strategic finance and Technology advisor, and Angela Liu, founder of Gap Savvy Bridging the gap between AI innovation and accounting. Thank you to the both of you for joining us today. So let's start with the big picture. I think we can all agree that AI is rapidly evolving and quite frankly can be overwhelming to probably all of us. But the real story isn't just about new AI tools. It's also about the roles, workflows and responsibilities of finance teams. So, Angela, let me start with you. How do you think AI, uh, is reshaping the role of FP&A accounting and controllership?

Speaker C: Thanks, Lynn. I think that's a great way to start at the bigger picture because at the end of the day we are people and we take on roles. And today I feel like in those functions of FP and A accounting and finance, we often feel like we have to wear many hats and, and I feel like with this innovation we are able to scale and almost consolidate a lot of those types of roles into a more seamless thought process. Maybe I'll start there because the thought process is then led by data. So I think that, um, there are some similarities of what will not change which If I thought about what absolutely won't change, it's our, um, need to take responsibility for the numbers, which I think many of us that have chosen these careers are very good at. Taking responsibility. Responsibility, um, but what will change is the tools that we use, the way we think about the data, the accuracy of the data that comes in, the types of questions that we ask it, and the tools that we use. Um, I think accounting and finance and fpa, there have been many, many, many years where we often were coding in Excel, and I'll use that word coding, because I think it nicely stretches into this new world, which is we were taking our thought process and what wasn't and didn't exist in the applications that we had access to. We were actually building around those applications. We were building and stitching together our human processes. We were stitching together our calculations and thought process in those Excel formulas, and we were sharing them and then summarizing them in decks to the rest of our organizations so that we can make meaning out of the numbers that were produced. I think that that thought process, um, it's almost like we have the tools that work at the speed in which we think and so we can cross more verticals. We don't have to stay as siloed. And I actually think that part of what will make people successful in this new world are those wanderers that never stayed in the silo anyways, were always curious enough to go out to the other pieces, never washed their hands and said, that's not me, I'm accounting, I'm FP&A. Or if I was an FP&A, that's not my fault. It's accuracy. Um, but it's really bridging those worlds, which I think we are really able to do with these new tools.

Speaker B: So how do you build a unified conversation between FPA and the controllership? As you said, Angela, we all know as either a CFO or in finance and accounting, we own the numbers. But there's talk about how sometimes there's a lack of accountability when you start implementing these AI tools. So when you start integrating them into processes and people start working with them, how do you start that unified conversation to make sure everyone is on the same page?

Speaker C: It's an interesting question because I think that unifying is a word that's very popular because it sounds great, but in actual reality with human beings, it's difficult. Right? So how do you get people to think in the same direction and work together? That is actually probably the hardest question, um, versus any kind of accounting or data question. Um, and I think that the answer actually doesn't change much. It's that it must be the tone of the top. Um, I think that when people start playing with these tools, it will maybe initially exacerbate some of the divides in the silos, but it'll also give you the ability to see where the fragments live. Meaning when you start with the adoption, I actually think it's okay for everybody to experiment within their own spaces as long as you have uh, something set up or a mechanism where like on Fridays you all show each other what you built and then you're able to start the conversations from there of like, oh, you're using that, or I'm using this. And I would use the example of I remember being at uh, a couple different companies and like, maybe Sales Ops was using Asana and we were using Jira and then someone else was using Monday and then someone else had an Excel spreadsheet tracker for their to dos. Right, right. Everybody went to the means of the tools that they had access to and now it's like, oh, it's very clear that you're using a different tool or maybe someone else built an app for that. But if you have that system in place and as a leader at the tone of the top, you're saying, well, on Fridays you know, go and explore and build anything you want and then on Fridays let's come together and see what we all built. I think those types of mechanisms and systems in place for you to start stitching together moving in the same direction is actually going to be more meaningful than sitting back and doing a two week design session of how you're going to implement AI. Um, because that way you're learning in the process and you're learning how to build that connective tissue over a period of time as your team trusts each other.

Speaker D: Yeah. Add on to what Angela's saying, because I'm advising multiple clients right now while they adopt AI. And what I think is really fascinating is it's taking a lot longer than you would expect because it's not a technology adoption or financial transformation project. It's a human, ah, project. There's communication, there's leadership. I have to meet with the chief executives and the CEOs and CFOs and get them on board and get them to understand the technology. So there's a lot of executive training in it. And then how do we get their leaders to understand the technology and not be afraid of what the impact might be to their teams? And, and that's really where it all starts. Like I get engaged for implementing AI to these billion dollar companies and a lot of it's just teaching them the foundational models and what they could do and what they could accomplish with it. And then once the leader, like Angela said, the tone of the top has empowered his, his or her leaders to explore and to dive into that, then it starts to unlock. But then it becomes more of an organizational design and a change management project than an AI adoption project. So like learning AI individually can ah, be very powerful. Like maybe I could change 100 hour project to a 15 minute project if I really know what I'm doing. But to scale that across an organization isn't an adoption project, that's an organization change management project. And so it's very interesting to see how that's playing out. But if someone wanted to do it with Angela, you need it, you understand the tools yourself, set the tone at the top and, and then start investing, at least training into your leaders so they can understand and unlock what's possible. And an uh, anecdote, a real example is I spent six months on one of these projects and finally got a big unlock when I was finally able to walk through and have access to the leaders of the senior executive and walk through what the tools could do. And then they finally were able to realize like, oh, I could actually use this on my team for this or that or they could start to visualize it and then they could start implementing it and doing POCs and making changes to their organization. But until the leaders get on board and have that vision, how do you expect the team to do it now?

Speaker B: Absolutely. Well said. So Devin, um, at ah, some of your clients, where are you already seeing places where AI is having a visible impact? So some examples of maybe some AI tools you're implementing or processes around either FPA forecasting, scenario planning.

Speaker D: Yeah. So interestingly enough, uh, I say the AI tools that are having the largest impact are the foundational models. So Claude Code, Claude Cowork Gemini ChatGPT. Um, and really I would say not their chatbot models, I would say you're talking about their agentic AI models which you actually give access to your computers, can build things with with them. Uh, and there's an infinite amount of use cases which make this a bit problematic. Right. Where I've seen, I just got off a phone call with a procurement team who's looking to standardize. They have 1500 onboardings every 90 days. Uh, on um, in Coupa, they're trying to standardize that process and understand the negotiation process and where they could have leverage against their standard terms. And so we're working on how could you streamline that through AI to process all those contracts, store all the, store all that data and then give strategic insights. I ah, could do something like that. I could also. I had another client recently. This is again all, these are all projects from last week. So this is how recent it is of uh, they have their bfo, their business finance officers meetings and they have a deck that's presented, that's created by all these different parties that they have to fill out information, get this deck summarized. And it's very challenging to get everyone in a room to fill that out in a quality way to get it together. And there's a whole role, a whole job, person's job is to make sure that's ready. Like a chief of staff kind of role. Well, we worked on building an AI tool that had forms. You fill out the forms and then the AI will create a beautiful deck in the similar, in the exact same template and fill out any missing information. And now that person doesn't need that chief of staff role or the chief of staff role has now gone to a review capacity rather than a preparation capacity. So 40 hours a week of time saved instead of having to chase down everybody to find out all this information. So like huge sweeping impacts. You could imagine anything from building valuation models to building FPA models, to standardizing and streamlining contracts to anything you can think of can build it. But it's not going to change those higher level executive functions of decision making, responsibility taking. And I've had a meeting over the weekend from senior executives that I know at major top tech companies that are in the AI space. And what's very fascinating is there's a culture of less technology for those people. So I was talking to them over uh, over lunch and a workout, we, we were having a conversation and they essentially said that their leaders have this new um, new mandate of like no laptops during meetings, try to minimize the use of AI and try to minimize the use of technology because what's most valuable in this age is being able to see through the clutter and focus on being able to make decisions. And so it's actually for the leaders, it's being able to cut through the clutter of all this noise. And for the staff and seniors or managers who are using this, how is it to use this effectively to automate work? So I see there's two camps there too that are very fascinating that we can talk about.

Speaker B: So let's maybe take a deeper dive into how do we relook at process and workflows when you're actually automating this within an AI tool. And then after that let's take it to the next level. You've automated it, but from a scenario to evaluation standpoint, right from a driving insights perspective, how is that working with some of your clients?

Speaker C: So I think that when it comes to that decision making portion, it's almost like these tools are very generic in many ways. And I love this idea that keeps now rising more, which is like AI is like the new electricity. Meaning when we think in narrow use cases we're almost missing a lot of the point. And sometimes that's why it's hard to get started because electricity is everywhere. Um, so I'll go back to your question, which is how is this impacting those FP and A decisions, those higher value pieces? And one of the genres that I think AI is very good at is one understanding and seeing lots of relationships in lots of data. So because AI is built off of neural networks and neural networks are the small decisions weighted, um, that as it reasons it goes through that neural network, essentially it's able to see a lot of patterns that you probably in your many years of experience were also able to see. We almost do it so fast we don't even recognize we're doing it. That's exactly how AI is built and exactly what it's doing. So when you get to that high stakes decision level, I think one way that's really nice to use AI is actually as a thought partner. Um, and a really great way to use it is actually to challenge your own assumptions. So ask it to um, argue the opposite perspective. Um, ask it to. I use this prompt a lot which is take a breath and reflect. Is there anything I missed? Is there another data set that would make this more effective? Use it to brainstorm and extend where you're thinking now that might sometimes get to information overload. And I think that's why you have to keep your responsibility hat on really tightly. Um, because I do think that um, there is no one probably in this network that's like I'm going to go totally autopilot on this. And if you are, I would say step away from the ledge. Right. Because the other thing that AI is really good at is it is a probability based completion and it will always sound right. And so with knowing that that's kind of the cap that you live in, I think that it will build an amazing financial forecast. Claude just came out with their, um, Excel plugin. Um, I think ChatGPT came out with it very, very quickly afterwards. Even the cowork functionalities that Devin just talked about literally came out in February. And today we are in March. So for anyone feeling behind, I feel like it's important to also say, it's okay, just start playing. Um, but those are basically the AI tools in our safe space. Like, we know how to use these tools. And so your financial model will just be built faster. Um, Devin and I actually did a recording, um, last week where we just asked it to build out a forecast off of a set of, um, NetSuite journal entry line data. Um, and it asked, you know, here are the assumptions. The assumptions are in blue. You know, like, it'll color code. Because there are certain plugins or skill sets within a lot of these models now. So in Claude skills, which is what they call, um, these preset instructions that you can build exactly like you would with a staff person, you can stack those types of thinking, whether they're skills or like, maybe if you're building a forecaster evaluation model. Give me the analyst view as if you're from this bank, right? Give me the view as if you are an SEC reviewer. Give me the view. Right? Like, you are almost creating this advisory board of geniuses from different perspectives that will poke holes and see things that you might not see under your cap, but you also probably know the edges of, uh, from your experience. And I feel like that that is a really, really powerful way at that level. And again, there's so many different phases of it, but this is where I feel like it's at that review really stretching your thinking again, not turning off your thinking, but almost helping you think harder and deeper, um, so that you really can get those rich insights. And often. And I find this all the time, actually, it will find an insight that I actually didn't think about. And then I might dig down that hole and I might reevaluate it and I might find that I need to plug in a different value set. Do I need to connect it with my Salesforce data? Did I actually have a sub ledger that I didn't have that line detail at? Um, I think that, that if I put that all into the construction is kind of context engineering in a way, which is you're bringing in the relevant data sources to make some of these decisions, but you're also diving in and stepping back at the right moments so that you're getting that clarity for the numbers, which would typically take you tons of time to judge. And now you're able to just make quicker decisions on how to layer in and weave together that, that thought process.

Speaker D: And everything Angela said is dead on. And practically I've been helping this over the last year. A bunch of financial due diligence clients and partnering with P and A teams so that they can figure out if they're going to acquire entities. What would it look like pro forma? What would uh. How could I go and help decide if this is a good purchase or not? Earnings quality is. And a lot of these times these are private companies that don't have the best audited financial statements. So there's a lot of quality gaps. And I'd say AI can be fascinating and amazing at if you pull in all the information and ask it to run quality of earnings analysis, uh, risk risk, uh, and uh, risk analysis or um, technical accounting analysis, whatever you want to do. But put all the hats that Angela said you'll get a bunch of robust documentation that is not. When people say hallucination risk generally I've seen it tied to the T that it like matches the numbers dead on. The problem now becomes I have too much information where there would be no way. This would have taken uh, a big four or consulting company hundreds of hours of work to generate a comparable report that you can now get in 15 minutes. So you can kind of let that go endlessly. At the end of the day your CIO, CFO, senior executives, they can't read 100,000 different pages of reports that you can now generate. So how do you pull through all the noise and actually make decisions? How can you take responsibility for. Oh, I understand now all of these risks I wasn't aware of that maybe I would have never been aware of because I wouldn't have the time to investigate them. But is this still a good purchase? Is this still a good acquisition for the business regardless? So that executive functioning, that executive decision making I think becomes more important because you're going to have more information funneled at you which is really fascinating. It's a fascinating paradox where you actually need less lower staff people. Everyone needs to be more up leveled to be able to make decisions on what data is relevant versus not.

Speaker B: No, absolutely. And you're highlighting the importance of judgment and accountability and making sure that you're applying your expertise. So I love Angela, the way you said apply your expertise, review and endeavin and use your judgment but also stress test your judgment, challenge yourself to say let me make sure I'm not missing anything. So I love the fact that it could be used both ways. Now Devin, you mentioned hallucination, and that's a common word that we keep hearing with AI and, and it happens more often than we think. And that's where the judgment steps in. But how do you make sure you minimize those hallucinations? It sounds like when you set a lot of your clients, when you're working with them, most of the responses are dead on. Is that because you've worked with them on the process, the structure, the data scrubbing before that, so you don't have bad data in and it's kind of tight controls of how it's kind of running some of these scenarios and applications?

Speaker D: Yeah. So I'll say what my process is and I bet Angela has a great perspective as well. But you'll find the funny thing is like, numbers are very telling. And so the accountant in me, the CPA in me prior, before my finance days, loves the fact that if you feed it certain context, like hey, here's what the asset, total asset should be, here's what income should be, your, uh, trailing 12 months, here's what the valuation multiple should be for your enterprise value, whatever you feed it, I can tie those numbers and make sure that it's accurate. And what I, what I've noticed, especially with Claude and lately OpenAI, but mostly Claude, it makes sure it's using dead on accurate numbers, which is a major change over the last few years. At the beginning, when I started playing with this, I would never do it. The numbers always tie and the reasoning around the numbers always looks correct when I feed it the context. So checking for facts. Now here's an interesting thing and here's the nefarious problem is if you don't feed it the right context, it will still give you facts, things that look like facts that aren't facts. And so for a good example, I might make a report like that where I put very detailed financial due diligence, mock numbers up, uh, and it will pull the exact numbers and it'll look very authoritative. But then it might make up a hallucination of like, let's say I'm buying a CPA company and it will, and I don't tell it that how many CPAs are in the market. But I want that data point. It might make up that percentage and be like 50% of CPAs have worked at big four or something like that. So now realize how dangerous that is. So now you have a report that is 90% dead on, 10% hallucinated. And our natural biases, especially as uh, decision makers, executives, professionals are going to be, wow, this looks really professional, legitimate. And you're going to accept the report dead on. So there's a real danger there. And I'd say anytime you're actually asserting any fact with AI, you should have some kind of checking mechanism built into it.

Speaker C: I have many thoughts on this hallucination thing, and I actually want to start quickly with what is a hallucination, because I actually think that that also helps as you, um, engineer that context. So, um, the models are trained in a way where they have to meet goals. They're trained in, here's your goal, do you meet it? And they are evaluated basically on a binary of, you met the goal, you didn't meet the goal, you met the goal, you didn't meet the goal, you met the goal, you didn't meet the goal. What does that mean? They're trained to be incredibly helpful and they're trained to meet the goal. Nowhere in that is saying, I don't know. So it is incredibly rare, almost uncommon for a model to return, I don't know. Which is very different than a human, which some of us are like, I really don't. I don't have the knowledge in my base to know that you can give guardrails in your prompts to say, if you don't know, return to me that you don't know. That would help it meet its goal, right? Because that's its goal. Similarly, um, you can also prompt it. One of my favorite prompt hacks is, um, write whatever you're asking for, write whatever your goal is, and say, ask me any clarifying questions, because what is it doing? As it's moving through the neural net, it's making little tiny decisions and reasoning to figure out which way, in its context to dig and provide that next answer. So instead of making those assumptions and jumping through that neural net, instead it is saying, okay, I'm going to give you the chance to fill in the holes. And now that you've told me these things and this is what you want, do you want public and private companies, do you want me to focus on this? Do you want me to work under ASC606 or a specific standard? Do you want me to take a different perspective? Um, you give it a much higher chance to increase its accuracy. Um, and you'll see that, like Devin said, the models are actually getting significantly, significantly better. Because as they're training them, they are making them more accurate. Specifically anthropics. Uh, Opus 4.6 was kind of mind blowing because of Its accuracy and its nuance. Um, I'm sure that today we will probably see the worst model we will ever see, which is wild, but that's true because it's just going to get better on two fronts. One, it's working on its own accuracy in terms of how many loops is it jumping through in that neural net in its reasoning steps. Two, what is your responsibility as a prompter? And then many of the models actually, and I think Cloud started this, I've been seeing ChatGPT do this as well. But like they will actually prompt you with those clarifying questions without you asking now. Like you'll ask it to do something and it'll say, did you mean this, this or this? And they make it very easy. You're like, I meant one and two. And so in a way we are all constantly honing in on the accuracy. Um, so there are many things that you can do in the prompting, many things you can do in understanding just how the model works and how it's trained. Um, and then I'm going to wrap around with exactly what Devin said, which is it is totally your responsibility. The buck stops at you.

Speaker B: Absolutely. Well said. So we talked about FPA evaluation, scenario modeling. Uh, where have you also seen, uh, AI's visible impact in other types of accounting functions, whether it be accounts payable, AR, the close process reconciliations, um, or even accounting research.

Speaker D: Yes, it does everything right. I, I advise a, uh, company called Tabot that's primarily a technical accounting writing software that is trained against, uh, trained against the guidance specifically rather than just a general learning model. So it has very low hallucination rates. But we've seen, we've seen pros and cons that everything we've discussed here, I've seen it with all these models where I've seen what used to take hundreds of hours of research and putting together very technical documentations that was really limited for accountants who went to national office or very specific types of accountants called technical accountants who were focused on filings with the SEC and the PCAOB and making sure that it could get through the top echelons of partner review. Um, I've seen that kind of skill set now be brought into. Everyone can go make a mellow a technical memorandum around very complex transactions, um, pretty quickly. And there's a very pro, there's a pro to that. It saves hundreds of hours of preparation time and staff time where you don't have to write these technical memos anymore. But there's a con where it looks really good and then it goes to audit review and the auditors decide that it's completely trash. Like you didn't have the judgment or expertise to make the decision that that memo was good or not. You bring Angela, someone like Angela or me into the room or anyone who has the experience, they could tell you that and polish it. But there is a very large risk for accounting and finance leaders, or leaders in general, where you hire people who use these tools that look legitimate but actually have no clue what they're doing where you need that executive judgment. So I am um, seeing a skill gap there. There used to be Intelligize, there still is Intelligize where you could look at SEC comparable SEC reports or public M information and get competitive analysis on how they're their disclosures and SEC preparation. Well now any of these AI models have access to the SEC's database because it's public information and you can do any of that research within, within the AI models. So questions on IntelliJ. IntelliGize has some extra functionality, but it's putting companies like that really I think at a different risk because anyone can now go and do the research. Pretty, pretty with pretty common language. Um, and then something, something as simple as summarizing AP journal entries into a new. Into your netsuite, uh, uh, templates from finance, doing a cash flow projection to just teaching new junior team members on what your expense policy are because you have an expense policy chatbot. We've seen all of those use cases that are all completely relevant and uh, it is disrupting the industry on multiple levels.

Speaker C: Devin and I squirrel hole um, a lot about this on our podcast. Um, and I'm so glad we're doing this series because of that. Um, I'll add on with a couple of other use cases. Techno accounting like Devin said, I wrote that um, paper on context engineering specifically and breaking down a thought process. Um, I do think maybe just one little hint hack there is that, um, a lot of us think a lot that have gone into technical accounting and research. Um, I find it very helpful actually to verbalize that with any of the tools and then ask it to see the patterns in the way that I think. And then that becomes a skill or a framework that I feed it back into it so it can duplicate the way that I think. Um, that is a really helpful tool in those types of spaces. Um, for like apar. Um, I'm finding reconciliations. So subledger general ledger. Um, right. Sometimes you have maybe three entries that make up one je line. It's able to, you know, we kept trying to create that logic that said try to find things that sum up to this one line. And I know I went super deep into that, but it's like giving instructions like that. But even today you can even just say reconcile. And I think it'll do it on its own because it now knows what reconcile means. Um, so, um, uh, within AP ar, like matching PO numbers to invoices. I've seen that one quite a bit. Um, collectibility, uh, with ar, um, credit checks pulling in different data sources, then in close. Right. I've lately I've been seeing people create their own closed checklists. Actually, one other really fun just mechanism of using AI that I learned recently is actually, um, if you like, for example, built in an app on your closed, um, checklist and then you have a meeting to go through your closed checklist with your team to actually record that meeting, use it as a transcript, use that transcript to feed back into your tool so that it updates what's been done. I'm like, oh my gosh, meetings might be useful again. Um, and so, uh, those are some examples. Um, because I think that one thing that is very useful to remember is that this is a very, very generic tool and our use cases are simply to help each other stretch our imagination.

Speaker D: I'm going to add on to what Angela said real quick. Like that last hack is so helpful, where I can't tell you how many meetings I go in with the executive teams now and I'll show them something, I'll be talking about something and they'll say, oh, the classic executive. Like, I wish this was the case, this is the case, this is the case and this needs to get fixed. Get it done. And when I was chief of staff, Google Cloud, that was like, man, that would be all month just like working through decks and fixing it up to their exact specifications. Now you can literally just take that transcript, take the action items against the deck, use the right tool and say, make these enhancements to the deck, or make these enhancements be done in five minutes. Right? That's completely been automated away and it makes something really impressive. So, real hack, huh? To give it the right context that you need.

Speaker B: I mean it's clear that AI is the potential to just reshape how finance operates. For a finance leader. If you're a finance leader and you could only implement one AI capability within the next 6 to 12 months, what would you choose? Um, and what, what, what capability or function do you think it would have the greatest impact in? So again, final thoughts.

Speaker D: You need someone with deep domain expertise that also understands this technology. And it just so happens to be us, uh, people kind of on the education and the corners who are academic plus consulting plus community building, who are willing to experiment and take risks and curiosity driven. You need someone like that in your corner to educate your team and from an executive education viewpoint you need to hire someone in your corner and you just need the basic tools you don't need. Uh, there's some great fine tuned tools like Tabot Advisor 4, many other great tools out there that are awesome numeric. Uh, I mean there's tons of great people that I've worked with that I really respect. Uh, but you shouldn't worry about any of that right now. You should worry about what is your team equipped to understand agentic AI so that they can get the best use cases off the General Enterprise Tools. ChatGPT, Gemini or Claude. Choose your tech stack and then go from there. If you haven't done that, you should do it today or you're going to be at a competitive disadvantage to all of your peers that are doing it. Not a question.

Speaker C: If I was to implement one tool today, it would be Anthropics Claude Claude. Cowork. Um, so I'm going to go straight on a specific tool and again this is March of 2026 and this changes rapidly so don't hold me to that. But today I feel like that is the easiest, best model Today in Excel there are finance plugins that work really well, kind of pre made out of the box. Um, it has all of the general functionality so that you can understand that it can summarize that it could work with data, that it could be a thought partner, that it can look at snapshots, that it can create presentations. And I feel like playing with that as a team in this safe environment where you set the tone at the top, where everyone admits that it's totally new and nobody is an expert necessarily at it, right? And that we are learning together. Plug into a community like appsavvy or have a training person, right, and explore there because your ability to automate the work that you're doing manually today is a wonderful place to start, right? You will learn so that you can build out that bigger picture. But getting started is more important than anything and getting started on one of the tools is more important than anything. I did happen to pick Anthropics Claude today because it is just moving mind blowingly fast specifically for the finance and accounting world, um, which I have been delightfully surprised to see. And again, I don't get anything from Anthropic. So I'm saying that just out of my last couple months of playing with it. Um, and specifically because Claude Cowork is their version of Claude code without having to go into a coding terminal. So the power to update things on your desktop to build power Excel presentations or Excel spreadsheets that feed into PowerPoint presentations, just kind of that bread and butter stuff that we do or the spaces that we work in, I feel like is a wonderful, wonderful training ground for your team to build that curiosity. I've also heard of some teams if you don't just go with one. I thought this model was really great, which is do two weeks of Claude, do two weeks of ChatGPT, do two weeks of Gemini, do two weeks of cowork or sorry, Copilot. Um, and, and play around with those so you know the general functionality. You will be better at buying all of those specific tools afterwards because you will really appreciate how people pulled together those workflows, how they refined and did the evals on the accuracy of all those prompts. Um, and you will be better at architecting those spaces in between while something's sitting on the roadmap. But you need to close. Um, so that would be my best advice, at least today in March of 2026.

Speaker B: No, that's. That's great advice. A great conversation. And that brings us to the end of, um, the AI Enabled Finance Driving Insights, FPA and Financial Governance Podcast. A huge thank you, uh, Angela and Devin, really, for providing, uh, some wonderful insights here that it'll be quite, um, helpful to our listeners. Um, it's clear from today's discussion that AI can accelerate and streamline processes, but the real advantage comes from when finance leaders turn that intelligence into accountable, strategic and auditable decisions. So if you enjoyed today's conversation, be on the lookout for the AI Enabled CFO Financial Executives Workshop series where we will continue the discussion covering the full finance landscape with practical use cases and candid conversations. Thanks for joining us. Uh, until next time, stay informed.

Speaker A: This episode is sponsored by AI denified. The nation's top RIAs and wirehouses use AI dentified to turn relationship data and money in motion events into organic growth without adding headcount. It's AI powered intelligence built for modern finance leaders. Learn more@, uh, aidentified.com thanks for listening to the Financial Executive's Edge. If today's episode sparked new ideas or help sharpen your perspective, be sure to follow and review us on your favorite podcast platform. You can also visit financialexecutivesjournal.com for more insights, articles and upcoming episodes. Until next time, stay sharp, stay strategic and maintain your edge. The financial executive's edge.