How CFOs can harness AI to transform financial decision-making

AI is here. It's not coming. It's here. And the skills of being a prompt engineer for an AI tool are going to be some of the most powerful skills of the next generation. We're already seeing it. Hello, and welcome. I'm Miles Coulson from EY, host of the EY Better Finance CFO Insights podcast. In this series, we explore the changing dynamics of the business world and what it means for finance leaders today and tomorrow by sharing insights from global leaders on key topics affecting the world of corporate finance. In this episode, you're going to hear from Isaac Heller, CEO and founder of Trillion, an AI powered software accounting company that helps clients streamline accounting workflows. Our discussion explores the transformative power of technology and how finance leaders can realize new value through AI. So let's get started. Isaac, welcome and thanks for joining us. Glad to be here. Thanks. Isaac, to kick off, can you tell us a little bit about your career journey and how you got to where you are today? I'm from the great state of Texas. I studied history. I started my career as a travel agent and then I went to night school and fell in love with accounting, actually a big passion for accounting. And then I had one of those internships where you ended up in a finance and strategy internship. And I was thrust into these roles around finance transformation, you know, Revrack and leasing and pre-IPO at a big travel tech company. And I both loved accounting and finance, but also started implementing it on the ground in terms of reading document systems, building spreadsheets, working with our auditors and our CFO. And I thought there was an opportunity to do some pretty cool innovations. So I'd say in about 2019, I started exploring artificial intelligence a few miles down. It's all the buzz, but that was the early days of artificial intelligence. And I started exploring it and seeing if it could be applied to some of the use cases that I experienced as a finance professional. And fast forward five to six years later, Trillion is one of the top AI-powered accounting companies with nearly a hundred people globally, working with a lot of amazing customers and partners. So that's been my trajectory and it's fun to be here today. Fantastic. We're going to explore a bit more about the AI and the impact on finance later on, but again, a fascinating journey, but grounded in finance and accounting. As we start, I was interested, you obviously got this background in finance and accounting, but you're now a CEO responsible for a rapid growth company building a business. What do you think is really important about the relationship between a CEO and a CFO in terms of the collaboration and what particular skills and attributes are you looking for in an effective CFO? It's funny you say that because working under a CFO is a finance professional, then working as a CEO with a CFO or two completely different perspectives. I couldn't have imagined the relationship I would need to have with a CFO as a CEO. Look, CFO, they're number two, or you could even call them 1B if a CEO is 1A. A CFO is the partner of a CEO. A CFO really protects the blind spot of the CEO. A CFO should be able to say no to a CEO, maybe in a way where a CEO has, I'll have 10 good ideas on any given day. I think they're good, but I have 10 ideas and a good CFO will help me whittle down three and then help me choose one. CFO becomes this extension of the CEO, the operational overlord, let's say, of the company. Quite frankly, over time, I rely a lot more on a good CFO for advice as an entrepreneur. That sort of role as a true business partner in setting strategy, tracking strategy, I think, is in a really important attribute. I think improvised comedy talks about always saying yes and with finance, I think sometimes it's a bit of a no, but maybe no, you can't do it this way, but to your point, here's three other different ways of thinking about it, and it's being involved early enough in the process to be able to shape and influence, not let things get too far progressed. That's really helpful. You've obviously had this, sort of been on this AI journey for six years, as you mentioned, and you've seen the evolution of the finance function. What challenges do you see as being new in this AI world, and which ones sort of endure and have continued to be important for finance leaders to be aware of and thinking about? First thing is, it's dizzying. It started to trickle in early 2000s, and then at the end of 2022, you had chat GPT come out, and I would say since then, it's been like a haymaker punch of the articles and videos. If I'm a CFO, I don't really know what to believe or what's the right stuff. Hopefully, they're getting the best stuff from Miles Yu and the podcast and stuff, but I would say, in terms of the challenges, the challenges that existed yesterday or last week are the same challenges that exist today for a CFO and the team, and I think that's data. I think that data, we still don't have a wrangle full handle on our financial data. It could be in different structures, whether it's documents and systems and spreadsheets, or it could be across different entities. How many times have we done an M&A or a transaction or a new team, and you just have a blind spot of that data. I actually think that AI is helping solve that challenge, but the challenge is more broad than just being some magic wand with AI. We've got to get our data in good shape, whether we're going to do AI, automation, insights. It's the same problem today that it was yesterday. Let us pick up on that point, Isaac, about how do you stay relevant and on top of all of this technology and all this voocal world of disruption that we're facing? One of the things that you do to stay up to speed and stay relevant? I look over the shoulder of the newest team member at Trulion, and I say, "Hey, what's that?" Maybe I'll even ask them, "What show are you watching? What book are you reading? Well, it usually shows now. What app did you download?" I try to stay relevant, just in terms of being in touch with the entry-level professionals at our company. I read a lot. I try not to read too much of the short-form, buzzy articles on LinkedIn or Twitter, so to speak. I try to listen to a little bit more of the long-form podcasts and maybe some books or technical articles for practitioners that are in the space. That lets me absorb it and come up with an opinion of all that. But overall, I think the easiest way is just to look at the youth, so to speak. I don't know if I'm old enough to say that, but look over your shoulder and learn what the next generation of professionals are focused on. That helps the most. A little bit of that reverse mentoring and making sure you're staying connected. I love that. That's really helpful. Go back to this question about the problems finance is trying to solve for. Obviously, you talked about the importance of getting data right, but I think one of the things we've talked about a lot here is, again, really understanding what is the vision for finance organization. What are the problems that you're trying to solve, and how does technology do that? One of the things that we see through the research is this increasing importance for finance leaders to be supporting the value creation agenda and particularly focusing on growth, rather than just some of the traditional responsibilities about cost management, value protection, value optimization. Again, as you think about how finance can step up and play that more value creation role, more of that business partnering, what are some of the strategies that you think finance leaders should be thinking about to be effective? The first thing is I would just reemphasize that finance leaders are in the poll position in terms of taking advantage of AI and new technologies to drive value, meaning we go back to the CFO as the best partner for the CEO, year number one or number two. If you think about AI being applied to all these different areas of business, well, customer support has their own, and engineering has their own, and maybe legal has their own. Each of them have these pockets of AI applied to their little vertical within the business, but ultimately if you're a chief financial officer, you're looking at the broader business. You're looking at an entire P&L of every type of headcount. You're looking at broader models of delivery and pricing and packaging, and so you have perv you into the whole business, and you think about the whole business. The only thing that's limited you from unlocking more value is probably the data, is having all the insights. For example, you could have a customer support line or a support line on your P&L, but if you could actually dig in to their success rate, their response time, all of that data within one place, you could make more operational decisions for the business. The first thing I would say is just double down on the reminder that if you crack this as the finance leader, you can unlock a lot more value across the business. Then the second, which is kind of your question, what strategies would you implement? Well, first, I would be aware of those different areas across the business where they either are or can't implement AI. If a product team is implementing AI for, let's say, a product wiki, you may want to have perv you on that and access to some of the things they're coming up with. How are they allocating different efforts to different features? Ultimately, that could flow into an efficiency conversation or even a revenue allocation conversation, how much I'm making versus costing on all those areas. Once you start to dip into those, hopefully they're already doing a lot of initiatives. For better and worse, I find that the product engineering and even sales teams become more advanced with the AI in the early days, whereas finance, because of the accuracy concerns, is a little bit of a late bloomer. Once you start to get a hold on what they're doing or what they can't implement, then I think you could start to put together a bigger picture for a mocking value. That's really helpful. This question about data quality is obviously a really important one in the AI context. Obviously, a lot of organizations are spending a lot of money on data strategy, data governance, looking at data lakes, other structural approaches. Is there a role that AI can play in actually helping with that iterative improvement in data quality? I think for a lot of organizations, obviously, they spend a lot of time and effort trying to get to this nirvana about clean data and it's expensive and it's time consuming. Obviously, one of the things with AI is the opportunity to drive impact quicker. I'm just interested in your views on that and then maybe also talk about how you see the responsibility for data within organizations evolving. Who do CFOs need to be collaborating with to really come together to drive that data agenda? I'll characterize it in two ways to unlock data. The first is bottoms up and the second is topstone. Bottoms up is how do you get your data into an accessible format? It's in different structures and different systems. That's bottoms up. That's always been a problem. Now, topstone is becoming more and more valuable because with AI, the genetic AI, in theory, you could ask complex questions on the data like a forecasting or a modeling question and the AI could answer it more quickly or more effectively than multiple days or weeks of iterations with an Excel model. That's the topstone. In order to get to the topstone, that genetic AI queering capabilities, you got to go bottoms up. My recommendation, what I've seen work well is best of breed approaches from a bottoms up perspective. The bigger the company, you say data lake, which is the nirvana, is like one lake, but you've got these little ponds and puddles all across the organization. You have to start very, very small. If you're a finance leader, you're probably looking at what are turnkey areas where I could get my data into a structured format. Maybe it's documents. Maybe it's a bunch of spreadsheets. Maybe it's a CRM system. Then once you identify those systems, you should start to bring them into not necessarily the data lake, but start to see how AI can bring them into one format. Just a hammer at home like historically building a data lake or building data warehouse, so to speak, it was a data mapping exercise. I have data element ABC in this, call it file or system. I want to put that in the warehouse ABC, but A needed to map to A and B to B and C to C. AI is changing the game. You can look at a document and turn all that into a structured data format. You can look at multiple spreadsheets and do matching and reconciliation and cleansing to ask if then questions to decide whether to put that data in a spreadsheet. You can't get to the nirvana anytime soon, but to get close, you've got to do those best to breed bottoms up data exercises to get it closer to a lake, let's say. We talked there again around AI sort of connecting the dots across the tech ecosystem within organizations and there's obviously I think this role that AI is one tool in the toolbox. You have to think about it in terms of your overall technology stack as well. Are there other technology advances that you see as being really important as we go on this journey that will be important alongside AI? Wow, it's a good question because there's just so much to catch out with when you talk about AI and specifically things like NLP that are very interesting to me. I think the biggest unlock is going to be agentic AI. Applying agentic principles on top of your data, what does that mean? Being able to deploy agents to answer questions or to automate tasks that were not previously available yesterday or last year. Classically, to put it in a box, you could think about it as RPA. RPA is high-volume process automation. The difference with RPA versus agents is RPA needed that A to A, B to B, C to C mapping, what you would call deterministic in agent theory. Whereas with agents, you can ask them complex questions and force them to reason and find the tools to answer those questions within the data and present them back to you. I think agentic AI is a massive advancement. I know it's already here, but I would say it's got another 10 years to just materialize. That's going to be a co-pilot or a best friend for CFOs in the next decade. Can you explain some of the different areas that you think organizations should be looking to start to think about implementing AI now? Because one of the questions we get a lot is what are the tangible use cases now? Obviously, there's incremental things, but it's important to be putting the foundation now for some of the more transformative things that will come down, particularly with agentic AI, as you mentioned. What would be your thoughts on areas to prioritize now? I'll give two perspectives. One, you're the CFO of an organization and the other is you're the CFO of the finance organization. Of the broad organization, you should absolutely have AI and agents implemented in your customer support, even for complex responses and ticketing that should absolutely be implemented in your IT management. AI and agents should be implemented. That's for your internal support around IT service management. I think that if you have a sales force team, like literally a sales force, account executives and sales development, business development, those teams should be using advanced call recording, advanced coaching, and all of those because you as a CFO could actually get insights into key deals without having to bug the sales team, and that all trickles up. Those are broader operational areas. From a finance perspective, look, you've got to get all of your contracts flowing through a system. Like from a finance perspective, I know there's a legal perspective as well, but whether it's sales orders, revenue contracts, partner contracts, vendor contract, leasing agreements, fixed asset agreements, all of those should be flowing in through an AI-enabled system. All of your sub-ledgers, so if you're doing anything offline, whether it's equity, revenue recognition, lease accounting, all these things, those sub-ledgers should not be in Excel or in legacy systems or point solutions. Those should be in AI-enabled systems. I think that's the place to start because that's the data that is before you go into the general ledger. Even if you want to put AI on your general ledger data, which is certainly feasible, you're going to be limited to the amount of insights you could get out of that because that's the end point, not the beginning point. The beginning point is the contracts, the sub-ledgers, a lot of the broader inputs. As a finance organization, I would jump into that unstructured data specifically related to accounting and finance. As you can tell it from truly, and we do a lot with accounting and technical accounting, and the reason we like that stuff is because we feel like it's compliance related. We feel like it's almost a mandatory investment. When you're fighting for budget, whether it's for, I guess, yourself if you're the CFO or cross-department resources, you need to be compliant and you need to be accurate in your financial reporting. If you can use that budget to unlock more AI innovation in the finance area, that's like a practical win for us with the contracts and the sub-ledgers. You raised the question of budget. I think one of the things we're hearing right now is people are starting to, like all technology adoption, figure out this actually may be a bit more expensive than they thought originally. I think there's a big question about how do you measure return on investment for new technologies where it's unclear what the return, you can measure the I, the investment pretty clearly, but the R may be more difficult. You're obviously talking to customers about this. How do you help clients make the business case for the investments they should be making? I think it's tough. I wish I had a magic wand, but look, AI is unlocking efficiency and insights. That's some of the hardest ROI to quantify. If someone saves me money, I can unlock budget for that. If something makes me more revenue, I can unlock budget for that. I have a business case. I'll get that much incremental revenue. When it comes to AI and incremental revenue, if AI can automate your front office, your strategic areas, let's say you're a manufacturing supply chain or life sciences doing really costly research, if it can implement your core areas, that's a revenue uptick, but otherwise it's really hard. Again, I go back. Look, call me conservative or old school now, but I love compliance. I love financial compliance. I love broader compliance. If you have to invest in compliance and security for your organization, whether it's financial, cyber security, and you're the CFO, you should work with an AI vendor or an AI-enabled partner like Ernst & Young to make sure that within that investment or of compliance and reporting, you're getting a taste of AI because you don't have to, I just don't think that's the kind of budget that you need to spend that money. It's critical. It's a huge risk deterrent, and so I'm really big instead of the big multi-million dollar AI overhaul, let's focus on these compliance and finance automation projects that will deliver incremental value tomorrow with the AI. I think that's really smart. I think it goes back to this point that human behavior with the programs to focus on risk aversion and loss avoidance rather than capturing upside. I think you're right. Obviously, compliance activities give you an opportunity to make the case that people will not want a shortcut where there's risk associated with that. It's a great point. Miles, I'd be interested what you think because I always get this sense that CFOs and finance and accounting professionals are more quantitatively savvy than we give ourselves credit for, meaning if you look at the history of automation, companies like Intuit and SAP, some of the early use cases for tech innovation who are around debits and credits and ledgers and people talk about Excel like it's not technology, that's a piece of technology. It's a beautiful piece of technology, and we've all mastered it in different ways. I don't know, do you seek some sort of renaissance or new wind of energy coming from the finance professionals that it relates to the technology? I love your example of Excel because when people say, "Oh, finance people can't innovate." Well, show me a problem, finance people can't solve with Excel. We've been innovating for a long time with that. But I do think the exciting thing and where the opportunities for finance are is getting more engaged in the commercial activities. You mentioned things like CRM systems, the ability to connect the historic insight that finance executives have from the historic data, put that together with CRM information and drive better, more real-time commercial decision-making, pricing, which customers you need to be focused on. That, I think to me is where finance teams and finance professionals can really start to demonstrate the value they can bring. I think there's some of this concern about, well, if we're going to be automating elements of finance roles, it's the time that gets freed up to do these more interesting value-adding roles. We've talked a lot about the technology side, but I did want to come back and wrap up, talk about the human and talent aspects and this point about creating an innovation mindset in finance. I'm going to joke about the people using Excel, but I think one of the big challenges for finance leaders is do you have enough of that innovation mindset within finance organizations that historically in many cases, particularly around the accounting, have been very focused on compliance, zero tolerance for failure to eliminate risk and focus on controls? We're now asking them to take on a different role. I'm just interested, from your perspective, how do we help bridge that gap for finance professionals to take on more of these expanded roles? I think you encourage and you push, because if you look at the history of tech and tech innovation, AI is here, it's not coming, it's here, and the skills of being a prompt engineer for an AI tool are going to be some of the most powerful skills of the next generation. We're already seeing it. A lot of the tech will play catch up, but the mentality, the people that grew up on apps versus the previous generation, that's already here. I would just encourage and nudge a little. Let me give you an example. I saw all these cool tools in marketing, coming out AI for marketing, content and design and creative videos, and I wanted to encourage our head of marketing to use those with the team. That was a little bit hesitant. What if I say, hey, could you use a little bit more AI? Then there was some sort of, why is Isaac, why is the CEO talking about more AI? Should I be worried? Does he not think we're doing a good job? Is there costs? Those types of things. Eventually, we rolled out some of these initiatives in terms of just how you could use AI within your organization. You would have been shocked at how many people came back, some that I'd never interacted with much and said, "I'm just so happy that this company is encouraging the use of this." There's a lot of people who are coming into the workforce who've been working the past five, seven years that are eager to use more AI, the role they're eager to adopt this. They have no fears. This is how they operate. I think us as finance leaders, we don't want to get stuck in the middle. We don't want to be the one who's trying to please the past and the future. We want to be the one that's preserving the future. In order for us to be the future, to be the leader of the future, I think we embrace that next cohort of prompt engineers and AI enthusiasm as we support it, and then that propels us as leaders. I just think it's as simple as jumping in. It's not going to be revolutionary. We're going to desperately need risk assurance. We're going to desperately need all of these advisory and audit and real strong partners to advise us on those areas, but we're going to need to do it within the context of AI. I think there is just this general encouragement and an urge to jump forward that everyone needs. I guess kicking the butt is one way to summarize. I love that example, and I think what you're describing in terms of tone from the top and actually letting people know it's okay to embrace these technologies and experiment is really important. I think that's one of the key attributes of effective leadership through these disruptions. As we wrap up, Isaac, what recommendations would you have for our listeners as they think about embarking on an AI adoption journey and getting started on this? Well, I think make a list. Let's use a rule of three. Choose three areas of your business where you feel like AI could provide an impact and task your teams with providing a strategy in 2025 or the next six months on how they could implement AI in that area. They should have the choice between working with a best of breed AI vendor or building internally. There's a lot of powerful foundational models and tooling that you can use to build internally. There might be some data concerns. I think that's it. Write those notes of those three vendors, maybe five if you're in a big organization, maybe if you're in a multi-country enterprise, task your CIO with maybe that project for each of his teams, and then track the ROI for the next six months or the next year. Fantastic. As we wrap up, we'd like to close with some rapid fire questions as to get know you a bit better. Is there an all-time favorite quote that you refer back to, and if so, why? I have no special talents. I am only passionately curious. It's an Albert Einstein quote. I saw it on the wall in a freshman year orientation at University of Texas, and I love it because it frames the idea of not assuming that you have any special innate unique qualities except the fact that you want to keep learning, and that's served me well as I've gone through a lot of changes in my career. Fantastic. Anything that encourages curiosity, I think is a good one, so I appreciate you sharing that. As you reflect back on your career and your journey so far, has there been a piece of advice or mentoring that's been most impactful for you? Yeah, definitely. It's simple, and it's get a mentor or get many mentors, and I'll double down because I think that we fall into a bit of a trap nowadays where there's so much content online about all these subjects, whether it's AI or well-being or whatever it is, but there's a lot of wisdom in having one or two or three people that you really trust that you talk to on a monthly basis because they're going to share that similar information with you, but it's going to be tailored to who you are uniquely as a person. I just want to remind people that it's better to have someone who knows you as a mentor than a self-help book that 100,000 people are reading. Get mentors that actually know you and can actually go on the journey with you. That's the best way to grow. That's wonderful advice, and again, it's that sort of comment around you're the sum total of the five people that you spend most time with, and so pick them wisely. That's great. You mentioned well-being in terms of all the stuff that's out there all night. Is there anything you do to maintain your personal well-being and balance? Well, good one. I don't necessarily believe in balance. I think when you strive for this idea of work-life balance, you just end up perpetually disappointed. I have a wonderful wife and four kids, and so that is the core of me, and that's who I am, and as long as I put that first, then everything else fits within the picture, whether it's friends or career or community or anything like that, so I'm very grateful for my wife and family. That gives me whatever balance I'll allow myself. Keeps us all grounded. That's great. Well, Isaac, it's been a wonderful conversation. Really appreciate you sharing so many great insights. I'm sure the listeners will appreciate. Thank you again, and I look forward to speaking to you again sometime in the future. Thanks, Miles. Thanks for your time. Good to see you. If you've enjoyed this or any episode of the EY Better Finance CFO Insights podcast, please leave a rating or review, and don't forget to subscribe so you get access to future episodes. You'll find related links in the show notes or ey.com/betterfinance. As always, thank you for listening, and if you have ideas or topics you'd like to see covered or guests you'd like to see featured, please don't hesitate to reach out. I look forward to speaking to you next time on the EY Better Finance CFO Insights podcast.