Workiva Amplify 2026 | Speed Without Trust is a Liability

Clips
More from Workiva Amplify 2026

Christie Kozlik

CAO

Accel Entertainment

play_circle_outline Enterprise AI Adoption: ChatGPT, Claude, and SaaS Tools for Slides, Meeting Notes, and Productivity amid Quality Variability

play_circle_outline Trust challenge: AI outputs appear confident but may be incorrect

play_circle_outline Proving AI ROI: KPIs for Accelerating Financial Close, SEC Reporting, and ERP Integration

play_circle_outline Regulatory and mature control environments (gaming/SEC) slow cautious AI adoption

Info
Transcript

Speed Without Trust is a Liability

Christie Kozlik

CAO Accel Entertainment

In this interview from Amplify 2026, Christie Kozlik, chief accounting officer of Accel Entertainment, joins theCUBE's Krista Case and Alison Kosik to discuss why trust, not cost, is the biggest barrier to AI adoption in finance. Kozlik details how Accel's AI journey evolved from cautious experimentation with an enterprise ChatGPT license to a broader multi-tool strategy that now includes Claude. She recounts a cautionary tale of turning on a low-cost AI feature that delivered no real output, underscoring why not all AI tools are built the same. Kozlik argues... Read more

explore Keep Exploring

How did AI get adopted in your company and what lessons did you learn about choosing, governing, and using AI tools? add

Is trust—rather than cost—one of the biggest hurdles to AI adoption in finance? add

What will distinguish finance teams that realize real value from AI from those that do not? add

How does operating within a highly regulated, mature control environment influence where, how, and how quickly you adopt AI technologies? add

bolt Powered by CUBE AI

Speed Without Trust is a Liability

search

Alison Kosik

>> Welcome back to Workiva Amplify 2026. We are coming to you live from Las Vegas. I'm Alison Kosik along with Krista Case, and we are about to talk with a customer of Workiva who is pretty happy about how everything has been going, to say the least.

Krista Case

>> Certainly. Yeah, we had kind of a conversation off camera about how they're listening to the needs and requirements. So it'll be great to hear more.

Alison Kosik

>> Yeah, let's dig in and talk with her. Her name is Christie Kozlik.

Christie Kozlik

>> Kozlik.

Alison Kosik

>> Yep. Almost like my

Christie Kozlik

>> name.

Christie Kozlik

>> Yes, that's very

Alison Kosik

>> close.

Alison Kosik

>> Christie is the Chief Accounting Officer of Accel Entertainment.

Christie Kozlik

>> Welcome to

Alison Kosik

>> theCUBE.

>> really appreciate your time.

Alison Kosik

>> Thank you so much for having me. I really appreciate

Alison Kosik

>> it.

Alison Kosik

>> So talk us through your role at Accel.

Alison Kosik

>> Yeah, so I work with Accel Entertainment. I'm their Chief Accounting Officer. And first about Accel Entertainment, a very exciting company. It's actually in the casino gaming space, so very familiar for Las Vegas, but it's distributed. So it's actually outside the four walls of casinos. We are in bars, taverns, convenience stores, gas stations. So we say we are a locals gaming market. So if you wanna be out there and not in a big casino, that's where we are for slot machines. But we actually operate 29,000 slot machines in 4,700 locations over 10 different states. So when I say distributed, it's a very different market. So as their Chief Accounting Officer, I own the accounting program, treasury management, SEC reporting, taxes, and actually their cash processing operations. So it's kind of a unique role in that I get to do operations and your normal chief accounting roles that you would see as well.

Alison Kosik

>> So as Accel Entertainment's Chief Accounting Officer, you're responsible for the numbers that the company ultimately has to stand behind. Yes. So I'm curious, where is AI already helping your team today?

Alison Kosik

>> Yeah, so AI has started to come into our company and just like I think most people, it started kind of slow and now it's like a snowball. So I get really excited about technology, really excited about adoption. So when our VP of technology started to say, hey, we got an enterprise contract with ChatGPT we're starting, I was like, ooh, when can I get that? And he was like, hold on, Christie. He's like, I'm gonna need to make sure that you're actually gonna do some governance over the data and make sure things are correct. And I was like, Jamal, I'm your chief accounting officer. If anybody's going to care about, governance and data and making sure things are correct, it really should start with us, right? So I was like, give me the license. And he said, okay, I'll give you a license. And now it's expanded and we have Claude. And so we are really trying to think, how do we use this? How do we use these different tools that are out there? I will say for me, writing PowerPoint presentations has become a heck of a lot easier. Taking meeting notes becomes a lot easier. So some of the basic things you will hear people use it for, I think have been great. But as I started thinking about the challenges with my team, right? If we talk about it, those are kind of overarching AIs, but then all of a sudden you have the different AIs within all of your SaaS tools. So Workiva, your ERP systems. And like I said, I get really excited about technology. So one of my tool vendors came to me and they said, hey, we turned on AI in our system, in our world. And I was like, that's amazing. I went to my IT team and I said, can we turn this on? They did a little checking. They're like, absolutely, Christie, we can turn it on. So one thing I'd ask myself though, and say, when you're turning on AI, what's the right AI? Because they signed off on it and they said, turn it on, Christie. So I went to my controller and I was like, we got it. We turned on AI in this space. You can go in there, we can start using it. So my team jumped in, they went in and started going through it, and they came back like 2 weeks later and I was like, what did you guys achieve? And they said, absolutely nothing. I said, but I turned on this new tool for you. This is supposed to be so great. And one of them was like, How much did we pay for this AI tool? And I said, $4.99 a month. And they were like, so you got us the garage sale version of AI, not like the high-end department store. what did you actually buy us? And so as I was sitting there thinking about it, I was like, all right, so not all AI is built the same. So Christie, you can get as excited about all these technologies as you want, but if you start selling your team a bill of goods, like really bad stuff, like they're not going to be super excited about what you're putting out there. And then guess what? Adoption's not going to be great. The output's not going to be great. And all the things we're talking about at this conference of governance and data and integrity really goes out the window. So as I think through it, we really need to be thoughtful on how we want to turn things on, what we want to use. And we are still trying to figure out our plan. And it's one thing to turn on Claude. It's another thing to use it smart. It's another thing to actually get the output from it that you really want.

Krista Case

>> Absolutely, Christie. So there's a couple of things that I want to dig in there. And I think the first one is how do you view your role and even the role of the team that you oversee I know we had a conversation off camera before we started this interview.

Alison Kosik

>> Yeah.

Krista Case

>> But how do you see your role evolving as maybe there are some opportunities for AI to strategically take some things off of your plate?

Alison Kosik

>> Yeah.

Krista Case

>> And then I think the other question is, how do you think about, picking your battles where AI can maybe safely and securely, help you to move the needle in the direction of where you think your role needs to evolve?

Alison Kosik

>> Yeah, as you think about any of us in the accounting finance space, it's just data's everywhere. It's in 25 different systems coming from different places. I think I mentioned we have 4,700 locations that we work with. We pay these customers on an ongoing basis. There's data coming in from machines from everywhere. There's just opportunity to help from a sales organization standpoint, from a footprint of who's going to go play our machines when and where. How do we take the power of these machines, which again, you can't even fathom how much data I can put into them and in microseconds it comes back with an answer. Now, is that answer right? Totally different conversation that we'll have. But as I look at my team and you keep asking for more and more output from your teams, how do you give them the right tools to get that better answer for you? And I think it really comes from which ones are we going to use? How do you get that data right into the right place so that it can be used and then fed into the right model? And I think that answer could be, is it going to be Workiva? That's probably going to be a big place for us. We have a lot of data flowing into there coming from different source systems versus an overarching enterprise-wide cloud, which is going to touch different places. We're really going to have to map our world out correctly to make sure we pull the right pieces into the right system. And I think that's really a big exercise that a lot of us are going through right now to make sure we get that right answer.

Alison Kosik

>> Yeah. Are you feeling that one of the biggest hurdles in AI adoption in finance isn't necessarily money, it's trust?

Alison Kosik

>> Oh, absolutely. Absolutely. I know somebody said 86% of executives are comfortable with the data that's coming out, and I think we have to get more and more comfortable with it. But so I'm a big communication person, right? And I don't know about you guys, but when you put something into the AI model, output comes from it is like so confident, right? It's like, here is your answer, Allison. And then you'll look at it and you'll be like, you know what, Allison? Claude, is it— are you sure this is right? And it's like, oh, you know what, Allison, that was a great question. Nope, it's absolutely wrong. But it's actually the answer that you want.

Alison Kosik

>> I know, it's like, well, why did you do it in the first place?

Alison Kosik

>> But the output is so incredibly believable and confident. And when I think about it, as I sit in my Teams meetings, right? And if we go over month-end results and I can sit there and be like, all right, Christie, tell me what's going on. How did the results look? And you'll start talking to me and I'll hear you answer. And I can be like, I can just watch you respond. And I'm like, you know what? She is not super confident in that month-end result. I can just hear it in her voice. I can hear it in the output. I'm like, let me just ask you a couple more follow-up questions. And I can just hear it a little bit more shaky where I'm like, Christie, are you 100% sure that you know that that is the output that should be there? Reading the body language, hearing how you're saying something can go a long way in knowing if that output is correct or not correct. When you're talking about AI, you don't get that. You get the absolute confidence. Or in my world, when I talk to ChatGPT, I'm a fairly positive person, so of course I put in there, Chat, when I give you something, please come back to me with very positive responses. Make sure I feel good about myself when you give me a response. So it always does, right? But that doesn't mean it's giving me a right answer. It's giving me the answer I would like to hear, that Christie, you're really great and you did a really great job writing this email. Oh, thank you. that doesn't mean it gave me the actual output that I need, right? And so it's hard when you don't have that body language element, that communication element to it. It does make it harder to know, is the output really what you need at the end of the day?

Krista Case

>> And Christie, if we're looking to begin trusting AI for accounting functions, for financial reporting functions, even leaning into sort of risk and compliance.

Alison Kosik

>> Yeah.

Krista Case

>> How do we determine when the AI is correct?

Alison Kosik

>> Yep.

Krista Case

>> And when it is not correct?

Alison Kosik

>> So I feel like it goes back to— it's going to sound like old school, right? But your checks and balances. What is the input that I am feeding into this AI? What is the output that I'm expecting? What is the root cause that I'm trying to solve for? Knowing those inputs, outputs, and expectations become so important. And when it comes to data, when it comes to numbers, like in the accounting space, right, we always want it to balance to a penny. We know an outcome. We're dealing with actuals. We want it to be right every time. FP&A, our financial planning and analysis folks, they get to play a little bit more in the future. Their numbers can be a little bit more squishy. So the outcome might be different or my answer might be different depending on what space I'm playing in, but there's always something that went in and there's always something that I'm expecting to come out and making sure I'm documenting and looking at that. There always has to be some sort of review done. And it was interesting in one of the sessions I was at, there was a big debate of like, well, does that review have to be a human or could you develop another bot to check the other bot and then another bot to check that bot? And I'm like, at what point does it become so circular that you've just checked yourself 8 different ways? But that is going to be a discussion that's going to keep coming as these agents keep getting developed.

Alison Kosik

>> Yeah, there seems to be a lot of pressure on companies to show some ROI on AI.

Krista Case

>> Yes.

Alison Kosik

>> So what's going to separate the finance teams that get real value from AI versus the ones that don't?

Alison Kosik

>> I think the real value is going to come from— I will use my example that I think about, right? The financial close process. We're always getting pressure to close it faster, more efficiently. So an ROI for my world would definitely be, can I take it from 7 to 5? And how did I do it? Did I do it with my team? Did I do it with bots? Did I do it with my close process? Or in the SEC space, can I take writing the 10-Q and the 10-K, and can I get that done faster as well? the SEC deadlines don't change from 45 days. They're still there. But can I get everything done so I can get the board, the audit committee, all those groups, their information quicker, faster, so they can make decisions that are more reliable? So, I think when you look at any process, finding that KPI or that metric that you think is most relevant and most reliable, is going to be important because today I think people still struggle to say, I've invested this much in AI and what do I actually get back for it? Because everybody's just being told, do AI, AI is going to make it better. Okay, well, what does that really mean? You really have to think about that KPI and every place that you're putting it into. And what is that walkaway point that makes you say this has been a success? And if you're not trying to do that upfront, I think you're going to walk away being like, to your point, what did I actually get out of this and was it successful for me? I think it's a hard question to answer sometimes, but that KPI metric, I think, is just critical.

Krista Case

>> It is difficult and I agree it's an evolving conversation. You mentioned and we do tend to think a lot about speed when we think about AI. But what about things like using AI to better detect variances, right? And kind of in your reporting or using AI to maybe kind of expand the scope and coverage of your team. Are there other KPIs that we should be thinking about as we're considering the business use cases for AI?

Alison Kosik

>> Yeah, I alluded to it earlier, in my world being distributed, right? We have, you know, it's not just my company trying to go out there and get that local gaming market. We want everybody to want to come and play for us. So from a player reward standpoint, from a footprint of when will people be out wanting to go gaming and want to go find that entertainment, how do we attract that customer? What's going to be that sales pitch or when does everybody get off work at a certain point in time and when should I then target that customer to want to come and play and go to my establishment? And all of our customers, it's like local establishments are trying to get these customers in to get revenue to their business. So how do we help those local businesses figure out how to market themselves as well? AI is going to be a huge factor in that because of the amount of mass data that it can take in, churn and give us an output and figure out again, will that output be right? Maybe, maybe not. But then we put more information in, get another outcome, and then keep learning and growing and going along. So I think from a sales standpoint, that'll be hugely helpful, which that sales then will come back into my world. What am I going to project for my revenue for the future? How does that look in my 5-year plan? Right. It can all start to build and waterfall on itself.

Krista Case

>> And your use case is interesting, right? Because gaming is highly regulated. It is actually. So, I know you mentioned you're very tech-forward, you're very enthusiastic about it. How does the fact that you do have to operate within this very mature control environment impact how you think about where you're going to adopt AI and even how fast?

Alison Kosik

>> Yeah, and I think that's for a lot of companies. It's again, like I said, that output control that I check it. And that's just not even for highly regulated, that's SEC people as well, right? All of our results go out to the street. They go out to our shareholders. And so you certainly don't want to be that first company, just like the Enron, SOX era of the world. You don't want to be that company that suddenly all the new regulation for AI comes into play because you were that company that made that big mistake that suddenly made all the regulation come into play. So you want to be that very thoughtful one to say, have I really thought about all the risks around what I'm putting out there? And have I thought about, am I making a decision based on something that's factual or am I making a decision based on something that just got spit out of a machine that I haven't really thought through. And like I said, that's where I think that, that governance, that committee, that knowledge point of, do I know what I'm looking at? Am I making a very verifiable decision based on something that's been vetted, something that's been checked, something that's been risk graded? And a lot of that comes from, to your question, what data is, you know, repeatable? Can I vet it against a third-party source? Does it go to another area? Is it checked against another third-party source? those things become very important. And then also, does the data that I'm using have a lot of management judgment with it? is there something that I'm putting in there that I'm putting in a lot of estimates, a lot of guesses? And so then that makes the output even more highly suspect to being not correct, right? So all of those pieces really have to just be documented, thought through. I would say to a lot of people, you know, especially in my space, get your internal audit team, get your external auditor sitting at the table with you so that you have everybody kind of saying, have we thought through everything? Because there's nothing wrong with having them all sit at the table and going on this ride with you because they're eventually going to be the ones who audit the heck out of what you're doing. And if they say what you do sounds pretty good, and we'll go along for the ride with you. And that way, if something does happen, you can all figure out where did we miss something back here. And that's normally what you're going to ask if something goes wrong.

Krista Case

>> Absolutely. It's about that auditability and that traceability. You know what, I guess you talk a lot to the human judgment piece of it. And it's a delicate balance because we need that human judgment and expertise, but at the same time, we do need to be able to lean into AI and move faster. So if you were to project over the next 12 months or so, where do you think that human judgment will be particularly critical? If you have any thoughts on that and where we'll maybe start to lean into AI more.

Alison Kosik

>> So I would like to lean into AI more in my SEC reporting space, in my ERP buildout space. Those are the two areas where I'm going to focus quite a bit of time. I know other areas in my company are really focused hugely on it, but those two areas I think are a big opportunity for me. But to your point, I am interested to see, regulations come out from the other regulators too, right, as they're all scrutinizing the field we're in. But I just see true opportunity where we have streamlined, we really know what data is going into those two systems very strongly. I know all the input sources, I know where that data's going. My team is very strong on those two areas, so I feel confident that we can move forward with AI and building out, we also use another tool that feeds data in that I feel very confident with. So as a team, we can sit down and really sketch out those processes. So those are kind of my two target areas right now, just because they're heavily governed, they're heavily audited, mm-hmm. They're very much scrutinized, which means I feel comfortable that I'll have enough eyes on that, that I won't fail, that I'll succeed in those worlds.

Krista Case

>> That's fair. That's an angle I guess I hadn't really thought about was, like you say, because it is so heavily regulated, there are going to be, more eyes on it. So, and I know we've talked about this a little bit, but, what would you bring to the table to kind of really prove, again, these outputs?

Alison Kosik

>> So I'll use my ERP space, right?

Krista Case

>> Yeah.

Alison Kosik

>> So, we could use revenue alone. So that feeds from a couple of different source systems which are tied to meter data that come from the machines used in like the gaming space. So I already know how all of that data feeds in. I know the controls around that, I know how it ends up. So if there's a way that I can use an AI bot to help feed that information through, help pull that data together, feed it into my income statement, that then I can— when it gets into my, we'll talk about Workiva workspace, If that can help me, then if I can build an agent to really build out that income statement that flows into my footnotes and then flows in even further to my analysis of what is going on with my revenue that feeds into my board presentations, my audit committee, all of those pieces then kind of getting self-built out. And they just literally showed it today that that's what they're trying to do. That would just be such a huge improvement in my world because then I could just really dig in and start looking at stuff because I know exactly where all of that information comes from. I know how I pull it in. I know what I look at. I don't feel like there's a huge risk that I won't understand the output. But now, as you talk about that, that's because I truly know that process. I feel very confident in it. As you start to peel back the onion or peel back the layers, there's probably other ones where you'd be like, how confident are you in, I don't know, Equity Method investments? I'd be like, well, that data comes from another source that I don't fully know how that's getting put together or how that's rolled up. I'm not really sure I'd want a bot just going in and pulling that in and putting that whole footnote together for me. I'd probably need to study that and make sure I'm comfortable with that information. So I really think when I keep saying data governance, it's probably more risk rating that data. Like if you look at anybody's 10-K or Q, like rate each footnote. Could this one be AI bot-ready? Yes, no, 1, 2, 3, and then kind of talk through each one of them, put a game plan together for each one of them.

Alison Kosik

>> All right, Christie, great conversation. Thanks for stopping by theCUBE.

Alison Kosik

>> Thank you so much for having me, guys. I appreciate it.

Alison Kosik

>> Got it. All right. You're watching theCUBE, the leader in live tech coverage and in-depth expert analysis. We'll be right back.