Private Equity AI ROI in the Office of the CFO

Lloyd Metz: [00:00:00] Get ready to peer behind the curtain of the private equity universe with each episode of Best But Never Final. Hi, I'm Lloyd Metz, joined by Doug McCormick and Sean Mooney. Together, we'll navigate the corridors of private equity, revealing the uncommon knowledge, challenges, successes, and lessons that drive the world of private equity and business forward.

Let's go.

Sean Mooney: It is great to be back with Lloyd and Doug. Doug and Lloyd, how are you guys?

Doug McCormick: Great. Good to be here.

Lloyd Metz: Hey, Sean. Hey, Doug. Good to see you guys. It's been a minute.

Doug McCormick: It has. It has.

Sean Mooney: I've missed you guys.

Lloyd Metz: I did too, actually, low-key.

Doug McCormick: I'm not going to go there.

Sean Mooney: Yeah.

All right. What was really interesting, I think the last one, we got a lot of compliments on the episode, [00:01:00] because it was like setting the stage for the things that you can do on AI, and to be candid, guys, I think there's AI exhaustion, but there's also AI resistance and kind of failure to act going on. I think it would still be good to drill down on this topic a little more, but really, like, how do you practically do it and get ROI versus everyone having fancy Google?

Lloyd Metz: Absolutely.

Doug McCormick: The concept of AI exhaustion doesn't resonate with me. I think this is going to be front and center here for a long time, so if you're exhausted, get ready. It's going to get worse.

Sean Mooney: Ding, ding. Guess what's going to win? Change. You're 100% spot on. If we all run towards it, the world will be our oyster. If we resist it and ostrich it, change will win.

I don't know about you guys, but I don't want to get left behind.

Doug McCormick: FOMO.

Sean Mooney: All right. So one of the things that I thought would be helpful, and we talked about this beforehand, is we happen to have a wizard of AI enablement within the office of the CFO in our midst here at BluWave. And so one of the biggest areas that we see just through equipping the PE [00:02:00] industry with AI enablers and tools is one of the areas of highest ROI for sure, in terms of things you can do right now and get payback, is in the office of the CFO.

So I invited our CFO, Jeff Berry, to join us here today to pull back the curtain and just actually say, "Here's what we're doing, and you can too." So Jeff, thanks for joining.

Jeff Berry: Thanks for having me.

Sean Mooney: So Jeff, you want to share just a little bit on your background, and then we'll jump in.

Jeff Berry: Sure. I'm a former public equity investor turned operator.

Been in finance leadership seats in middle market companies for the last 8 years or so, and have gone from having perfect public market, SEC-approved data to very messy data inside growth stage companies. So very excited for all the unlocks that AI is is now bringing.

Sean Mooney: And what I like about Jeff, he went to a football college. It's called Vanderbilt University. Many people didn't know it was a football college, but it is now. And then Doug and Lloyd, like closest to their hearts, because I always see them with their rings on and their flags behind them, [00:03:00] he went to this little school right outside of Boston that you guys went to as well.

I try not to hold that against him. I know that's going to mean a lot to Doug and Lloyd.

Jeff Berry: Try to keep that part of my background secret.

Doug McCormick: What do you ever do with that, Lloyd? Where, where do you go?

Lloyd Metz: I do nothing with it. I just let it go. I let it wash over me.

Sean Mooney: They're not taking the bait. I'm poking them on, I'm poking them... sometimes they don't take the bait, but it's okay. They resist it. They're learning to resist my pokes. So I play the everyday guy, man of the people here role- ... in this conversation here, and that's where I feel best.

Doug McCormick: There's nothing about you that speaks man of the people.

Lloyd Metz: Jeff it's really good to finally meet you.

I've heard Sean rant and rave about you for the better part of a year, so I'm really excited for this conversation.

Jeff Berry: Great to be here.

Sean Mooney: So Doug, why don't you kick us off? How do you want to take this?

Doug McCormick: I think this is a super interesting conversation for a couple reasons. In my little ecosystem, 10 portfolio companies, I often feel like the CFO is in this interesting place of being the traffic cop between AI initiatives that the private equity sponsor wants to [00:04:00] initiate, and trying to figure out how to access and prioritize in the company.

And I think that's probably because people in your seat are generally pretty tech-savvy. You sit in the very data-intensive part of the business. As you think about your own ecosystem, it feels to me like there's a foundational element for everything that you accomplish that can then be translated across the organization.

In that crazy world of, you're at that critical juncture in something that's happening so quickly, I'm just interested in hearing your thoughts on what's changing, where are you finding opportunities, and, if you could start with maybe a little bit of a conversation around: do you have a framework that you're using as you think about prioritizing the things you want to accomplish in this area?

Jeff Berry: Yeah, it's a great question, and I think what doesn't change with AI is the sort of same North Star as before AI, which is: you have to have good data, you have to have good processes, and you've got to be able to understand working backwards what you want to know about your business daily, weekly, monthly.

And [00:05:00] once you figure out those things, that's what AI can really supercharge. And so, when I think about enablers of AI in finance and accounting functions, I think about five things. First, I think about data and the data structure. Two, I think about understanding your processes. If you have bad processes, AI can't fix them, and so it's really important to get your arms around them before AI can accelerate and improve your processes.

Three, you've got to understand prompting and increasingly more so how to use the coding tools that have come out, understanding the policies around what you can put into LLMs and what's secure. And then five, understanding the AI-enabled tools that you can bolt on top of your core systems and data.

And, happy to go into each one of those, but I think the two that I would continue to circle are making sure that you have that foundational data layer that's very structured, really understanding across multi dimensions how different data points are connected and what you want to do with that data.

As it relates to processes, there are free low-cost $20 a month tools that will follow along [00:06:00] clicks on your computer that will actually map out what your teams are doing, what buttons they're pushing, and then that will help you say, "Hey, I've got a hundred step process to do bank reconciliations that shouldn't be 100 steps.

And now that I've actually mapped it out and know where the bodies are buried, I can start to layer on AI to automate some of those manual tasks."

I threw a lot at you. I'll pause there. Happy to unpack any one of those five things.

Lloyd Metz: Let's stick a pin in it. I'm curious, when you started at BluWave, did you come in already with that five-part framework, or did that framework evolve over time as you spent more time figuring out what you needed at BluWave?

Jeff Berry: It's definitely evolved, and I'm very fortunate thanks to Sean's foresight around the structure of our data, which has really been a very strong sort of base camp from which we could build off of.

Most companies, I would say, aren't as clairvoyant in terms of investing in their data early and getting it as structured as we have here. And so [00:07:00] I think the framework has definitely been evolving as we've seen what AI can do. And as it evolves, so too does the framework. For example, the models two years ago couldn't do reliable variance commentary.

Now they can do that. And so I think you have to evolve your framework as the current changes.

Lloyd Metz: Can we just unpack that data part a little bit more? It's mentioned a lot. I'm still trying to get my brain around what exactly does that mean. Can you give an example of what you mean by having the data at BluWave structured?

What does data encompass? Are you talking about just financial and accounting numbers, or are you talking about other elements of the business?

Jeff Berry: I think it really starts with connecting your CRM and your ERP and joining that data together, and what you want in your data lake or your data cube, you don't want just sales by customer.

You want sales by rep, by customer, by geography, by product, as many dimensions as you can possibly capture. [00:08:00] You want that, think of it in columns and rows running across as if it were in a giant Excel file. You'd want as many columns and rows of data joined together as possible so that you could pivot the data or look at the data in as many different dimensions as possible.

And when you have that, with AI building a narrative layer on top of it, it can be really powerful in terms of how you attack the questions that you want to ask of your data.

Sean Mooney: When you get feeds out of your ERP system and directly out of your CRM, is it structured or is it just in, like, a big stream of consciousness, and are there things that we have to do to, quote-unquote, organize it or transform the data to make it useful?

Jeff Berry: You really do have to transform it and use unique identifiers to be able to join data from your CRM and your ERP. So if you wanted to be able to say, let's use that sales by rep, by product, by geography example. In a standard ERP or in QuickBooks, you wouldn't be able to see that level of data, and so what you would want is that unique identifier that can join it from your CRM [00:09:00] to enrich the data with as many dimensions as possible.

And that way, an analyst or someone in the CFO's org doesn't have to do Excel gymnastics to be able to get to that question by downloading data from a CRM that's probably not clean and then joining it with, in Excel, with ERP system data that probably has its flaws as well. You want that in sort of one single source of truth, ideally in a data lake.

Sean Mooney: So we have a full-time data architect at BluWave who's been on our team for a number of years. A large part of his job is to take the data from these different systems. And at this point it's our CRM, it's our sales automation and management system, it's our marketing automation system.

And he will pull all of that data, and it comes in, like, streams. And he will put it, say, "Well, okay, this is what sales equals. This is what equals cost of goods." And he'll put them literally in tables. So he organizes it. because what comes out of your systems is really just, like, a bunch of gobbledygook.

And then what he'll do is [00:10:00] what's called 'semantic layering.' And for that, we have a software system. He'll type in, "Sales. This is the revenue that we earn. Rep is, like, there's two types of rep. We have business development reps and account management reps." And so he's literally describing in type form what these fields mean so that the AI, when it reads it, it goes, "Oh, this makes sense."

And then what will happen is for the things that we're always asking pretty regularly, and this is, like, literally every week we're doing new common calculations. He'll say, "Hey, in order to make this calculation," if someone asks it on our internal system we casually call Jarvis, that everyone has access to, it's like, "Take this field, plus that field, divided by this field to get this answer or something similar." So there's been a huge investment just saying, A, keep our data clean. B, put the data in structure. And then C, like, literally typing notes saying, "This is what this data means," so the AI can interpret it.

But anyone can do this in, like, three months.

Doug McCormick: The way you described it there, I'm going to say two things, and I'm looking for a little feedback. The first is, I assume the [00:11:00] lift to go back and recreate a lot of this data is pretty data intensive or time intensive. And so you guys kind of picked a spot and said, "Hey, this is the point we're going to start collecting the appropriate architecture that we can analyze it on a go-forward basis."

And early on, you don't have much to compare it to historically because you just don't have the data. So you kind of pick a spot and move forward. That's question one.

And then the way you described it, though it does feel like this is perpetually evolving, as you deliver new insights, you have new questions, and so you continue to, like, I guess, increase the different fields that are going to be relevant for you on a decision-making basis.

Sean Mooney: I'll defer to Jeff, but what I will say, though, is if you have common processes and a common way you're capturing things, and that's been consistent over time, you can then look back in time and go back in time. But if you're doing things differently all the time, and you're constantly changing things, then your year-over-year comparisons get harder.

But that also, in my mind, raises the urgency of, like, you better get going [00:12:00] now to get a common standard in.

Doug McCormick: Beause you want that year-over-year, right?

Sean Mooney: Yeah, that problem just gets bigger and bigger over time.

Jeff, what's your perspective on that?

Jeff Berry: Yeah, I think, yesterday is when companies should have done this.

I'll never not have the data that I have now at BluWave. That just will be a very foundational investment I'll always want to make, because we're just so dynamic with what we're able to do with it.

Sean: This episode is brought to you today by Oridian Capital Partners, a lower-middle market private equity firm focused on partnering with family and founder-owned manufacturing, service, and distribution companies.

ICV Partners, an innovative private equity firm supporting management teams of leading companies at the lower end of the middle market.

And BluWave, connecting the most proactive business builders in the world with the best-of-the-best service providers, interim executives, AI advisors, for critical, variable, on-point, and on-time due diligence and value creation [00:13:00] needs.

Lloyd Metz: Sean, a question for you: what did you see or what gave you the conviction to make the investment in data? Because that's the starting point. It sounds like a lot of work. It sounds like it's expensive. Sounds like you've got to hire some people that you've never had on the team before, but you must have seen something that gave you the oomph, the conviction to plow ahead.

What did you see and what was it? And I'm asking on behalf of our listeners who may be wrestling with, "Hey, I'm not there. We're not there. Our data's not clean. I barely understand data. How do I get started," right? That's part of the theme of the series of conversations. How do you get started and then how do you get in the game?

What was it that got you going and committed to making that investment in data?

Sean Mooney: This all started 15, 20 years ago. So it really started at my prior PE firm, where I was co-leading the area that invested in information, [00:14:00] data, analytically enabled businesses. And so those were the types of companies where we could see every day, like what you could do with this data, how you should structure it, understanding the data in itself are just like ingredients like flour, sugar, salt. But if you put them together, you can create cakes, proverbially. This goes back to 2008, right? Where we would immediately get in and start structuring the data, making sure the things we can do with it, use it, the capabilities weren't that great. So when we started in 2017, effectively, my aspirations for what BluWave could do right out of the gates was much grander than the robotic capabilities that existed, right?

So, like, we just couldn't do the algorithmic things that we can do very easily now. But I knew someday the robots were going to get there. And so there was a commitment, like every day, we're going to work on keeping this data as structured as we can, because eventually the robots are going to get there and be smart enough.

Almost every year, it used to be called machine learning before AI, and I would call the groups in our network, like, "Is this [00:15:00] stuff ready?" In most years, it was, "No, you have to have like a Google or, then, Facebook budget to use these tools." But then what happened is like Facebook started like rolling out these tools.

And so about a year and a half before ChatGPT came out, the groups that we were talking were like, "Hey, it's ready." A lot of these products have been commercialized, like PyTorch and things like that. And I was like, "All right, let's go." So then we actually started building our own AI engines and things like that.

But it really came from like a deep understanding, frankly, just by the virtue of the mentors I had at the last firm, and the types of companies I've invested in realized how important this data is. So it was built in from day one. But I'll tell you, like literally every week, we're getting projects going with lower middle market PortCos to start getting their data structured, and you can do it relatively reasonably because now the AI can help you structure the data, and it's not that expensive in the grand scheme of things.

You need the audacity to get moving now, because as Jeff will tell you, it limits what you can do in the future if you don't do it.

Lloyd Metz: And [00:16:00] ballpark, Sean, give us a sense for when you say it's more reasonable now, bigger than a breadbox, smaller than?

Sean Mooney: Put like huge brackets on this because it's going to depend on the complexity of your organization, the complexity of your data stack, and the status of it.

But in general, you can do quite a lot to get your data structured, call it $25 to $75,000 for a project, right? And there's going to be pluses and minuses, probably more on the plus. And then the one thing that you're going to have to do is you're also going to have to pay someone to kind of keep it evolving. We actually have a full-time data architect that does that, but we also use an outsourced partner because they're getting so popular.

And you can do that depending on your needs and use cases. You're going to spend probably another $50 to $100 grand a year on an outsourced partner to keep it evolving and going, and all the cool stuff we're doing now. But you can get to a really pretty high level of capability in three months if you have the conviction to do it.

Is that fair, Jeff?

Jeff Berry: Yeah, totally agree.

Doug McCormick: We talked about [00:17:00] data structure for a second. Just hit on processes a little bit. You mentioned these tools. That sounds very Orwellian. Just walk me through how the tools work and how easy that is.

Jeff Berry: I've been in situations where you join a company, billing is in the person's head, and then the person decides to leave a week after you join, and you've got to figure out how I'm going to go and bill 150 customers.

With or without AI, you want documentation, so if someone leaves, someone gets sick, your next person up has documentation to be able to complete the accounting and finance workflows. That's just good hygiene in general. Now there are AI-enabled tools that are either free, or $20 bucks a month that follow your screen. Every time you click, it will generate a screenshot with a description of what you're doing to basically form SOPs so that someone doesn't have to manually and painfully write out each step that they're taking. And then as an added benefit, your auditors will like it because you've got documentation. Two, once you have the documentation, you can put it into an [00:18:00] LLM and ask it for free consulting to diagnose your process.

And you can say, "Hey, I'm doing this hundred-step accounting workflow through manual Excel manipulation. How can I make it better?" And so AI can both help create the documentation, help improve what you're doing, but I think it's that initial step to just get it down on paper that, with or without AI, is just good practice.

Sean Mooney: I'll share that Jeff did a great job for us as we had a lot of the things structured, but our processes weren't really mapped when he joined. And so that was the tool that he used. Put huge brackets on this, because this world changes very rapidly, but what's the name of the tool that you used for that when you first did it?

And I don't know the state of it today or not, but...

Jeff Berry: The tool that we used is called Scribe, and again, there's a free version, then there's a $20 a month team version. The founder was a former McKinsey consultant who used to go, and she said that she would sit with the best person at the company where she was doing an engagement and would record all the steps that they were taking to suggest process improvements.

And that's where this $1.5 billion dollar company, [00:19:00] that idea came out of those McKinsey engagements. And so really it becomes your brain for processes, and so I highly recommend, whether it's Scribe or other tools like it, to just get everything down in one central place so that you don't have single points of failure.

Sean Mooney: And literally what it's doing is it's just like recording what the person's doing step by step, and it turns it into a value stream map that's documented, notated, and I believe even visualized.

Jeff Berry: That's right. Screenshots of exactly what they're doing. And I'll tell you what, for me, the way that I learn, if I were to sit down with an accountant and say, "Hey, tell me what you're doing," I would get lost by step three because I'm just not that kind of learner.

I'm more of an active learner and visual learner. But being able to actually review these documents and say, "Okay, now I understand this process," that gives me the tools I need to figure out how to actually improve the process.

Sean Mooney: And once again, this is an AI-enabled tool that's relatively inexpensive, that anyone can use day one, and you can map your processes, which in itself is really valuable because very few companies end up doing that, and it's the unfun, [00:20:00] unsexy part of anyone's job.

But now you just have them do their job, and it maps it. Two, that becomes a huge benefit if you have unexpected turnover, because someone can now jump into that role and say, "Oh, here's how you do it," and I can visually see each step. So their ramp time is a lot slower. And then the third benefit that Jeff does is then he feeds it into one of our large language models and say, "Act as a Lean Six Sigma expert. Improve my process for me." And it'll look through all this and find the abilities to simplify.

Lloyd Metz: And you've actually done this, Jeff?

Jeff Berry: Actually done it, and it's led to meaningful time savings across several of our work streams.

Lloyd Metz: Can you give us an example?

Jeff Berry: Let's talk about AP. We had a really old system, cards and expense system, that didn't have approval workflows.

And so what we would be doing is an accountant would be emailing different people in the organization saying, "We got this invoice over email." Forward the email to them and say, "Do you approve?" And then, okay, that [00:21:00] person approved. In my head, I know that it's over twenty-five thousand dollars, so I've got to escalate it to Sean because that's our threshold when it goes to the CEO for approval.

And it was taking forty to fifty hours a month for that person to just basically be sending emails saying, "Do you approve?" And then hounding people to get them to approve. What we did instead was put in an AI-enabled AP and expense system that basically had those approval workflows so that it could basically ping the people directly to say, "Do you approve of this invoice?", read the invoice through AI and then code it to the right general ledger account. So if you took an Uber, it would say, "This is a travel and entertainment expense, not a software expense." So there's no human error related to coding. And then it would push the vendor directly into our ERP system so that you wouldn't have to download the data out of the AP system and then manually re-upload it into the ERP system to capture the transaction level data.

And so through all of that, that's saving us forty, fifty hours a month. [00:22:00]

Doug McCormick: And when you first used the tool to kind of just codify what you're doing, it was able to look at that system that you just articulated and come to that conclusion and then suggest the system, or was the new system kind of your thought creation?

Jeff Berry: I've been fortunate to be in several CFO groups where people are trading best practices, and so this one came out of that one. An LLM would probably also recommend sort of the top systems as well. I'm sure that you could do it that way, but always I recommend coming to BluWave for more vetted recommendations on your unique set of circumstances.

Doug McCormick: I think one of the interesting takeaways for me, as you describe that, is we talked about the importance of data upfront, which I think is like really the holy grail in terms of like better business decisions. But this is an area that's really less about having good data and more about having good documented processes that allow you to drive efficiencies in a different way.

So for those that are still struggling with the data problem, this is an interesting entry point in my mind of continuing to make progress while you solve the bigger problems. [00:23:00]

Jeff Berry: You don't even need clean data. You just need to know what are the workflows that I want, and how do I get everyone to adopt it is really the challenge.

Sean Mooney: And the thing that I think really helps with mapping these processes is like everyone has an intuitive sense of like how big of an opportunity or challenge it is. Until you map it, you don't really know how big the opportunity is or the mess is, right? And then when you do that, suddenly you can start force ranking what you're going to work against.

Because I think where a lot of companies get in trouble is they're working on a hundred projects. Some are a little small and have no impact, some are really big. And this really helps Jeff focus on ease of impact and size of opportunity in terms of what we tick off next. And that was almost impossible to do until he mapped the processes out.

Doug McCormick: So Jeff, I'll be a little bit maybe controversial here. So when you described your five areas, data, I was like, "Wow, I can't wait to hear about that." Processes, I was like, "Wow, I can't wait to hear about that." Enabled tools, super exciting to me, and I kind of like, [00:24:00] "Eh, prompting, policies, interesting, but like less so."

And so like make the counterargument. What about those two are super interesting in your mind or critical to the whole ecosystem?

Jeff Berry: I think policies is an easy one. I think you just need to do two things One, you need to set a culture of AI adoption, and that's got to come top-down, where everyone should be using the tools, and they shouldn't make it seem like it's an afterthought, or they don't have time for it. It's got to come into their workflows day-to-day.

One of the things I give Sean a lot of credit for is mandating that everyone go through a specific Claude training and the Anthropic specific training that's free that they put out there, and everyone gets a certificate of completion. Just that culture around, "Hey, these things are coming. If you want to work here, you've got to use these tools." That's a policy that I think is worth enforcing.

Two, it's setting clear guardrails around what you can and can't put into the system. We said, "Hey, we're not going to put Social Security numbers and banking information into the system, but we can put in our data such [00:25:00] that we can chat with our data inside of Claude."

So I think setting those guardrails and just getting everyone to opt in is an important thing. Prompting's really evolving. You used to have to do very specific prompts and very structured prompts. I think the models are getting better where you can give a little bit less context. But where it's getting really interesting is inside of the coding tools.

I've been using Claude Code for, I'd call it advanced FP&A-related workstreams. You're doing more software engineering type work, and you've got to be very specific around what you want it to do if you're building advanced models.

Doug McCormick: Super helpful.

Lloyd Metz: People hear Claude Code, and people hear work streams, they're used frequently in the context of AI discussions or commentary.

Can you give a more specific, tangible example of how do you use Claude Code and what were you trying to figure out, build, or solve as it relates to your FP&A function?

Jeff Berry: I'll give two examples that I think are really interesting recently. So like most companies, you close the deal, you book [00:26:00] the deal, you don't bill it until later on.

And so that naturally creates a bookings versus billings gap, and you want to understand when you're going to bill and when your cash is going to, is going to come in. And so what I was able to do through one of our AI-enabled tools was pull into Excel all of our bookings data, and then pull in into another tab all of our billings data.

What I worked on with Claude Code is I said, "Hey, in column A, I have the date of the booking. In column B, I have the amount of the booking. Then in the other tab, I've got, this is my data structure. What I want you to do is I want you to join the booking and the billing by the unique identifier, and I want you to tell me when we think everything is going to bill. And then I want to be able to create a waterfall chart that shows this is how much I've booked, this is how much we're going to bill in the future, this is how much we billed against prior year cohorts of bookings, and this is our ending billings amount. So I can understand exactly where the bookings versus billings gap is coming from."

And, it built a... what I would give to a FP&A analyst [00:27:00] to be able to do, or something that probably would've taken me a couple hours in Excel, it probably did in twenty minutes. That's one example.

I think more for a private equity audience, this is a really interesting one. We recently renewed a credit facility, and it had some amortization features related to the credit facility that I haven't modeled since I was a second-year analyst at Morgan Stanley.

And so what I said was, "Hey, here's a PDF of the term sheet. I have a seven hundred-row, three-statement financial model. What I want you to do is build in a debt schedule so that if we choose to draw on it, create a switch where we draw on it and then create the amortization based on the term sheet. Then I want you to wire in everything into the three financial statements so that the balance sheet balances, so that it reflects the draw on the cash flow statement, so that the interest expense comes into the income statement, and so on."

And it just did it. And again, it probably would've taken me probably several hours to figure that out, whereas the models are just doing it in twenty minutes.

Lloyd Metz: Gosh. So to that point, now you've got me thinking, how many [00:28:00] people, out of curiosity, are in your office of the CFO and your FP&A team at BluWave today?

Jeff Berry: I'm the FP&A team/CFO. I initially thought coming in I was going to need to hire someone in FP&A. And then because our data is so structured, and because of some of the tools that we've been able to bolt on to our systems, I don't think I'm going to need someone.

Sean Mooney: So Jeff has that in his budget and has had it, but he keeps on saying, "I don't think we need it."

Lloyd Metz: Get out of here. What did your team look like when you started at BluWave?

Jeff Berry: Yeah, so we had four people in accounting. We've got fairly complex billings. We had two people focused on billing, one on the general ledger. And then we had a controller who was really systems-minded, and I thought he was too good for daily AP, AR, general ledger stuff.

And I said, "Hey, I'm going to path you. You're going to help me Agendify all of our systems." And so I moved him into a financial systems director role to help build out even more automation. And now we have an accounting manager who's overseeing the kind of day-to-day [00:29:00] workflows

Lloyd Metz: So when you started, there were four people?

Jeff Berry: Still have four, but we've reshuffled.

Lloyd Metz: Got it.

Same four people broadly defined, maybe one moved on, but still works closely with you, added another. But the company is what? Three, four, five times bigger?

Jeff Berry: Yeah, probably three times bigger.

Doug McCormick: It's great operating leverage.

Sean Mooney: If you were to map forward five years, how much more do you think you would need?

Jeff Berry: I don't think we're going to need more people.

Sean Mooney: At multiple times bigger.

Doug McCormick: I think that's an amazing setup for kind of the enabled tools discussion. Maybe, Jeff, you just frame it out as: here are the activities that the office of the CFO manages, and then give us a kind of play-by-play of use cases in each of those major mandates where you're deploying AI and what the tools can do for us.

Jeff Berry: Yeah. Let's say AP, AR, general ledger, FP&A, which are core office of the CFO activities, and then we can go broader from there.

AP, we talked about. There are very good [00:30:00] tech-first systems, and there are some companies that are just raising billions of dollars every year to solve accounting workflows.

And they've really mastered a lot of these workflows so that you don't have to manually code invoices. You have approval workflows. You can basically build agents for policies. So someone can ask the agent, "Hey, is this office chair in budget? When I come to headquarters is getting a dog sitter for my dog in budget?"

And then the policies can read your policy. It can literally take your handbook and then give a person guidance on the policies. So AP is really good, and those tools tend to be very low cost because you get cash back on the credit cards that come with the tools. And so it's basically, it's virtually free to get fifty hours a month of time savings.

AR is harder. Every company I've been at, billing is difficult. We put in an AI-enabled tool that can read contracts, create revenue schedules from those contracts, and then push the data from the revenue system of record directly into the ERP system. That's probably saving us another fifty hours a month, and [00:31:00] the people in charge of billing are no longer air traffic controlling and spending time reading contracts.

They're being able to do higher level work, and that's been a big unlock. General ledger, there are different tools that are automating reconciliations that I think are really interesting, like your monthly bank recs, or your balance sheet reconciliations, automating tasks on the month-end close checklist.

We've actually deployed our FP&A tool for accounting purposes. Because of our connected data, we're able to bring multiple data sources together to do those reconciliations. So we've repurposed an FP&A tool versus getting a general ledger tool. The FP&A space has gotten really interesting, and I'd say in the last five years, for all these tools, there's just been an explosion of capital coming to solve the office of the CFO challenges.

In the FP&A tool, what people have to decide is, do they want something that's cloud-based, or do they want something that's Excel-based? I personally like to work in Excel, and I think our company's at a maturity stage where Excel is going to take us through a very large revenue number, where we don't have to move to a cloud-based [00:32:00] type of planning tool.

And so what's great about it is, in the budget process, I can build out an Excel template with an Excel add-in, push that data to the cloud, and then download all of the data into a centralized model. Versus, when I've done this previously, I'd be consolidating and deconsolidating twenty-five different department level budgets, and then just the hours of manual work that goes into that can go away with these FP&A tools.

So it makes reporting and forecasting very straightforward.

Sean Mooney: What's really interesting on this, just from color commentary from my lens is, Jeff can basically pull in all of the data anywhere in the company at a snap of a finger, and then he has plug-in tools that can help him analyze.

We'll be in leadership team meetings or our one-on-ones, and someone will have a question that three years ago would've taken, like, a week, and Jeff will be able to say, "Here, give me one moment." He'll pull the data in real time. He'll do some quick analysis, both just intuitively, but then also cranking and having some... run some ad hoc modeling. [00:33:00] He'll give us a key insight to a real-time question in real time, in ways before that would take weeks. And you think about the art of business, so much of it is just speed, and reps, and iteration, and A/B testing, and going micro left and micro right, and removing blocks of time. And what Jeff is able to do and what he's built for us, it just enables us to move so much faster without sacrificing quality or excellence.

Lloyd Metz: Say more about that, Sean. People talk about feedback loops and OODA loops, and it sounds like that's what you're describing. How have you seen that shape decision-making, launching initiatives at BluWave?

What does that look like and feel like today versus a year ago or three years ago?

Sean Mooney: Two things. One, I just think the speed of insight, you have a key business question, it would take weeks, and we were still probably then three years ago better than probably the vast majority of companies, just even with what we had.[00:34:00]

Because normally what you would do is say, "Okay, we have a question," and Jeff would say, "Okay, I'm going to go with my team. We're going to go pull the data. We're going to join it. We're going to do some analysis. Then we're going to bring it back, and then we're going to say, 'Oh, well, actually do it this way, not that way.'" And then another week would go because everyone's got 100 balls in the air.

And so a multi-week process goes down to multi-minute in terms of just being able to get key answers because we have the data, it's structured. We know it'll give us at least a pretty confident directionally correct answer. I think there's usually another iteration that you want to finalize on. And we do that across our entire organization, like, "Hey, how are the reps doing?"

And then Jeff can pull it all down, like, okay, we're pretty meaningfully expanding the go-to-market part of our business right now. And he can pull down in real time every single rep, and how they're doing versus others and blah, blah, blah, and he can do this like a whole performance analysis in seconds.

It's fascinating. That speed of business is something that we could never accomplish. The [00:35:00] other thing that it gives us is, I don't know about your companies or you all, like we were pretty early adopters of dashboarding, and we would make them visible, and gamify our business across key metrics where people could see what they were doing.

But that was a function of having a business intelligence analyst who knew Sigma, or knew Tableau, and could essentially engineer these dashboards. And any time you wanted a change, it was a huge exercise. Now, any time any of us want information, we go into our system that we call Jarvis, where all is there, and they just ask a question.

And so today, like if I want a dashboard, I just go into Claude in CoWork, and I have it build a dashboard that is exactly the way I want it. There's some iteration because like, oh no, I want it this way, that way. It's not a mind reader. Within minutes, it's probably a 30-minute exercise. I'll get something that I would have waited on for months, and then it's the exact way I want it.

But Jeff wants to look at things slightly different than I do, and so he can [00:36:00] have his dashboards the exact way he wants them. I can look at mine the way I want them, and you don't have to have an overworked BI person who's doing an engineering task. And so it's just like the customization of insights, and the speed of insights are exponentially better than they were even one year ago.

Jeff Berry: I completely agree. We're running into a problem where we have so much data, it's: what do you do with it? And that's where I think it comes down to processes, and what are the right ways to operationalize the data. What are the right cadences you want to get into so that with all this data, you don't turn left, then turn right.

Sean Mooney: And so like another example, like our head of technology, who we had on the last group, James, who's also really sharp. If people seem like, particularly in technology, there's a tendency to say, "I want to have a feature farm, and I'm going to do a thousand things." They all could be little things that no one wants, or they could all be like a couple big things, but they're going to take five years.

And so what he's built in our Jarvis system is essentially an ROI engine. And so anything that he's considering building and [00:37:00] proposing to the leadership team and me, he'll basically come in, and which is the other Shangri-La that every team aspires for, is like, "What's the ROI on this investment? How does it compare to other things?"

And he will go through his whole roadmap, and he's got a force ranked list of like, "Here's where I think the ROI is, relative to these others." And it includes an assessment of outcome and ease of impact. These are all the things that I always wanted, but we never really had the chance because it was just too hard and there are only 25 hours in a day.

And to Jeff's point, now it's the idea of, like, these things are happening faster. Now it's the idea of, like, it's probably the bigger thing is, like, editing and what matters most. because, just like anything, it now it probably turns into, like, you're doing a thousand analyses. What are the few that matters?

Lloyd Metz: That's a good segue to my question for both you and Jeff. With this increased automation, increased intelligence built in to your processes: time savings. Instead of months, you're getting answers in days, hours, or minutes. What are you doing with that freed up time? How has your [00:38:00] day, Jeff, changed? How does that look today versus last year versus three, four years ago?

And I'm going to come to you, Sean, with the same question.

Jeff Berry: It certainly changes how you work. It frees up time for higher-value work. It's less time fine-tuning the model, updating our financial model, so it's literally seven hundred rows. I can actualize the model and have a rolling forecast with two keystrokes for seven hundred rows.

And so instead of spending a few hours updating the model every month once we close the books, I can actually spend more time engaging with the data and understanding the drivers of the business, and figuring out what actually happened.

Another example, we have a fifty-slide reporting package. It's a similar process to update that every month. Couple of clicks in Excel and PowerPoint, and then I can give it to Claude to actually write the commentary. That would've taken a few days in my prior life to do that, so we can get insights faster. I can be more of a customer of the data versus a producer of the data. It leaves more time for experimentation.

It's: what are problems in the business or [00:39:00] problems in my function that I want to solve, and how can I use these tools to help me solve them?

Doug McCormick: I'm going to extrapolate a little bit, and just give me a reality check. You talked about AP and AR as a fifty hours of savings a week. You've got a four-person team.

I'm extrapolating and basically saying ten hours a week for a person, that was their function. It feels to me like you could realistically say to yourself, I think the AI tools across your entire suite of services, or your mandate, can generate efficiencies of about 25%, and importantly, it's the least value add 25% of your day.

And so, in terms of like capacity to actually drive the business versus deliver the business, it's gone up more. Is that a reasonable way to think about like the size of the prize here?

Jeff Berry: For sure. And it's... I don't want a six-figure employee being a professional sender of emails to follow up with people.

I want them figuring out where we have gaps in controls, where we are today when we're ten times bigger, what do we need to be doing [00:40:00] today to get in front of those potential gaps.

Sean Mooney: And I think that's an excellent point. So there are hard productivity savings. And for us, we don't view it as a deal that like, oh, this is like a cost savings, right?

For us, this is enabling the people we have to spend more time productively working on the business, versus in the business in doing the higher value added things. And we as an organization are still hiring lots of people. Probably more so than we otherwise would have if we didn't have these tools. Because the field and the frame of usage of things we can do is just so much bigger and faster.

And so, yeah, in a zero-sum game, if it was just us and we didn't have larger objectives and we just wanted to optimize around steady-as-she-goes, there'd probably be great cost/savings opportunities. But for any business that aspires to grow, and works in a matrix where you have competition, you're playing against the rest of the world.

And so we're redeploying any productivity into our team, and into our growth, so that we can do more [00:41:00] with less.

For us, I don't view this, and I think there's a lot of fear around, oh, jobs are going away. I don't view the jobs going away at all through our lens. We're hiring by leaps and bounds just because we're able to do more faster.

Doug McCormick: I agree with everything you said, and the whole premise in my mind is that you have a growth mentality where you're redeploying those assets to growth. And if you think big picture though, if a market's not growing, you've got to be taking share, right? And so like you think about, like, from a societal perspective, I still think there's huge productivity benefits that are going to allow us collectively as a society to grow faster.

Isn't there going to be a big share gain and loss between those that have adopted? You're using speed as a competitive tool now, where in your marketplace you're going to be a more dominant provider, is like the implication if you believe what your opportunity is. The tools don't really help you grow the market. It helps you be more competitive in the market, don't you think?

Sean Mooney: I think it's both. I think you're absolutely right. You can be more competitive in market, you [00:42:00] can take share, but as these tools increase productivity, inevitably what people worry is like, okay, well, the price points are going to come down. But inevitably, if you look at these periods of time where you increase productivity and price points maybe go down, usage goes way up.

What happened with computer chips? Oh, the price is going down down, down, down. But with Moore's Law, they got more and more competitive. Like the other great example is like radiologists. Seven years ago, they stopped training radiologists because one of the first use cases on machine learning, which now became AI, was like cancer cell recognition.

What they were able to figure out was like, "Oh my gosh, like robots are going to do all this stuff." But seven years ago, what you used to have to do is you'd have to wait a month for someone to read your scan, and it was really expensive, so no doctors ever would order X-rays, right? Or MRIs. Today, the cost of that has gone way down, but the radiologists are sold out across the country right now, because now the cost of a scan is so much lower, and you get it in like two hours versus [00:43:00] four weeks.

And so I do think there's like an expansion of the pie that it causes as well. I think it's more of an 'and' not 'or,' but it, it's a good point, but I do think it also will create more abundance within categories.

Doug McCormick: I like your take on it. It's the Goldilocks situation, right? It's growth of markets, anti-inflation, and anti-unemployment.

Sean Mooney: And to be fair, the sad reality of times like this is, there are going to be periods in time where jobs go away. And that's really hard when that impacts you, and it's super hard to justify. But when, if you look at the broader benefit, like if you were a scribe when the printing press came out, you were not too psyched, right?

But what did it do? It created this explosion of information across the world that accelerated the advancement of our societies. And I think this is maybe similar to that.

Doug McCormick: I sucked Sean, and he and I, into a conversation pontificating on where the society's headed.

Jeff, if you could just give us a little more, like, get back to the tactics [00:44:00] of, okay, we talked about AR, AP is really good tools, FP&A.

What are the other most compelling tools you're contemplating? And let's assume you have some purview into HR, and some of the other functions like sales.

A little bit more broadly: anything really interesting or compelling?

Jeff Berry: Yeah. I think one of the HR use-cases that I've been using personally is turning AI into my own executive coach.

There are a lot of really good executive coaches out there that have open sourced their process, and you can basically download their documents, create a project in Claude or ChatGPT and say, "Hey, I want you to be my executive coach. Here's a situation I'm dealing with. Use this person's open source documentation on how they coach executives, and guide me through this situation."

Sometimes, when I have to give difficult feedback, we use a system called Predictive Index, which is basically a Myers-Briggs sort of personality test, but a little bit more modern, how I would describe it. And I'll say, "Hey, I need to give feedback. This is the situation. I need to give feedback to this person. This is their personality type. Help me guide the [00:45:00] feedback so that it lands with them so they don't... So it doesn't come across critically."

And it'll walk me through, "Oh, okay, the way that you're going to give the feedback to a strategist is different from how you give feedback to a captain personality type.

You're a strategist, and so you're thinking about it this way, but you really need to have the empathy of this type of person to give that kind of feedback."

So that's a really good HR use case that doesn't cost anything outside of the cost of the personality test. In sales, I'm a little bit less dangerous here, but there are a lot of AI forward CRMs that are coming out.

There are revenue systems of record related to call recordings, and coaching and feedback, even role-playing avatars that are very interesting. There's certainly going to be a lot of good use cases for our teams with those tools.

Doug McCormick: Anything you can offer us in terms of, so within your core mandate, where do you think this is headed?

Big picture, big opportunities. Anything insightful there?

Jeff Berry: We're going to evolve from a month-end close to a continuous close, where you have agents that are continuously [00:46:00] reconciling transactions, and month-end close can get to be a day-one or day-zero type of activity, depending on the type of business that you run, because the agents are just doing the work that you'd be doing a few days after the month-end all the time.

That's certainly one. The AI-enabled ERP systems that are challenging the incumbents are really interesting. We'll see if they can get scale with really big businesses. That's certainly a space that we're going to watch closely.

Sean Mooney: There's a lot of lessons that you've learned, and we get a lot of questions for these things.

What are we doing with our customers who want to see kind of how we do these things in terms of rolling it out themselves?

Jeff Berry: I would say for middle-market private equity, and lower-middle market private equity owned businesses, people are still very early in their journey. We talked through a lot of the things that we did today, what are foundational low-to-no-cost things you can do if you want to invest a little bit.

This is further up the curve in terms of what you'd want to prioritize. And so we have a lot of conversations all the time with private equity firms, as well as their portfolio companies, around how to start deploying these things, and [00:47:00] what are just some quick wins that you can get.

Sean Mooney: If you're a customer of ours, just give us a call and we'll tell you exactly what we're doing and here are the tools to use.

And it's free. Commercial alert, but you know, if it's free, it's a pretty good price.

Lloyd Metz: I really like the points that you made, Jeff, and some of the points that you were making, Sean, about deploying time, mind space, and even headcount, because you can think about the business and work on the business. And as you think about growing, whether your market's growing fast or not, whether you're greenfield, or taking market share, thinking on the business and deploying people and creativity on opportunity is just fundamentally good for your company, good for your business.

That's pretty meaningful.

Sean Mooney: It has been profound. And I'll even say personally, like, what's interesting is maybe two years ago, and part of this [00:48:00] is just like the stage of our business and growing, like I never had a moment to think, in part because like there's just certain stuff I needed to do myself to feel and understand the data.

That's probably the PE background. Like sometimes I just had to play with it. But now I'm actually able to time box multiple hours across the week where it's just like free because I freed up time that I normally would have been, even as a CEO, just doing like very manual nonsense. And it's enabled me to have like multiple hours where I just keep it free, and it gives me a chance to think in ways that I've never had in my entire career.

I don't know if, Jeff, you've had that kind of opportunity as well, but it's like there was always that line, like, "Oh, work on the business versus in the business." For the vast majority of my adult life, I always felt like I was just in the business. But I would say, "Oh yeah, I'm working on it," but I never really was.

And so now for the first time ever, I think like I am.

Jeff Berry: Totally agree. And Philippe Lafont, the founder of Coté, was on CNBC yesterday, and he talked about the way that he thinks about it is, you can go to sleep now at night and have a thousand people [00:49:00] working for you, and you wake up and you can review their work.

And it's very different than doing the work of a thousand people, which is the world that we were in versus the one that we're in now. That framing really resonated with me.

Lloyd Metz: Enjoyed the conversation, Jeff. Thank you.

Doug McCormick: Yeah, me too. Appreciate the time, Jeff.

Jeff Berry: This was great. Thank you for having me.

Sean Mooney: Well, we're really fortunate to have Jeff on our team, as you guys can tell here, and so you better not come and get him.

Lloyd Metz: If any listeners want to follow up or ask questions, how should they go about it? Reach out to you? Reach out to Jeff? Post it on LinkedIn?

Sean Mooney: I think you can do all of those. We're easy to find, and we try to be everywhere to find. So you can go to our website, you can go on-- reach Jeff or me on LinkedIn.

Increasingly, we're going to have connectors that any of our customers or people who want to be customers can have that are built in directly into your teams, your Slack, your LLMs, and we're going to try to help you figure out how to actually [00:50:00] frame your challenges and opportunities easier and better, the way we do things, and make us easier to help.

We're pretty much findable anywhere, but thank you for asking.

Lloyd Metz: Great. Stay tuned, everybody.

Doug McCormick: Awesome. Appreciate the time, guys.

A special thanks to Oridian Capital Partners, a lower-middle market private equity firm focused on driving transformational growth through consolidation strategies, by partnering with family and founder-owned manufacturing, services, and distribution companies. Learn more at Oridiancapital.com.

ICV Partners, an innovative lower middle-market private equity firm supporting management teams of leading companies at the lower end of the middle market. Learn more about ICV at icvpartners.com.

And finally, BluWave, connecting the most proactive business builders in the world with the best-of-the-best service providers for critical, variable, on-point, and on-time due diligence and value creation needs. Learn more about [00:51:00] BluWave at bluwave.net.

For further information on Oridian, ICV, and BluWave and relevant topics discussed during the episode, please see the episode notes for links.

Sean Mooney: The views and opinions expressed in this program are those of the individuals presenting, and do not necessarily reflect the views or positions of any other persons or entities, including those referenced herein.

No representations, warranties, financial, legal, tax, or other advice are made herein. Consult your advisors regarding any topics discussed during this episode.