#287 Why AI Hesitation Could Cost CFOs More Than Bad Decisions Todd McElhatton COFO, Zuora

Kevin Appleby: And this probably needs investment from the board and so on. How would you suggest to any CFOs listening into this that their journey should be going?

Todd McElhatton: In my conversations with other peers, they're finding business models are evolving ever so quickly. Competitors are going this way and I don't have the tech stack to do that.

Kevin Appleby: How do I know that the stuff that I'm going to do today isn't going to be rendered obsolete in another 6 to 12 months?

Todd McElhatton: The impactful CFOs, I see you're leaning in. You're going to have to take chances. If you sit back and you wait too long, you're going to find yourself in a really bad position.

Kevin Appleby: Grow CFO is where finance leaders grow together. Join thousands of like minded professionals using Grow CFO to access the combined knowledge and experience of the finance leader community. You can join us today. Grow CFO of Net hello and welcome to the Grow CFO Show. I'm your host Kevin Appleby and today I've got with me Todd McElhatton who is the chief operating and financial officer of Zora. And we're going to talk about the increasing pressure that finance leaders are under to reassess their tech stacks and adopt AI. But Todd, welcome to the Grow CFO Show.

Todd McElhatton: Kevin, I appreciate you having me on today.

Kevin Appleby: Todd, before we get into that main subject, tell us a little bit about you because you've been in some fairly big fintech organizations. So give us a quick put at history of you, your career and then maybe a little bit about what the aura do.

Todd McElhatton: So my career, I started back in the day with HP and as I like to tell folks, when I started with hp, getting a job there was kind of like getting a job today at Apple or Google or any of the top tech companies. And I think that's kind of a really interesting learning because although HP's still around, they're certainly not the company that they once were. And I think that certainly speaks to the fact that it's interesting how quickly things can change and how incumbent it is upon management to make sure they're making good decisions. That being said, I spent a lot of time there, went through a lot of different areas, had the opportunity to work in our services business, our software business, also had the opportunity to work outside of the United States and was really fortunate to have some great mentors there, got a bunch of different assignments. I like to say I took a, uh, two year sabbatical and went to a.com company, went to WebMD during the early 2000s and after spending some time there and really getting kind of what I would say, 10 years worth of experience in two, had the opportunity to rejoin HP and a much larger, more significant role where ultimately ended up running about a four and a half billion dollar business unit. That gave me then an opportunity to go fully into software. I followed one of my mentors over to Oracle, had an opportunity to start their original uh, cloud business, the Oracle on Demand. Spent time there, then went over to VMware and started a new business unit there. And I think that was a real interesting experience for me. That business unit did not actually turn out the way everybody had hoped it would and so very quickly we had to re pivot it and ultimately ended up selling that business unit off. And I like to say sometimes things that are supposed to go a certain way and when they don't go that way are some of the best learnings that I had in my career. And ultimately ended up at SAP where I was a CFO for North America and, and then took over as CFO of all the cloud businesses. And that led me to Zora where I joined almost six years ago as a cfo. When we were a public company and we did the take private, I turned around and expanded uh, my responsibilities to be the chief operating officer. And I think that's a really nice hand in glove fit because you certainly see CFOs are impactful today, have a real understanding of how the business operates. What does ZORA do? ZORA is a quote to cash platform. We are helping some of the most dynamic businesses and some of the biggest companies in the world as they see the transformation of their business. Going from a one time product sale to an AI is absolutely disrupting a lot of our customers. How do they catch, how do they monetize those businesses, how do they take care of events that may have been once sold in a subscription, may have been once sold as a product, but are being sold in outcome or consumption based models so we can help companies with their quoting needs, their billing, revenue recognition and collection and cash application. So we really are that layer of helping the business run on that quote, the cash platform.

Kevin Appleby: So how are you finding AI is changing your business model?

Todd McElhatton: I think one of the things that we're seeing is a lot of our customers were on a seat basis and there's a ton of pressure to go out there and re monetize those businesses. Companies are looking for different outcomes, there's different business models coming around and so one of the things that we're seeing is a lot of Our customers or potential customers are finding they don't have the tech stack that allows them to adopt some of these new business models. We're really helping a lot of companies, whether they be current customers or customers that are inbound, figure out how do they monetize these new business models. And it's really fascinating to watch how fast this is occurring and how quickly some businesses are needing to adapt because the business models they had just a year ago that were seemed to be really solid are now looking to be really tentative as people are looking for new ways to uh, consume their products.

Kevin Appleby: What are you seeing happening as folk are reassessing their software stack? Are you seeing AI versions of traditional packages being implemented or are you seeing a lot of AI native solutions suddenly coming on scene?

Todd McElhatton: I would kind of put the software down into maybe two categories. I'm going to say one is there's a category of things, let's call it like workflows, content management, helping people with collaboration. Those are areas that are under crease amount of pressure. And I think quite frankly you're going to see a whole lot of those companies, if that is their main focus, a lot of what they do today is just going to be assumed by AI. And we're seeing the native AI tools really being able to take over a lot of the value that they previously delivered. And so that's causing us here at Zora to constantly reassess our tech stack. And I would sit there and say we've seen a significant weeding of different applications that we've had where we've seen that either some of our other ones have put in native AI tooling that's allowing some of that work to be done, or just able to go to those languages, large language models and, and have them build an agent and do the work and we no longer need the content. And then I think the other area are what I'm going to call systems of records, where there are things that have to be absolutely 100% precise. I think those are going to be less impacted. You're certainly going to see the data on how we interact with that have an impact, certainly going to cause a impact for how quickly our customers can expect us to innovate what it does to cost structures. But I think those are the two areas that I'm seeing that are playing out today in those two areas of software.

Kevin Appleby: We've been running in gross CFO quite a number of webinars with various aspects of AI being discussed. And in quite a few of them we've put Polls up at the beginning asking folk where they are on the AI journey. And 12, 18 months ago, I was expecting people to say, oh, we're looking at this. We're starting to double our toes in the water. But now my expectation is, oh, great, Kevin. We've got some serious projects that we're now starting to implement. However, I'm still seeing a lot of the, oh, uh, we're dabbling our toes in the water type reply. What do you think's holding folk back at the moment? Because there's an awful lot going on here that they can be getting a hold of and adopting.

Todd McElhatton: What I can do is I can speak what we did at zora. I think one of the things that we did at Zora maybe 18 months ago is made a conscious investment and maybe a dozen tools and really wanted to get people comfortable with the experimentation. You said, how do you, go ahead, try this. How do you build an agent? Where can I help you in your everyday work? And, uh, we spent a lot of time helping enabling different teams, making sure we had champions, making sure they had folks that were coaches or mentors to help them use the technology. And I think naturally, by the type of business we are, we have a lot of smart, curious people. They grabbed onto that. They saw the potential things that they had done in the past that were manual, that were repetitive, that they could really automate and really made them much more effective. And so I think where we've moved on as a company is now we've probably got, I would say, maybe five projects that we're working on that really have the ability to change the momentum of the business, whether that be a top line or bottom line. And those are the areas that we're focused on. So I say we are well past the experimentation. We've got our people comfortable with the tools, and you are now seeing them use those tools to really improve their efficiency and help us drive a much faster pace of growth and innovation.

Kevin Appleby: Brilliant approach. But there's a big difference between playing with Claude chatgpt and so on as a chatbot, uh, to actually getting agents that are doing things for you. How have you found that it's been to get people comfortable with actually agents getting the job done?

Todd McElhatton: I think once people can start to see the power of AI and again, it goes back to one of the things that you're able to eliminate some of these manual repetitive tasks. In finance in particular, nobody comes in and says, I'm super excited today about doing a huge reconciliation, but if they can see that there's this tool that maybe something they were spending six or eight hours on is done in 10 or 15 minutes and they're able to validate the accuracy of it. Now you've got someone who's a real advocate for this and they are looking for ways how can they do this? And I think they're also looking for how do we sit there and then move on from things that are less value add to really much more interesting and helping the business meet its objectives. So that's one of the things that we've done and I think people have appreciated it and that's one of the reasons there's been such a good embracing of the technology here at zora.

Kevin Appleby: I like that that says, hang on, what are the boring, repetitive things that I'm going to spend hours doing that I really hate doing? Can I get rid of them? That is the number one use to me of an agent. But at the same time, you're giving a lot of power to these agents coming along. How do we keep this under control?

Todd McElhatton: I think every time we have an agent, there's always a person working with that agent. There's a person that's overseeing the work that's especially important in finance. But that's probably the main key is how do you keep a human involved in it and understand and train that agent just like you would train a person.

Kevin Appleby: That is a repeating theme that I'm hearing from so many people human in the loop. We cannot take the person out of this. The other one that I keep hearing about, Todd, is segregation of duties. How do we make sure that the AI agent can't approve something that it's created and, um, effectively make a load of things happen? As a super agent, an AI agent shouldn't have any more power than an individual human.

Todd McElhatton: And that again, I think just becomes part of the process of how do you work with your internal audit, how do you work with your external auditors to walk through so you can actually see step by step what the agent's doing and be able to go back and have that traceable and validated. And I think that when I talk to other customers, I talk to other CFOs, that's what I see folks are doing. There's a little bit of getting comfortable with it, but people are coming along and they're getting quite comfortable with that.

Kevin Appleby: Brings us to a point that now AI is now so powerful, you can build an awful lot of your finance solution yourself, build agents yourself, you can even build software apps yourself. Where do you think the build it yourself versus buy it decision is going, Todd.

Todd McElhatton: I think where I sit there and look, whether you build it or you buy it, you have the same basic constraints in place. And the reason you think about systems of record, why do people buy that software? Well, first of all, it probably has an integration point into a dozen different systems. So you've got to make sure that you get those integrations correctly. Secondly, you're going to have really deep domain expertise that you need. And so the question is, as a company, do you want to take your engineers and have them working on your product and driving your business, or do you want to have them developing back office agents? You're always going to have tech debt that mounts, you're going to have compliance issues, you have auditability issues. And as you start looking at those and then you take a look at where are you using that software. For, for example, in systems of record, that's like flying an airplane. I have to hit the Runway and land that plane safely. One, uh, hundred percent of the time, one miss doesn't work. And we think, best case, people talk about maybe AI is 95% accurate. And so I start thinking about companies with systems of records. I'll look at Zora, where we're doing billing or revenue recognition and we have customers that have hundreds of thousands or millions of transactions a quarter. So let's just say you have 5 million transactions a quarter. Do you want 5% of your invoices a month? Do you want 5% of those being incorrect? Think about what that does to your revenue recognition. Think that what that does to your collections, Think about what that does to customer happiness and satisfaction. Think about the rework that you've got to do. And so AI is great for a lot of things. There's a lot of things that's helping us in finance, but I think things that need 100% precision is probably an area you're not going to see AI replacing those systems because the amount of effort to do it just isn't going to give you the return that you need.

Kevin Appleby: And rework is remarkably expensive. As a consultant, I've, uh, process mapped things on many occasions and looked at the FTE involved in a process and so on. You nearly always find that all of the effort is in the rework rather than processing the original transaction. And I think if you can use AI to reduce the rework, it's a very big use for it.

Todd McElhatton: Absolutely. But the other thing I would sit there and if I think about rework from a standpoint of how I'm Recognizing my revenue, that becomes potentially a restatement, that is a company's reputation, that is a finance leader's reputation. And so I think it becomes more than just the effort, but it becomes reputational damage. And then, um, can your investors can count on what you're telling them? Do you have the potential for an SEC investigation? And I think those become things that are even more top of mind than just the amount of rework. That's certainly important. But if I sit there and think about it, the worst thing that every CFO or CIO has to contemplate is, oh my goodness, I need to restate results that I've given because I had a weakness and they weren't accurate.

Kevin Appleby: I think you've got to assess the risk around any use of AI. And we put a page in the recent white paper we published about AI uh Native where we looked at governance and control and suggested you should assess tier naught through to tier three. What sort of risk is it where tier zero is? You're having a conversation and doing some research using ChatGPT right through to tier 3 where the AI uh is creating transactions in your finance system. I think it's really important to understand the risk, the potential for things to go wrong and control accordingly.

Todd McElhatton: Absolutely agree with you. And I think one of the things I'd say is one of the leading LLMs has recently chosen ZORA to run all of their revenue recognition. And as you talk to their chief accounting officer, what his statement is, look, we're doing a lot of fantastic things with it, but I take a look at systems of record like my ERP and my revenue recognition. Those are things that I want to go ahead and buy software because they've got the domain expertise, they're keeping up with the things that we talked about like compliance, tech debt. And it just doesn't make sense for me to try to recreate that. And certainly we're going to have AI functionality and we're going to be able to have where those AI tools from wherever they're coming from being able to be used with zora, the ZORA system and it can go outside of the ZORA system. But that zora, what is in the system is a good system of record and you can count on it, you can take it to the bank for being absolutely precise and accurate.

Kevin Appleby: And if you can't properly describe to an auditor where a transaction has come from, why it's happened, and give a full trail of evidence behind it, you've got a problem. So Todd, we're talking about folk that uh, are implementing at the moment. Technology is moving quickly and in fact you mentioned HP earlier and how HP isn't anywhere near the company it ah, was. We've got companies around like Anthropic and so on that perhaps two, three years ago we'd never heard of. If you're putting a crystal ball out, where do you think this is going over the next 12, 18 months?

Todd McElhatton: I don't know that I want to be predicting my crystal ball thing. If I had that crystal ball, I might be doing something else. But one of the things I do think that we're going to see is where I take a look at where ZORA is and I talk to an awful lot of CFOs. I think you're going to see companies running much more efficient. I think you're going to see an awful lot of manual work coming out. I think you're going to see a level of innovation that we have not seen in a long time. And I also think that you're going to see people are able to find that problems that they had that they hadn't been able to solve in a long time, that they're really able to harness this new technology to move much faster than they ever have. I think that's one of the things that we've seen here is like you said, things that we hadn't thought AI would be able to help us on a year ago that we wouldn't even been fathomable are absolutely within our reach. One of the things I've been super excited about is, for example, implementation. The implementation of our system takes a fair amount of time because we're having to bring a lot of data in from our customers. And when we bring that data in, it's oftentimes not in a great shape. And so how do we get it set up and go live? Our software is relatively simple to move forward, really reliant on the data from the customers and what do their processes look like. And um, we believe that we're going to be able to take maybe implementations that were taking six, nine and 12 months down to just a couple months. And we've already started to see huge improvements for that where we can go on day one, because of the information that we have for the 15 years for the transactions, we can go and with AI, go ahead and ingest all of a customer's data and we can give them a preloaded tenant with our software and they can get ready to go to do user testing. And that's like something that's A massive step forward. And how can you implement a technology so much faster and more efficient than you ever were able to before? And so I think as you start to see these things, you're really going to see where technology can really help companies drive some efficiencies that uh, they've never been able to do in the past.

Kevin Appleby: So efficiencies, does that mean fewer people?

Todd McElhatton: I think one of the things that when we're looking at it here, we're a growth company, as we have people that move on to other roles and do other things, we're constantly taking a look at, hey, should we backfill this role or is this a real candidate to take this role and can an agent do it? We've got a couple analyst roles where we sit there and believe that a vast majority of that we're going to have agents be able to do that. Now we also may reshuffle other people's work. And what I see is we're going to have people doing more high value ad work, which for our employees should be more exciting. But I think we're looking at it from a standpoint of when there's people that move through the organization, are there opportunities that we don't need to backfill? And I think as a company that's growing and that's got a fair amount of movement of people anyways, that certainly gives us an opportunity to take a look at what our staffing looks like and as we move forward, how many people do we need to add?

Kevin Appleby: Interesting times that we're finally seeing. Agents can take huge swathes of work out in the back office, huge repetitive tasks. So probably at the uh, bottom end of the finance team, the clerical end, it's likely to make a difference. Does that mean you're probably going to think about more senior people that can do the analytics part of the role, the FP and A roles and so on.

Todd McElhatton: So look, I was talking to one of our customers yesterday, one of the big language models, and one of the things that he was telling me is we absolutely are looking to hire people that are probably a little bit more senior so they understand how the business works and how then you can take what's been done manually and take that and help the business achieve its objectives. So we absolutely are going to be looking for people that understand how companies work and then understand the objectives of the company and help get to the end and help get to where we're trying to drive our businesses. So we're certainly looking for people that understand that and people that have good experience in the business I think are going to be in more demand than ever.

Kevin Appleby: We're really seeing some long term change starting to take place here now. Uh, I think the challenge a lot of CFOs have got right now is to say, okay, I've played around with some agents and so on, but now I've really got to evaluate my whole tech stack to say where am I going? And this probably needs investment from the board and so on. How would you suggest to any CFOs listening into this that uh, their journey should be going to kind of move this up from a gear? A gear from playing with an AI agent and getting a few things done to really refurbishing the tech stack?

Todd McElhatton: Yeah, it's a really interesting question, Kevin, and I think it's super relevant to CFOs today because one of the things in my conversations with other peers is they're finding business models are evolving ever so quickly. And we kind of talked about where we went maybe a decade ago where a lot of things were product based. We went to this recurring revenue model to now there are outcome based models. So we have a company that does manages call centers as a customer and now instead of paying on number of calls, they're paying on the outcome of those or they're getting paid on the outcome of those calls. And so now you're seeing a lot of times where CFOs and finance leaders are like, the market is asking me to price something in this way, competitors are going this way and I don't have the tech stack to do that. And So I think CFOs and finance leaders across the board need to understand not only where is their business today, but where do they expect it to be in two or three years and making sure they're building a tech stack that is agile enough to handle all of those different business models. And we're seeing where oftentimes there are companies that aren't able to handle different business models. I recently talked to a CFO that was telling me about an acquisition that they did. A big part of that acquisition was they were going to be able to bundle the products and services between the two companies. And what they learned is they had huge system limitations that didn't allow them, um, to be able to recognize revenue properly. And so that delayed their ability to bundle that. And So I think CFOs are absolutely going to need to be at the forefront of, of understanding the business strategy, where it's going, and making sure that they have the backend technology that will support those business models. Because I think the last thing any CFO wants to be in a situation is we got a product that's ready to ship, but I don't know how to bill it. And oftentimes as finance teams, we've been super agile and we've been able to figure out how to throw bodies at things and make things work. But the level of speed, as you said, and complexity that's coming, some of those you just can't throw people and bodies at. You really do need the technology today. And so CFOs need to be thinking about that front and center.

Kevin Appleby: I actually wonder if it's that very thing that's holding CFOs back that says, hang on a minute, all of this new stuff is happening. I know I've got to adopt it, I know it's imperative that this is in my business, but this is moving so fast. How do I know that the stuff that I'm going to do today isn't going to be rendered obsolete in another 6 to 12 months?

Todd McElhatton: The impactful CFOs. I see you're leaning in. You're going to have to take chances. Every chance you take everything. You lean in decisions, you got to be agile enough to see when you need to pivot. But I think if you sit back and you wait too long, you're going to find yourself in a really bad position. So waiting and waiting to do something is not the right answer. I think you need to be agile, you need to be thoughtful about the decisions that you're making. But you're going to need to lead in, you're going to need to lead much quicker. And, and I think kind of waiting to say, hey, where do things settle out is probably going to be a really bad outcome for a, uh, CFO or their company. We've got to lean in. I think we're seeing CFOs. I like to say our job isn't to say no. Our job is to understand what does the business need to do and figure out how to say yes and to manage that risk and to understand the different variables and levers in it. And it's becoming more complicated and it's changing more. But that's what we've got to do. We just can't sit and wait for the dust to sell because if we do, we're going to find that entire cycle has passed us.

Kevin Appleby: How about getting approval from the board then for the CFO taking a risk, how are we going to manage to get the sort of ROI information that the board's going to feel comfortable with. If we're jumping in and these risks are around.

Todd McElhatton: At the end of the day, when you're investing in systems and we're talking about what's happening with AI, it's a capital allocation strategy. It's no different if you're investing in people for a product. Going forward, you're investing in equipment or machinery or technology. I think the same things apply. There's areas that you lean in, there's guardrails that you put in, there's governors that you put in. There's hurdles that you're going to want to see. And every decision you're going to make isn't going to be the right decision. I think the question is, how do you continue to monitor those decisions? And when things are going right, you continue to accelerate those. And if things are not going where they want it to be, that you've got natural points to evaluate them, um, you have the proper time to say, this isn't working, let's stop. When you invest in certain AI technologies, I think you've got to be really clear on what your expectation is. When we're giving folks money for tokens to go out and take a look at what they can do with these LLMs, I expect the same return that I would expect on any other capital allocation that I gave them. And so we're going to have to make sure that we can help those teams measure it and also understand what the output that we're getting from those investments are.

Kevin Appleby: Absolutely. So, Todd, we've talked a lot about sort of biggish system implementations to take advantage of AI. On a more personal point, there any really interesting use cases that you, as an individual CFO are applying to for AI?

Todd McElhatton: There's a lot of just things when I do on a daily basis where I'm, um, wanting to get research, I'm wanting to understand what's happening. Where I'm using AI a lot, a lot of times if I get some analysis, I will double check it. I might take a board presentation, run it through and ask, what do you think the board's going to ask? Go feed in what we've done with previous models, help me prepare for different meetings. What are the scenarios that we might expect? What questions might we get based on what's happened from previous board meetings? Are there things that we missed on follow up? So I think it's just helping me be a whole lot more thorough and take advantage of all the data that we have and what we've done in the past to sit there and maybe spot something that I, ah, hadn't thought of or that AI kind of gives me a second set of eyes to look at and it could do it super fast.

Kevin Appleby: That's it. It's the extra set of eyes and super fast. I like that. And I'm coming to the conclusion that the limit of how you can take advantage of extra set of eyes and do it super fast is purely down to your own imagination and your own creativity and what you can think to do.

Todd McElhatton: No, absolutely. And there's a lot of things that when you're using AI and you're thoughtful about it, and I think the one thing that I think some good advice I had on AI is uh, you want to have a healthy skepticism of it. And I think a good example of this is recently someone on our revenue team was looking how do we book a particular transaction. And so it's great use case. Let's go ahead, go into the large language model. Let's get how do we book it? And so came out with something that was absolutely looked right, super compelling, went through step by step, this is what the analysis should be shut down. That's interesting. And said, you know, I think I'm going to run this through another language model and see what, what it says. It came back with literally 180 degree difference. And I think that's another thing you've got to be thinking about when you're using AI is hey, how do you have a healthy skepticism? Just as when I make a decision and if I was talking to you, you're one data point, I would certainly probably triangulate it with some other folks to make sure that that made sense and went forward. And I think that's one of the things that we've also got to be really thoughtful with AI and maybe not get lazy and say, oh, I went to ChatGPT, I went to Anthropic, I got this answer must be right. I think you need to have a healthy skepticism and make sure you're providing the same level and lens of rigor that you would make on a decision, that you're getting something from AI as you would making any decision. And I think that's probably one of the things that does worry me about a CFO is do we have people that maybe get a little bit lazy and they just went out and got one piece of information from AI and assumed that it is 100 accurate? And so I don't know that that's an assumption that we want to do. So I want to make sure that people have healthy skepticism and continue to think. I think that's one of the things that's really important with AI is that people don't lose their ability, their critical thinking ability, their critical judgment, and to make sure that they're validating what's coming out of these systems because they're not going to be 100% right.

Kevin Appleby: And that is an interesting one that you came up with there, getting AI to review AI. And at Gross fo, we've done that once or twice. Actually, only a couple of weeks ago we got a very complex legal agreement in from a potential global customer we're going to be working with. And Dan, our, uh, founder, looked at it with the help of Claude to suggest some changes. And then Dan fired it through to me because we'd been having a discussion about things that we were personally bothered about in this agreement. Dan's revised contract, then into ChatGPT and got another different set of answers. So, yeah, get AI to review AI. If Claude's produced one thing, get ChatGPT to check it, vice versa.

Todd McElhatton: Absolutely. But at the end of the day, the agent's not going to get fired. If we make the wrong call and really have a screw up on how we account for something. That's going to be me who's on the hot seat for that. I can't sit there and delegate my responsibility to an AI language model or agent. I own that. So I need to make sure that I have the proper diligence on those decisions I'm making.

Kevin Appleby: Absolutely. And that's the same all the way through everything we've discussed today. Right. From automating, reconciliations, processing transaction flows, whatever it is, the human has to be responsible for the output. The AI agent can't be fired, as you say, the human can. So you've got to be absolutely sure the AI you're implementing and supervising is right. Todd, that has been a really interesting canter through what's going on in the AI space and the adoption of new technology in the finance teams. And I think that point that CFOs have got to be prepared at this stage to lean in and just get on with it is very relevant. Todd, thank you hugely for being this week's guest on the Grow CFO Show.

Todd McElhatton: I really appreciate the opportunity to speak with you today.

Kevin Appleby: Kevin, brilliant. It.