AI in Finance: Use Cases That Actually Work

Transcript

Alex Diaz-Asper: [00:00:00] About just a little over 18 years ago, and we were acquired by Cherry Bekaert end of last year. So now I’m new to Cherry Bekaert.
Jon Cochrane: No, that’s been quite the journey. I’ve learned a lot from you and your journey and just what you’re seeing, and you’ve been a great partner to us as well.
Um, one thing I’d encourage the audience to do — it’s rare that we can always have somebody like Alex with us, and so I’d encourage you as we go, please use the chat. We do stop, we do take your questions. Feel free to drop those in as we get into it. One of the things that we’re gonna dive into is kind of the state of what Alex is seeing.
Um, one question I have for the audience — we’re gonna put up a poll here — I’m curious where all of you are in your own adoption of AI. So if you go to the polls, curious kind of where you all are [00:01:00] within your own journey, what are the barriers that you’re facing?
So go ahead, answer that, and then we’ll get into some of our topics for today. I mean, Alex, you sit across and you’ve managed a bunch of different functions all at once. Everything from bookkeeping to CFO work. As you all are approaching how you’re doing what you’re doing, practically, where is AI showing up, and what are you seeing today in how those functions are changing?
Alex Diaz-Asper: That’s a great question. It’s showing up in many different forms in many different parts of the Cherry Bekaert world now. I’ll stick to the FAO part, which is my part of the world. Clearly Claude or ChatGPT, those kinds of tools are assisting people in ad hoc analysis for sure.
We’re also trying to be what I call AI humble as far as the tools out there, and it’s trying to stay on top of what is actually coming out by third [00:02:00] parties. So like if it’s the Relics or the Campfires or the tools that we can use in our practice to make ourselves more efficient.
But there’s also the fear factor of, hey, what are we exposing ourselves to from a risk standpoint, which is also a little bit nerve-wracking about the whole thing. Or, you know, what kind of flaws it introduces into our processes. Like for example, if you overly rely on some analysis and that analysis turns out to be wrong, or you build a model and the model turns out to be wrong, that does reflect on us.
So it’s just a little bit of all that. A lot of, let’s say, chaotic thinking about it, and frankly, just concern — but also very excited about what it can do, and I have some examples of stuff where we used it and it’s been awesome.
Jon Cochrane: Yeah, I can’t wait to get into some of those examples.
I’m curious, do you find that your [00:03:00] clients are pulling you into this AI conversation? Are you bringing it to them? Like where is the push-pull?
Alex Diaz-Asper: That’s a great question. We work in — my day job I was the managing partner of Tarsus Bunnell, with my client work, but I do have a book of business of clients.
So on the client side, and what am I hearing from other CFOs on our team — we’re working with a lot of tech companies, so it’s not like they’re looking for us for guidance here. A lot of them are doing things on their own and have all kinds of challenges. They’re using it from everything across the stack, from go-to-market to product development, et cetera. So they do rely on us to handle the finance and accounting stack and solution and services. We bring it into them, Jon, to answer your question quickly.
Then they might be curious, they might be asking this [00:04:00] question, especially on the inbound. But where we see it a little bit more is on the new business side, where clients — some of these newer companies that are getting started or are hitting certain inflection points — are asking, as we go into discussions about different systems, “Why don’t we go straight to these newer systems versus, you know, the Intacct, Sages, the NetSuites and those things?”
So it’s a little bit like that.
Jon Cochrane: You know, one thing that I feel, Alex — and I think a lot of us in accounting feel this — is pressure to show that we’re making momentum and we’re actually using these tools. When somebody comes to us, whether it be a CEO or investors, they’re going, “How are you showing adoption of AI?”
And you have to have an answer for that.
Alex Diaz-Asper: Right.
Jon Cochrane: And six months, 12 months ago you could kind of get away with some of the check-the-box stuff — [00:05:00] like, “I’m using it in my email, I’m using it to do certain analyses a little bit quicker.”
Now people are starting to expect more and more. But I’m curious from your perspective — there’s still, at least from what I can see, a lot of hype. There’s still a lot of people making huge claims on LinkedIn or wherever you’re, especially some of these newer folks on this.
Anyway, there’s a lot of hype in the market. I’m curious from your perspective, what is real, what’s hype? And for this audience, what should they be thinking about — like, okay, these are some real practical things you should start looking at doing now?
Alex Diaz-Asper: Yeah. I think some of these basic tools out there are great junior analysts, and if you were a CFO you would not take a junior analyst’s work product and go into a board meeting with that or [00:06:00] into a pitch deck with that. So it’s just that context setting of where you are as far as the tools and how you can use them.
And then as part of that, I think part of the fear there is what I call AI laziness — it does look great, it can come out sounding perfect, but you do need to check these things. These things are not flawless by any shape or means, just like a junior analyst. So with that kind of framing, personally, that’s how I look at it within the finance and accounting structure.
Now, we all in the finance and accounting world are really working for businesses — on an outsource basis, but I assume everybody on this webinar, a lot of them are in-house at different firms. So we’re supporting businesses that are making bigger bets in this realm. One thing is, hey, look, can we find some efficiency in our finance function.
But much more important to these companies is [00:07:00] how do we assist them in making the decisions about how they use it across product development or go-to-market, and how we measure success in that. And that’s hard. That is really hard. A lot of it is just the basics — you can’t make any decisions unless you’re tracking.
It’s just making sure you’re tracking separately your token costs and the different elements of the cost structure and how that is changing over time, so you can sort of build that into your forecasting, pricing, et cetera. We find that part is probably the more mission-critical stuff.
And we can be as helpful to our clients, and any finance and accounting person within those organizations, in making those decisions, if that’s helpful.
Jon Cochrane: No, that’s super. One thing you mentioned a little bit ago, Alex — [00:08:00] and even you just mentioned this now with the concept of a junior analyst — people always had to show, prove out, “Okay, I can stand behind this number.” Like, I could never — let’s just say I was on your team, Alex, you had just hired me — I could never imagine coming to you and saying, “Well, Alex, Excel gave me this number and here’s my number.”
And I just feel like people need to start thinking about that with AI. You can’t go to your person and be like, “Oh, well Claude gave me this” or “Excel gave me.” It’s like, no, you have to explain it. You have to be able to stand behind the number, and it’s really important that you understand how those numbers were generated.
Just like you would train a junior analyst — “Okay, these are how the numbers work, here’s how you stand behind it, here’s how you explain it.” Same thing with AI.
Alex Diaz-Asper: Yeah. Totally agree. And sometimes it’s what it’s built on too, which is the data.
And there’s a really [00:09:00] important point as it gets to the data level — like, what is it looking at? And if that’s structured in a way that is consistent and clean, which is what finance and accounting has always struggled with. And there are some great tools out there that can help with that, but that’s fundamental too.
Like, you gotta have the good data for it to extrapolate, and you don’t ever trust… I mean, we’re both old enough to know about work papers and binders and the good old days when you used to —
Jon Cochrane: I used to check them in and out, Alex. It was like I was really nervous every time I had to sign my name to check out work papers.
I had to go down to the file room, I had to walk over and sign my life away, and I knew if I ever lost those work papers, it wasn’t good. I used to sweat. It is a little scary. It is a little nerve-wracking to think of — is Claude checking out my work papers? And who has copies of those and where are they going?
Right. So you do need to be thoughtful about access and all of that. We do have — I wanna shift a little bit here, but just to recap from the poll as to where some of you are: [00:10:00] 18% of you are just getting started. Honestly this is a mixed bag across everything.
So almost a quarter of you are just getting started. A quarter of you are casually using this in day-to-day tasks. A quarter of you are starting to pilot AI-specific workflows, and then a quarter of you are actually AI-embedded into how you do your work. If any of you are interested in sharing in the chat some of the workflows or things that you’re finding really practical, I’d love to share that with the audience here.
Alex, from what has actually changed in how you all are doing your work — can you share a couple examples of what has actually changed? And I’m excited too to start showing people some… well, we have a couple demos that we’ll get into here in just a minute.
But yeah, what’s actually changed in your day-to-day or your team’s day-to-day?
Alex Diaz-Asper: Yeah. Some examples. I think on the FP&A side, it’s been immeasurably helpful. It gives you such a great analyst to sort of point and figure [00:11:00] things out.
I was just gonna give you an example. We worked with this new type of business that I had never worked with before. It was a B2C company that’s in the free-to-rip, buy-and-collect digital card business. I didn’t even know what that was.
Jon Cochrane: I’m trying to wrap my head around this. So, free to rip — I’m not actually ripping open my… I still remember the thrill of opening Pokemon cards when I was young. So this is like, I’m not gonna have anything physical to rip. I can digitally rip these cards and they have… and it’s mine, it belongs to me.
Alex Diaz-Asper: Well, that was the interesting permutation on this business model. You could say that’s what I defaulted to — like, hey, the old baseball cards or the old Pokemon cards that I used to get my kids, et cetera.
But no, this one — it’s once you rip, you can buy. You have to decide whether to buy or you lose, and you have like [00:12:00] I don’t know, a minute to make that decision, and that sort of adrenaline moment is what drives the behavior part about this.
So what was fascinating to me was — if you think about it from a model standpoint, that sounds relatively easy to model: you have this event, do they buy or do they not buy? That’s revenue. But what Claude helped me figure out was how do you measure how people behave over time? And that kind of cohort analysis and retention economics is so different in these worlds, because it’s divided up into three classes of users — you have your whales, the minnows, and your average bear.
The high-power users tend to have retention that stabilizes. The longer you become a client, the less likely you are to churn. Versus a traditional SaaS model, it’s kind of a linear decay in your revenue base as people churn. This is the exact opposite — [00:13:00] so it kind of logarithmically declines.
And that difference between a constant churn and that logarithmic decline can be up to 60% of the tail revenue, so massively important to this business model, which I did not know anything about. And the great thing about it too is that, like a lot of people on this call, I needed to see the math.
I needed to see it in Excel specifically — nothing exists in this world if it’s not in Excel. So I had to build those different retention dynamics and compare the two, and I could look into the cells and see the math, and that’s how I got to understand it.
Jon Cochrane: How would you have done this in the past? I mean — don’t get me wrong, I was pretty dangerous. I still like to think I’m pretty dangerous with Excel. Even Google Sheets, which is a little bit like breaking religion [00:14:00] within the Excel-passionate folks. But how would you have modeled this type of business in the past?
Alex Diaz-Asper: You know, frankly, we probably would’ve modeled it poorly. We probably would’ve figured somebody on the team — my group within Cherry is about 180 people, we’ve got about 15 people in FP&A and a handful of CFOs.
I would’ve gone out to them and said, “Hey.” And I did talk to somebody who had worked on something similar, but we wouldn’t have picked up this super important dynamic in retention. So frankly, because of our lack of experience specifically in this sector, we would’ve modeled it poorly.
Jon Cochrane: Well, that sounds complicated. I’m just trying to think around how I would do that today. [00:15:00] Anyway. Are there other tasks that your team… You know, one of the things that is starting to come up — there have been so many waves over the last few years. So 2021 was growth at all costs. 2022 to ’23 was, okay, now everybody needs to be efficient. And then ChatGPT launched and it was like everybody has to use AI. And then 2025 was like — Anthropic caught up, and now all of a sudden it’s like, okay, are we using Claude or are we using ChatGPT or Codex?
And now everybody’s going, okay, what do tokens cost? But I feel this new shift is coming where efficiency is table stakes, understanding your tokens and your costs is table stakes. [00:16:00] Now it’s moving beyond efficiency, and now it’s about — how are you able to take on clients that maybe you couldn’t support before? Or what are the things that you’re able to move into that you weren’t able to take on before? And how do you show that?
I’m just curious on your end — moving beyond efficiency, what are some of the things that your team is starting to lean into, take on, that maybe you weren’t able to take on before?
Alex Diaz-Asper: This was a good example — this particular client, we feel now we’re in a good position to support them, now that we see this and understand it and can start talking about what investors are looking for, which is that traction. So everything from modeling to tracking — whether it’s the cash flow or the KPIs — it’s so important in what we do.
Jon Cochrane: Oh yeah, access to information. [00:17:00] I was talking with one of our PE/VC partners, and they were saying, “Our ability to help our clients is so much greater because we have just more access to information and data than we’ve ever had before.”
Where previously they would send a PBC list — it’s more of an audit request — but there’s just so much information that can be consumed and digested in record time.
Alex Diaz-Asper: Yeah. And specifically what you’re talking about — in so much of my software world, my B2B world, it’s all about what’s happening with the customer dynamics and the cohorts, and having the data in a good place in a way that can be consumed by these tools is so important. And obviously you guys do a great job with Maxio on that.
Because that’s still driving what the new… at least what we see for our clients and what investor decisions [00:18:00] are based on is what’s happening with customer turns. Gross retention seems to be the number now as a proxy for ARR risk. So it’s really helpful to have the data in a way that you can really understand that.
Jon Cochrane: Now, we have another poll question I wanna throw out to the audience — essentially about what are the barriers that are holding you all back today. So we’re gonna launch that poll. Curious to see what some of you are saying, “Hey, what’s holding me back today from leaning into these things?”
One of the things I like to do, Alex, is just to show people some practical examples. And I have — one of the things I’ve pulled up here — this is not real data that I’m gonna show folks because, anyway, [00:19:00] this is as close to a real thing as I could put together just the other day.
In the past, I think many times finance teams and accounting teams — one of the biggest intersections between finance and the rest of the business comes down to commission plans. And it got a chuckle out of you. It’s a heated topic. And one of the things I always wanted to be able to do was to take somebody’s commission plan.
So here’s a sample commission plan I pulled together. Let me share my screen really quick. Let’s go over to screen one. All right. And this has — there’s no name in here, but it’s like, all right, here’s my base salary, here’s my commission, here’s kind of my accelerators. But every rep sometimes can have — ideally, if you have a perfect comp plan and everything goes according to accounting’s desires and hopes [00:20:00] and dreams, there’s one plan for the whole team. It never works out that way.
One of the things I always wanted to do was to demystify how do you get paid on this plan. In past lives, we’ve brought in consultants to build out calculators for everybody on the team. There are lots of ways to go about it. Sometimes the accounting team has enough bandwidth to do that.
One of the things I wanted to show here was this is a blank Excel workbook with the Claude plugin, and I wanted to show some folks some ways that I’ve gone about solving this in the past.
So what I’m gonna do is take this comp plan, drop it in here, and ask Claude for help in building out the calculator. So let’s say, “I have a comp plan. I need help modeling how my rep is going to get paid on a monthly basis. I’d like to keep this calculator [00:21:00] simple so they can enter their attainment by month and see the payout.” And let’s see what Claude comes back with. This is live, so who knows.
I think this is one thing for the audience — these are some real things as you think about… I was tempted to do, Alex, maybe we could do this later if the audience so desires — you can actually take Claude, point it at a website. Maybe we could point it at a digital token-type website.
Alex Diaz-Asper: Mm-hmm.
Jon Cochrane: And say, “Hey, help us model out this business. This is a new client we wanna win, this is some new work.” And so the cool thing here is this has gone through, it’s read the whole plan and said, “Hey, these are the elements I’m seeing.
Is this true?” And then it’s saying, “Here’s what I’m proposing on how we build the comp calculator.” How much detail should we enter per rep per month? Let’s go [00:22:00] simple. Nobody wants anything too complex. And it’s saying, “Okay, is this a ramped customer or is this somebody who’s fully ramped?” — very important in modeling out your quota attainment and budget for the year. So we’ll go, no, this is a full-year ramp, and let’s see what it comes back with.
While this thing is cooking, Alex — I’ve talked to some folks, and everybody likes to make big claims, but I’ve talked to a couple of analysts and they’re like, “I’ll never write an Excel formula again.” Is that too much, or how real is that?
Alex Diaz-Asper: I think — gosh, that’s an interesting question. There’s a certain amount of… I mean, in FP&A and I used to do banking too prior, there’s nothing worse in the world than getting another person’s model. It’s so hard to validate a model if you don’t understand it. So I’m leaning toward the answer to that question being no — [00:23:00] I do want me and my team to have to use this as a tool, but not as a replacement. But hey, I could be wrong about that.
Alex Diaz-Asper: What’s fascinating about this too is so much of what we do is based on these assumptions. Like, everything about this — you wanna know, is this a good plan? Specifically if you’re dealing with this issue and you’re a CFO at a particular client, you wanna make sure that you get this right. The compensation plan is super important, really helpful to have your people pointed in the right direction and excited about selling.
So the next question is, is this comparable to other companies at this stage in the market? And that’s another element where Claude could be really, really helpful, or other tools can be really helpful — it’s benchmarking. [00:24:00] So much of what we do in finance and accounting is tracking things and then comparing how we’re doing versus benchmark. We know what gross retention should be, we know what net retention should be. So if we can calculate our own, we wanna see where each one of our clients falls, and that’s where these tools can be super, super helpful too.
Jon Cochrane: Ooh, now we do… Jack is stirring the pot here in the comment section. Jack had a comment saying, “Everything’s pretty much fun and games until you run out of tokens.”
Alex Diaz-Asper: Right.
Jon Cochrane: As this is building out here — I’m curious, Alex, has the token conversation come up for you all? Like, how do you control that?
Alex Diaz-Asper: So right now within Cherry, it is still pretty restricted. We are a top-20 tax and CPA firm, so we have that culture where it is a bit restricted.
But personally, [00:25:00] we do have individual accounts and corporate accounts with individual monitoring of those, and I’ve had issues. To Jack’s point, I’ve had issues. Historically, one time I was having it build a deck for me where I was giving a presentation on some KPI analysis, and I was using Claude, and I ran out of tokens right on my last iteration.
Jon Cochrane: When you run out of tokens — I mean, Alex, you have a pretty senior role in the firm, and so I’m curious, when you run out of tokens, who do you have to go to to get more tokens?
Alex Diaz-Asper: Well, I just — because I could acquire the new tokens, and I really needed them, so —
Jon Cochrane: Oh, okay. Sorry, I missed that. I get to go acquire new tokens.
Alex Diaz-Asper: But others within the firm might have a separate issue. And that could be a real problem. I think to Jack’s point, so much of this — we don’t yet know the true economic cost [00:26:00] of these tools. So much is still in flux.
Jon Cochrane: Yeah. A couple things that just came in here. We actually have assumptions built out here. One of the cool things with modeling best practice is to call out your assumptions with blue cells, so you know — “Hey, my annual quota is $1 million, here’s my base salary, here’s the different ratios, my OTE.” And now my rep can actually model their net ARR for the period.
One of the things I really appreciate about using Claude within Excel is I can actually trace the formulas. I can click in and actually see where these things are pulling from, and it makes it very easy to understand how the calculation was built. And sometimes Claude likes to get a little crazy and uses, you know, maybe nested [00:27:00] index match formulas — I actually find that some of these models are pretty good at keeping the formulas simple, but you can prompt it to keep them simple. You can say, “I prefer using SUMIF, XLOOKUP,” whatever it is, and it will keep it simple so that you can understand it.
And if you have the formulas, you can take somebody else’s model that has formulas and very quickly understand prior models, demystify them, make it your own, to your point.
When I left a company, the first thing they did was throw out my 50-megabyte FP&A forecast. It was a beautiful piece of art. I’m very sad about it. But to your point, nobody wants to use somebody else’s model because it’s very hard to explain.
Alex Diaz-Asper: Right.
Jon Cochrane: Hopefully this is helpful to some of the people in the audience here.
I wanna go back to the [00:28:00] poll here, which just came in — what is the biggest barrier to doing more with AI in your finance function? Trusting the accuracy — about a third of you said trusting the accuracy of the output. Protecting sensitive data, about 10% of you said that. Not knowing which tools are actually ready for real work — I wanna come back to that one in a minute. Honestly, the biggest barrier so far is time or bandwidth to experiment.
Alex, you manage a team of hundreds of people. Like, how do you handle — I’m sure everybody there is focused on utilization and making sure everybody’s working on valuable stuff — how do you think about encouraging your teams to take the time to experiment? How do you think about encouraging folks to dive into this?
Alex Diaz-Asper: So we [00:29:00] set up a team headed by a gentleman on our team called Andrew Miller, who was responsible for innovation — he was the one who was gonna come up with use cases, do the testing, do the pilots, and then come back with results to the rest of the group.
Other than that, we are very close to signing a pilot. This is where we are in our journey — just getting started. We’re about to set up and start a pilot with one of the AI GL packages to try it out for a client engagement.
Alex Diaz-Asper: So we’re not using it a ton in native workflows, like doing the month-end close or anything like that. It’s more ad hoc. For example, we had a new client with two loan agreements. We needed some loan schedules. That was super easy — [00:30:00] you plug it into Claude, tell it, “Summarize these two loan agreements and build me two amortization schedules.” Stuff like that just kind of speeds up ad hoc tasks. It’s more on that, frankly.
And I will say it’s kind of hard to drive change in an accounting function firm because accountants — I always say controllers, it’s in the name. They like to have control, and they can sometimes be difficult to get —
Jon Cochrane: I love control, Alex.
Alex Diaz-Asper: Yes, exactly. And I think most people go into accounting because of that kind of mindset. So changes can be difficult in our world.
Jon Cochrane: Going back to the time or bandwidth to experiment. [00:31:00] One thing that actually came up last week as we were chatting with Nick over at Curve Dental — their team actually hosted a finance and HR hackathon. Their CFO sponsored it and put real money on the line. He basically said whoever comes up with the best idea — “Hey, what are all the workflows that you all are working on? How can you make… Throw AI at it. I’m not going to tell you how to throw AI at it, just see what you can do.”
The team went after it. They had two winners, and now they have a number of different automated skills and processes that have made their process much more efficient. So making it fun — [00:32:00] finding ways to adopt new tools is important. There have been different waves throughout the year, and I think it’s important that teams pause for a minute, or find an opportunity to pause — maybe a one-hour or two-hour session where you can say, “Hey, we’re all going to do this. We’re going to make it fun.”
In a past company, one of the things we actually did — we knew there were some things that our team needed to get done. We had a lot of processes and documentation that needed to get done. One of the people on our team hosted a doc day.
Alex Diaz-Asper: Right.
Jon Cochrane: We made T-shirts. We brought in donuts and bagels and lunch. Everybody on the team rolled up their sleeves and we made it fun. And I think that’s one thing that people in the audience could think about — how do you make it fun? Especially if you’re leading the team, how do you take a minute to pause and think about — if your team is slammed, nobody has time — how do you take an intentional period of time [00:33:00] just to roll up your sleeves and get comfortable with these tools?
Alex Diaz-Asper: Yeah. And I don’t know how applicable this is to other people on this call, but in our firm it’s always great to find champions or centers of progress, let’s call it — not necessarily centers of excellence because that sounds negative to the rest of the team. But it’s places where you’ll have people — maybe a controller who’s a little bit more open to these changes — they implement the change, you use that as a case study for the other team members and show how it helps. We certainly have done that prior to AI repeatedly, and it’s been helpful to us — finding these kind of champions and then leaning in and expanding with progress.
Jon Cochrane: Now, one of the other poll questions was trying to figure out what tools are actually ready for real work. [00:34:00] Alex, I feel like you all — as GA accounting firms in general — get pitched by new tools all the time. How do you go about figuring out which ones are real, have real depth? How do you go about testing them, and how should people think about evaluating these tools and which ones are ready for real work?
Alex Diaz-Asper: Yeah, that’s a great question. Like, what is even real work? Well, for example — early on we had this company (I won’t mention their name) that came and said they were a pure AI, ASC 606-ready rev rec KPI tool, and you could just give it the contracts and sit back and watch it build your revenue schedules, your invoicing, and so on.
When we saw the pitch, we were like, “Oh my gosh, this is going to save us so much time. Can we give you three contracts?” So the pilot was, we gave them three contracts and we wanted them to show us [00:35:00] everything from the revenue schedules to the deferred commissions — just show it to us. We sent them three contracts. It came back, and all three were wrong. Not even close.
So yeah, to your point you made earlier, there’s just so much hype. It’s a little bit overwhelming — all the stuff you’re hearing about how it’s gonna solve everything, you’re never gonna have to do another thing in the world, and your life is gonna be perfect once you implement this tool.
Jon Cochrane: It sounds wonderful.
Alex Diaz-Asper: It does. It does. And a lot of the things are wonderful, but yeah, like everything in life, it’s never that easy.
Alex Diaz-Asper: So test. Just test. We’re signing pilots. In this pilot, we’re gonna run it parallel with another client system. So let’s say it’s on one stack over here traditionally, and then we’re gonna start a new stack with this new tool and see how it compares. [00:36:00]
Jon Cochrane: So yeah, if you’re trying to figure it out — I even rewind to when I was evaluating my first ERP purchase at another company, trying to figure out which ERP was right for us.
Alex Diaz-Asper: Right.
Jon Cochrane: The same thing you did is what we did at the time of the evaluation. And the litmus test I always used with folks was very similar to what you said, Alex — “Give me a real example. Here are my three contracts, and I want you to show me live [00:37:00] how you would process this.” And I think that is the key —
Alex Diaz-Asper: Right.
Jon Cochrane: — thing for this audience. Show me live, and you show me. Don’t go off and get your one system expert and bring them back. And I mean, I do wanna be fair within this too, because sometimes things are complex and you do need to be methodical and think about how does this plug into the system.
So let’s have some grace there. But having somebody actually show you live how it works — whether or not it gets it right, whether it flows through the system, and then after it gets through the system, how do you actually handle a change or modification or something that comes up, maybe something that’s not part of the ordinary workflow.
Alex Diaz-Asper: Exactly.
Jon Cochrane: It’s not just some random thing out of left field, but okay, how do you actually handle this? And I remember it was very easy for us to figure out who could actually handle our business and who couldn’t —
Alex Diaz-Asper: Absolutely.
Jon Cochrane: — because we used real-life examples. And I think now more than ever that’s a really good approach. You can figure out pretty quickly which is real and which isn’t. But you also need to come up with “Hey, here are the 10 to 20 most important things for us as we’re evaluating these tools.”
Alex Diaz-Asper: That is key — understanding what are you trying to solve. Because nothing, [00:38:00] no new system, no new process is gonna be easy to implement and change. So it’s really focusing on: what are you trying to solve? Is it the revenue recognition processes? Is it the close the cash cycle? Whatever it is you’re trying to solve — the KPIs, whatever — fully understand that and make sure that is your rubric for how you evaluate these new tools.
Jon Cochrane: Now, we’re almost at the end of the webinar here, Alex, but I’m curious — two more questions for you. As you zoom out and think more broadly about Cherry Bekaert, some of the advice that you all are putting out to market — what guidance are you all putting out to your teams and to your clients, especially in the context of this whole conversation? What are the key pieces of guidance or advice that you all are rolling out internally and to your clients? [00:39:00]
Alex Diaz-Asper: That’s a great question. So I like to call it being AI humble — meaning there’s a lot of capital being deployed in this world, and there are a lot of very smart people trying to do lots of things. So understanding that what you can accomplish maybe internally with your own systems, or trying to do it on your own — just keeping that perspective that you may be in a better position just to choose winners here.
Like, do you build your own GL package or cloud-code your own GL package, or do you pick a Campfire or a Relic? That’s an absurd example, but somewhere along that line, just being humble about where you are and what you can accomplish internally.
Because I hear a lot of [00:40:00] bravado. A lot of people are like, “Hey, we’ve done this, that, and this,” and then you poke under the hood and it’s really not working as they say. So yeah, that’s a big part of it — staying AI humble and very open, just keeping your eyes open on the horizon of what’s coming as far as tools.
Jon Cochrane: I like what you’re saying there about being AI humble. I like to think that I’m pretty dangerous with these AI tools — I’ve actually deployed whole applications to two different URLs and stuff like that. The thing that I have a new appreciation for is just maintenance of these tools.
And as new models come out, models get retired, and then all of a sudden — [00:41:00] if you’re gonna start relying on these tools, especially for a GL or an ERP as your system of record and your source of truth —
Alex Diaz-Asper: Very good.
Jon Cochrane: — that’ll get hairy, if not today, in six months whenever your auditors show up or you go to fundraise, and then you have to show how everything is made.
The way our team thinks about it is we’re pretty obsessed with having reliable data sources — systems where we know that the numbers we’re pulling from are accurate. And then where we’re kind of going ham on AI are the analytics that we can run on top of that and the insights that we can unlock on top of that, because we know that we can trust our data sources underneath.
Alex Diaz-Asper: Jon, how does Maxio view the security concerns and the —
Jon Cochrane: Oh man, putting me on the spot, Alex. I think you have to — I mean, before AI came out, people had IT policies. They had security policies.
And [00:42:00] you had to be thoughtful and methodical around just protecting data in general. And I don’t think that has changed at all. And within our own team, whenever you’re connecting to those sources of data, you still have to have access to them. Like, you can’t necessarily go and access certain confidential things within your HR system unless you have a login to the HR system or something like that.
I think the thing for teams to be thoughtful about — within IT, sharing data that may be sensitive or confidential, whether it be company data, employee data — it’s just a whole lot easier now to Control+C, Control+V your spreadsheet or your output and iterate. So you need to make sure you’re cognizant of where those data sources are going, especially in the world of accounting and finance. Hopefully you’re still being methodical and protective around that.
I’ve heard with other businesses — I think there’s a real need to be thoughtful and methodical about it, but you also don’t wanna be so restrictive that you’re actually holding back your team’s [00:43:00] ability to lean into these tools. Like, one of the things I like to think about — I like doing woodworking on the side. I haven’t done it in a few years, but —
Alex Diaz-Asper: Okay.
Jon Cochrane: — where I had my three kids. I’ve got a five-year-old, a three-year-old, and an eight-month-old, so time is lacking these days.
But it’s like somebody dropped all these new fancy shiny tools that are way better than the tools I had in the past. I just don’t have an instruction manual, and I need to be experimenting and using these tools because what I can create is so much more powerful. So I think there’s a happy middle ground within IT security postures.
If you haven’t spoken to your CTO, your CIO, or whoever kind of governs your data and infrastructure and systems — like, you should be thoughtful and methodical and at least be communicating with them on how do we be mindful and protective of who has access to this stuff? What can they do? What can they access? Where can they ship the data?
You probably should be thoughtful and have [00:44:00] a policy and some training. It doesn’t have to be over-engineered, but that’s how I think about that stuff.
Alex Diaz-Asper: That’s very well put. All right. Well, I didn’t know about this woodworking side. Have you made… What’s your proudest achievement on the woodworking side?
Jon Cochrane: Cutting boards is a go-to. Cutting boards is like the gateway entry into this whole thing. I feel like I’m like, “Oh, here’s what woodworking taught me about AI.”
But —
Alex Diaz-Asper: But this is your fallback if the AI takes over everything. You can just go back to your woodworking.
Jon Cochrane: This is great. I’ll send you my Etsy shop as soon as I roll it out.
Alex Diaz-Asper: Okay.
Jon Cochrane: Fantastic. But I think to build a good cutting board, you need to figure out how to — there are lots of things you can build with woodworking, but you need to figure out how do I get a board straight, plain, 90 degree, 45 degree angles. Get everything cut properly. [00:45:00] And to have a very nice looking cutting board — if you can get that down and get your lines straight, no gaps in between, you can build a desk, you can build a table, you can build whatever.
Because it’s the same skill set. What you have for building a tiny cutting board is the same type of skills you would use on a larger project. It comes back to some of these basics. And that’s how I think about AI too, Alex. I couldn’t resist it.
Alex Diaz-Asper: Way to nail it. Way to land the plane, Jon.
Jon Cochrane: Oh my goodness. All right. And yeah, now we can watch our audience numbers drop off because of that. No, I’m kidding. I hope that was helpful for all of you as well. But I do think starting small, you build skill sets. And I encourage my team to do this too — start small, because don’t underestimate how the small things allow you to do much, much larger and meaningful and impressive things down the road.
Alex, I very much appreciate you [00:46:00] dialing in from Chicago, spending time with us. Do you have any final parting words of wisdom for people tuning in right now or who maybe watch the replay down the road?
Alex Diaz-Asper: Just be curious, keep talking, join webinars, and just hear what other people are doing. There’s a lot of experimentation, a lot of people doing really cool stuff. So yeah.
Jon Cochrane: Awesome. Well, thank you. Very much appreciate you taking time with us.
Alex Diaz-Asper: My pleasure.
Jon Cochrane: Hope you have a good, safe trip back home after your time in Chicago.
Alex Diaz-Asper: Thank you. I look forward to getting a cutting board as a gift.
Jon Cochrane: Yes. Once I get sleep again. All right. Thanks, everyone. We’ll catch you — next week we’re dialing in with Casey from K1. I think that’ll be a really great conversation as well. So we’ll see some of you — actually, that’s in two weeks from now. We’ll see some of you in two weeks.
Jon Cochrane: All right. [00:47:00]