How CFOs Balance Automation with Human Oversight for Strategic Advantage

Megan - 0:18: Today, my guest is Billy Newman, CFO for Redwood Software. Billy is a chief financial officer with over twenty-five years of experience leading finance teams of high-growth software as a service companies in both private and public company settings and guiding companies through various transactions, including initial public offerings, take-private transactions, M&A, and debt financing. As an executive leadership team member and trusted business partner, Billy is skilled in providing financial leadership and strategic insights to all parts of an organization to drive organic and inorganic growth. Billy, thank you so much for joining me on CFO Weekly. It's great to have you here.

Billy - 1:33: Well, thanks for having me, Megan.

Megan - 1:35: Over the past few years, we've seen automation and AI move from buzzwords to real tangible forces reshaping how finance operates. For many leaders, that shift has redefined everything from forecasting and reporting to decision-making and even talent strategy. So to start us off, I'd like to go back to the beginning of your own journey with this transformation. When was the moment you first realized that automation and AI could fundamentally change the way finance operates? And how did that insight shape your approach as a CFO?

Billy - 02:08: Great question. I'm really excited to be here today and discuss such a top-of-mind topic. I'd say it was back in 2017, so not some time ago. I think AI has become a much bigger thing recently, but we were finding ways to automate processes for some time. Back in 2017, there was a manual process—it was quote-to-cash process, frankly—and it just started to break everywhere. It was very manual, lots of manual touch points, and it was really starting to have a material negative impact on our external and internal customers. The age-old adage, what's gotten us here, which was previously just throwing heads at the problem, was not going to get us to where we wanted to go. We were very fortunate. The company was growing very fast, which is great. But like a lot of companies, when you get sucked into the daily firefighting of everything that's going on—and there was a lot of firefighting because, as I said, the process was broken—it's hard to take that step back and take the time to automate the process. It just got to such a point where we were forced to do that. And so one of the things as CFO today, I really impress upon my team that when you're creating a new process, you really have to take that extra time and effort upfront to try to automate as much of that process while I would say the plane is on the ground, so to speak. Because once that plane gets in the air, it takes a lot more effort to do it while you're up there. And so that's really my takeaway from that and how I've tried to run my teams today.

Megan - 03:44: And before we get into the meat of the conversation, just so the audience has some idea of where you've been, what you've seen. Walk us through your career to date.

Billy - 03:54: So I started with PwC, not in public auditing or anything like that. It was actually for their financial advisory services group. I do have a background in accounting and actually was a CPA at some point, but never did the auditing. I've spent my career primarily on the FP&A side of the house, and I think having that accounting background really helped from an FP&A point of view. But really spent the majority of my career with two SaaS companies. The first one was an online banking and bill payment company called Online Resources. And then the second, the last ten years prior to joining Redwood, is a company called Cvent, which is an event management technology company. That's when I got the opportunity to move into the CFO office. And so that was a great run at two great companies. Cvent, especially, we were public when I joined in 2014. I was taken private by Vista Equity Partners in 2015, became CFO in 2018, survived the COVID pandemic as an event management technology company when  of your revenue is coming from in-person meetings and events and those aren't happening. You can imagine that was tough, but we eventually created a virtual solution and were able to get out of that such that we went public again and then taken private by Blackstone. So I've since moved on to Redwood. I wanted to take my learnings from Cvent and start with a company that frankly was at where Cvent was when I first started there. So I'm trying to do that same play here at Redwood Software.

Megan - 05:23: Alright. Billy, tell us a little bit about Redwood Software.

Billy - 05:27: Sure. Redwood Software is the leader in full stack automation fabric solutions for mission-critical business processes. Think about getting containers off the port to the other side of the world or automating your financial close process across hundreds of entities. These are mission-critical processes that if they don't go smoothly, if they don't go quickly, really bad things happen. And, obviously, when there are so many processes that need to be run, it's really important that you can automate and piece together all those processes into one seamless process. We have the first SaaS-based composable automation platform specifically built for the ERP, and we like to say our mission is to provide lights out automation because we do believe in the transformative power of automation. We empower our clients to orchestrate, manage, and monitor their workflows across any application, service, or server, and we can do that in a cloud environment or on premise. And we allow our clients to do these things with confidence and control.

Megan - 06:35: So once that AI light bulb moment happened, how did it start to show up in your work? In what ways do you see CFOs leveraging AI to improve forecasting accuracy and business agility?

Billy - 06:47: I think starting with the forecast accuracy part, there's a few ways there. Obviously, you can do a lot of predictive analytics. The AI bots can learn over time and figure out how to maybe forecast for different things. I will say, this one, I think, still needs a bit of work. You need to build up enough dataset for it to really be useful. And because of that, if you've got a dataset where there's just a lot of anomalies where it's not consistent, that can make it very difficult to do that. It does still require a fair amount of QA, but I think over time, that can be a very useful thing. It allows you to do a lot more scenario planning and analysis. And I think in today's fast-changing macro and geopolitical environment, that's a big help. Anomaly detection—you can go through datasets very quickly from a forecast perspective, and you can see if there's something that just doesn't look quite right in your forecast. And then there's a lot of real-time data integration that can happen between systems. You're not porting data from one system to the next that could potentially lead to errors. That just really helps in all the things I just said. Similarly, from a business agility perspective, you're allowed to move a lot more quickly. You can create a lot of AI-driven dashboards and alerts that you wouldn't be able to do today to allow for real-time notifications. From a dashboard perspective, it does allow you to create automated reports and insights with very little effort. And so you can get information in the hands of executives a lot more quickly and in ways that they can access that maybe they wouldn't have been able to access in such a real-time manner before. And I think all of this—being able to automate all these things—just frees up the finance team to be able to do much higher value strategic work, which ultimately makes the company more agile.

Megan - 08:39: AI is amazing, and I look forward to seeing where we are in three to five years. But with all the technology that's now embedded in finance, there's always the question of balance. So how do you integrate automation while still maintaining the human oversight, judgment, and accountability that are so critical to sound financial management?

Billy - 08:58: I think the first thing you have to do is just define clear roles, what you can automate, and what still requires human oversight. As much as there are so many things you can automate, there is a limit at some point to what you can do. And so in terms of what things you should be looking to automate, things that are repetitive, rules-based, and ones that really lend themselves to easy monitoring for accuracy and quality. I mentioned before with predictive analytics that you really want to be able to still test to make sure that the outputs that you're getting seem accurate. You need to implement robust governance frameworks. You need to be able to have audit trails of all the processes that are being automated to make sure that if something broke, you can get back to what broke. And also, the more and more you're using automation to do things from an accounting perspective, your auditor is going to want to better understand what's going on there, and so they're going to want to be able to see those audit trails. You want to make sure that the same people, that you've got some segregation of duties there and access controls because you don't want people getting in and changing how these automations work. They might not understand the impact that they might have downstream. So you definitely want to have some checks and balances there. You want to be able to continuously monitor what's going on. Things might work well at first, and so that's great. But, again, you still want some human oversight there to make sure that as the business changes, as processes change, that the automation tool is still working. I like to say that AI agents are like toddlers. They're constantly learning, which is great, but they do need to be watched and monitored to make sure that they're not out breaking things. Similarly, you've got to make sure that you're regularly reviewing all the automation rules that are out there. Again, companies change. They're fast-growing. That's going to change some of those rules that you need to have. And then also, if you have a lot of attrition on your team, there's just a lot of knowledge that can leave, and so you need to make sure that those things fall through the cracks because somebody has left. Well, you need to make sure that you're validating all those things in a timely manner. And then finally, you need to make sure you have the right skills on your team. A lot of people have this misconception that AI is going to replace all these people and that you don't need human oversight, and I don't think that could be any farther from the truth. It's going to transform how people work, but it's not going to eliminate them. And so you need to make sure you have the right skills on the team to just maintain all the different automation tools that you have out there.

Megan - 11:40: And implementing automation at scale is never easy, especially in fast-moving SaaS environments. So what have been some of the biggest challenges that you face when rolling out AI-driven tools or processes, and what lessons have you learned along the way?

Billy - 11:55: I think the first one is we keep coming back to data. The data that these tools need to have clean data in order for them to be successful. And so the data, like I said, needs to be clean. It needs to be labeled in order to be used. I know one of the things that I've unfortunately encountered too many times is that the datasets just don't exist in that way. And so there has to be a lot of effort upfront to make sure that it's clean and that it stays that way. A lot of us are using legacy systems. They weren't built with AI in mind. And so sometimes, in order to automate things, you lose the cost-effectiveness of running with these legacy systems because there's so much work that you have to do upfront just to maintain the systems that it might not make sense to use AI in that world. Data security is a big concern of a lot of people, both within your company and when you're using third parties. So I think that's a challenge that we need to be making sure is always top of mind. We just mentioned what's the right balance of human oversight over these tools. That's more art than science. You kind of have to iterate as you go and learn from mistakes that are made. That's definitely something that you need to make sure that you're working on as you implement these tools. I think one of the biggest things I mentioned, the people aspect of this and the perception that AI replaces people, that leads to uneven adoption of a lot of these tools throughout an organization. I think it makes it key that people understand that AI tools make people more valuable, more engaged because they're working on value-add tasks. And so people shouldn't be caught up in, "Oh, am I working myself out of a job?" It's the opposite. You're actually making your job better because you're not going to have to do a lot of these repetitive tasks. You'll be doing more value-add fun tasks. That comes down to training and onboarding. You've got to make sure that these people that you're trying to sell the dream of AI to, that they're trained, they're onboarded, they understand how the tools work, they understand the why, not just the how. Because if they don't, they're going to fall back into their old ways. You're going to lose the adoption that you get. And then finally, resource constraints. These tools don't build themselves. You're going to need technical resources to help build them. But then also within your team, you need subject matter experts to assist those technical teams. And a lot of cases, those people have day jobs. They still need to close the books. They need to put together forecasts. And so a lot of times, they're kind of doing this outside of their day jobs.

Megan - 14:28: And beyond finance itself, AI is changing how teams collaborate across the business. So how has automation influenced the way your finance team interacts with departments like operations, sales, or customer success?

Billy - 14:41: First off, it makes for faster decision making. You can get information out to your stakeholders a lot more quickly in real time. There can be alerts that you set up. And so there's a lot more pushing of data that's happening as opposed to the pulling of data from the departments. It will result in, because of that, an improved cross-departmental planning. Instead of the data sitting in silos and you're just exchanging it one-on-one with departments, you can make that data visible across the entire organization. And so it really allows different departments to work more closely together because all that data's out there, and it's much easier to digest, and it's all in one place. We see this within the finance team, but it helps because it does cross over to other departments where you're trying to get data from one department into another. Like, for example, the FP&A team works very closely with the HR team, and we use Workday, for example, at Redwood, and we use Adaptive. Fortunately, they're very highly integrated because Adaptive Insights is owned by Workday. We're able to get a lot of the personnel information directly into Adaptive, so there's less impact on the HR team to get data to us and, obviously, less impact on the FP&A team. You know, another example is just using Salesforce and integrating that with our ERP system to make sure that data is moving seamlessly. And I think the offshoot of those things is that it does improve data integrity. It creates a single source of truth. You're spending less time reconciling data with business owners and more time really using that data on the strategic side. There are less manual touches. Like I said, there's not data that has to be shared between one department and another. And as that data is being shared, again, this AI can look for anomalies in the data and can send alerts if it looks like something's a little off. You can't just assume that all the data is being transferred or digested by these automation tools all the time. You still need to have some level of QA there. And so I think all this adds up to finance being viewed as a strategic partner as opposed to maybe a task master like they've been in the past. It really frees up the finance team, like I said, to provide strategic insights and not just reporting numbers to somebody. And when the finance team is positioned in that way, that view of finance now shifts in the eyes of other folks to be value add, which I think if the departments are getting what they want out of us, which is these strategic insights, they're going to be more open to getting the data we need because, obviously, we're dependent upon inputs from them to do our day jobs. That's not their day jobs to be providing data to us. It's their day job to make sales, to build product, to service our customers. And so it does grease the skids a little bit there. It's a win-win for everybody.

Megan - 17:38: Yeah. Couldn't agree more with what you just said. AI is very empowering to a finance organization. So can you share a specific example where automation or AI directly drove measurable efficiency, accuracy, or even revenue growth—something that maybe made you think, "This is the future of finance?"

Billy - 17:58: I'll take you back to the example I gave when we first started. It was that impactful on me. I can give you a little bit more details in terms of what was broken, what the impacts were, what we did, and what the results were. And like I said, it was our quote-to-cash function. It was broken. It was creating a huge drag on our sales and account management teams and was creating a major customer satisfaction issue as well. So across the board, things were not right. And like I said, it was an extremely manual process. Systems weren't integrated. We had Salesforce. We had our internal production system. We had our Oracle ERP data that was literally being manually entered in each of the three systems. It was incredible what was happening. And the data that was being brought in, the opportunities that were being closed-won by the sales team, you're dependent upon the sales team to be providing you clean data so you can get it implemented into our production system correctly and get the contracts set up properly. And there was just no data validation being done when that data was being provided. So it's effectively garbage in, garbage out. But in order to try to clean up the garbage, every contract had to be manually reviewed and, like I said, entered in these three systems. And when you have, and we had tens of thousands of contracts coming in. When you have that kind of volume, it doesn't matter how many people you have and how skilled they are. Things are going to fall through the cracks, and so that's what we were seeing. And so you had invoices that were wrong, but they were taking so long to go out. And so that created a lot of back and forth with our customers because they're saying, "Where's my invoice? This invoice looks wrong." Obviously, that leads to very poor collections, so we're spending a lot of time trying to go out and collect money. At times, it's just a very wonky collections effort because we're not even using the right data in our systems to try to collect. And, ultimately, that then feeds down to the sales executives and the account managers because they're getting pinged by their customers. We're leveraging them to get to the bottom of things. And so those teams were spending way too much time on fixing problems and not selling, which is where I really want them spending their time. And so what we did is, I mentioned, sometimes you get sucked into this firefighting and you don't see the forest for the trees. We temporarily increased heads because we had to create bandwidth within the team so that the subject matter experts could spend time fixing the issues and automating the issues. There were just too many problems that were happening for them to have the time to really do this. And so those individuals mapped out the process end to end. We worked with sales to create that automated data validation checks on the front end. We worked with IT to integrate all the systems together so that the data seamlessly would come in, contracts closed-won. It updates our internal production system, then integrates into our Oracle ERP system. Invoices go out, and there's really nothing that really has to be touched other than we built in some QA along the way to make sure that everything was working. And then, like I said, you've got to train the team. And so we did a lot of heavy training on our finance team to make sure they knew how to care and feed for the tools, became standard operating procedure for when new people were hired that they had to go through this training and they had to actually be tested to make sure that they were doing it correctly. And then finally, we set KPIs to make sure that what we were doing was actually having the effects that we wanted to have. And so we had ten key KPIs that we would report out to sales, to our account managers, to our CEO's meetings. It was that important, and so we wanted visibility for everybody so that everyone's being held accountable. And the result was that that team went from being just the bane of everybody's existence to a couple years later to being just a shining star for that overall process. Sales teams, account managers were really happy with what they were seeing. The amount of interactions they have with their customers from a quote-to-cash perspective went way down. They could focus on the positive sales conversations they're having. Thank god we did this because when COVID hit in 2020, there were so many contract amendments going on. Payment terms were being extended. It would have literally broken the company, I think, if we didn't have this in, and it worked great, which was a great proof point that we had done the right thing. And then from an efficiency perspective, as the company was scaling, we really didn't need to hire any incremental heads for two to three years after we got all those systems in place. The temporary heads that we had to bring on, those were digested into the organization, and we were able to use them to get the systems up and running, but then we didn't need anything from there on out. And so I think the best part, though, was the quote-to-cash team was in a really bad place in 2017. They were just down and out, and they were just beaten up. And getting through this and automating everything and getting that 180-degree shift just really skyrocketed the engagement of the team because they not only were getting great feedback, but the things they were doing were, again, more enjoyable, value-add work as opposed to manual data entry.

Megan - 23:30: That's a great success story, and I'm sure a great example of where people who are afraid of what AI might do to their jobs, they shouldn't be as afraid because it's only making their work more interesting.

Billy - 23:16: Exactly. And they saw it, fortunately. They played a key role. They saw the problems they were having. They wanted to solve the problems, and they saw the benefits that automation could have. And so they really dove headfirst into it, and it's a big source of pride for them today.

Megan - 23:32: And a lot of our listeners might be asking the question of where to start. So how do you evaluate which finance processes are ready for automation versus those that still require human expertise and intuition?

Billy - 23:45: I think first off, you need to find things that are very rules based. These automation tools, AI, you need to create rules for them. If you don't have a process that's based on rules, it makes it very difficult to automate them. So, some examples would be invoice data entry, bank reconciliations, expense approvals where the guidelines are very clear. If you're below this amount, okay, and you get approved. So I think those really lend themselves to automation. And those are also examples of things that have high volume and high frequency. Because if they're really high volume, obviously, the more of those you can automate, the more efficient, the less heads you're going to need, and those tend to be more rules based, like accounts payable processing, payroll processing. There are a lot of just journal entries that are done every month for the same amount, same accounts. You can automate those things. They need to be data driven processes. Again, the automation tools are very heavily dependent upon large datasets. So if you don't have that data and it's not clean, you can't automate the process, frankly. And the process needs to be stable and mature. If it's not stable, if it's not mature, you're still kind of learning how the process needs to work, it's going to result in a lot of garbage, frankly, the garbage in, garbage out saying again, because you just won't be able to trust the output that's coming out of it because you're not certain that the process itself is the right process to actually be performed. Those are things that describe processes that are ready for automation. I think the things that really require human judgment would be anything that involves strategic decision making. You can create data that's used for strategic decision making, but you don't want those actual decisions being made by AI, for example. You really want a human because there are things where the data could lead to give the wrong answer, frankly, because it doesn't have that strategic point of view. If there are a lot of exceptions that come out of a process, again, you're going to want a lot of human judgment involved there to make sure that those exceptions are being dealt with in the right way. There can be exceptions to a process. Like I said, using the Redwood example, if there is something that breaks in a process, if it's a standard sort of thing where it's like, "Hey. I've seen this scenario before. And in this scenario, this is always what needs to be done to be fixed." Alright. Great. You can use AI for that. But if it's complex and there are different decision trees involved, you really want a human stepping in there. And again, talk about the data point. If the data's not structured, if it's not complete, you're just not able to use automation to perform that process.

Megan - 26:32: I think a lot of people see AI as a data security risk. But from a risk perspective, how do you see AI playing a role in actually strengthening compliance controls and financial governance?

Billy - 26:44: I mentioned anomaly detection, scenario planning. The more of those things that you can do, it makes it a lot easier to identify risks and assess those risks. With all the scenario planning that you can do, there are things that maybe you wouldn't have considered before that are business risks that come out of that because you can look at so many different scenarios and, "Oh, I didn't think about this potential scenario happening." That's a big thing that I think you'll see more and more moving forward. That anomaly detection can really be helped for fraud detection and prevention. If you see something different in a pattern, that very well could be an indication that there's some fraud out there. Credit risk evaluation—you can go out in real time and get data from different sources and really assess, "Hey, is there a credit risk here when it comes to vendors, for example?" Something that might have been a much more manual process before you can really automate that and move more quickly when you're assessing vendors, for example. And from an audit and internal control perspective, you can use AI to continuously audit things very similar to fraud detection, anomaly detection, we talked about that. A lot of the tests that you might need a human to do in the past, you can automate that. But I think the biggest benefit from being able to do this from an audit and internal controls perspective is that over time, your external auditors will be able to rely on these tools such that they don't have to do as much testing because they know that, "Hey. This tool is doing it. It's working. I can see the audit log," and so they don't have to do as much sampling themselves because they can rely, if there are appropriate controls over how the tool works, they can rely on the items that come out of those tools.

Megan - 28:32: And last question. But as the pace of automation and AI adoption accelerates, what advice do you have for CFOs who are preparing their teams, not just technically, but culturally, to adapt and thrive in this new era?

Billy - 28:47: I talked about really embracing AI and the value of AI, getting people to understand that AI is not here to replace humans. It's here to transform what we do, not replace what we do. It's going to increase effectiveness. It's going to make our jobs more engaging. It's going to make our jobs frankly more important to an organization, which is all really good for people. I think in order to kick things off, you're going to want AI champions. You're going to want change agents within your team. You'll be surprised, I think, by the people who raise their hands and say, "Hey. This is really interesting. I want to do it." And you really want to leverage those people to push the team to come up with ideas and really get those ideas out there and embrace them once they're made live. But that being said, you don't want people just kind of going all over the place and creating bright, shiny objects. You do want to build a clear vision for the team and a strategy so that the efforts that you're making are you're going after the right things and hopefully the lowest hanging fruit at first to show some quick wins and then move on to the more meaty things from there. I think before you go down this journey, though, you do need to make sure that you've got the right capabilities on the team and address those gaps. Either it's going to be in the form of upskilling your existing team, training those folks, or in some cases, you need to bring in some experts from the outside to make sure that you're making good progress on these things and identifying where there are opportunities for automation. And, ultimately, this is going to result in, I think, changes in organizational structure. You're going to have roles that didn't necessarily exist before. It's going to be important that you consider that as you structure your organization, and that's going to result in different career paths for people on your team as well. And you need to make sure as you do PDPs and stuff like that that you're giving people the opportunities to use AI tools and as they go down their career because this is more and more it's not just going to be people that are accountants or FP&A folks. You're going to need to be able to not only do the accounting and the forecasting, but you're going to need to be able to use tools to think strategically and be that value-add person for the company. And I use the example before—I talked about increasing heads temporarily to be able to focus on AI. That's definitely something people should think about and hopefully, the executive team appreciates that and that, "Hey, there might be a slight cost increase in the short term, but we want to get to that longer-term efficiencies because you need those folks to be focused on building the AI tools." Then I think once you're all done, you need to be measuring and communicating what the impact is. I mentioned the KPIs. You need to create KPIs, be able to set goals with regards to what you want the end state to look like before you implement the tools. And then you want to measure how you're performing against those goals. And whether things are going well or not going well, you want to communicate that out so that you can quickly course-correct if needed or double down on what you're doing and leverage that to inspire people to do more with AI if things are going really well.

Megan - 31:58: Billy, this has been such a fascinating discussion. Thanks for sharing both the strategic and practical sides of leading through technology transformation. Really appreciate it.

Billy - 32:06: Thank you for having me on here, Megan, and I really enjoyed the conversation. Obviously, it's a very timely one. Something's very top of mind for CFOs and just finance folks in general. I think there's a lot more that can be done here, and it's pretty exciting.

Megan - 32:20: Yep. Couldn't agree more. And to all of our listeners, please tune in again next week. And until then, take care.

Outro - 32:28: You've been listening to CFO Weekly presented by Personiv. Please subscribe wherever you get your podcast to hear all of our episodes. Want to learn more? Check out personiv.com. Thanks for listening.