CAS thought leader Kane Polakoff, the head of the client advisory services practice at Top 100 Firm CohnReznick, discusses how artificial intelligence can be integrated into CAS now — and what that intersection will look like over the long term.
Transcription:
Dan Hood (00:03):
Welcome to On the Air with Accounting Today. I'm Editor-in-Chief Dan Hood. AI and CAS are two of the hottest subjects in accounting these days, but you don't hear much about the interaction of the two. I'm here to talk about all that as Kane Polakoff. He's a partner and client advisory services practice leader at Top 100 firm CohnReznick Advisory. And he's also a leading thinker in the CAS community. Kane, thanks for joining us.
Kane Polakoff (00:21):
Yeah, thanks for having me here,
Dan Hood (00:22):
Dan. I want to start with CohnReznick. It started to give us a little narrow focus because this is a huge topic. How are you guys using AI in your CAS practice or have you started using AI in CAS?
Kane Polakoff (00:32):
Yeah. So we have a couple of things. So we are using Copilot and ChatGPT and some of the basics for creating documentation, helping to facilitate the training process and doing emails. However, now we're venturing into what the fund has really begun. And so we're looking at really integrating AI into workflows and how we can actually execute and process work on behalf of our team. So we're at the very beginning of that. I'll talk shortly about what we're doing, but I'm pretty excited about what's coming our way.
Dan Hood (01:02):
Yeah. Well, I mean, there's a lot of different angles we can take about. I want to start on one thing, which I think I'm not sure enough people think about is training on AI and staff and getting your staff involved in that. Maybe talk about that. I mean, are they involved in the kinds of things you're using AI for? What kind of training are you giving them? All those sorts of aspects of the human side of AI.
Kane Polakoff (01:19):
Yeah. First, we've taken the last three years to really look and kind of evaluate in the marketplace. There's been so much growth in AI. And I'm sure you get three or four emails a day on AI. And so through the evaluation, kind of learning, meeting with different providers, we've kind of got an idea because this is going to be huge from a change management perspective. And you talked about training. We wanted to make sure we've spent the last three years building our CAS practice, which we've grown quite quickly. But now the key thing is, how is AI when we deploy it? How's that going to impact our clients? How's that going to impact our employees? And how are we going to change the way we're doing business? So we have been thinking about it from a training standpoint. So we are going through a pilot phase first, identifying five to seven clients across some of our different verticals, and having us partner very closely with our AI partner to kind of go through and start that process.
(02:12):
Now, we are going to have certification. So we're going to have the team go through certification. It's going to change a lot of what our focus points are, a couple of things. We're going to use it for a categorization, a classification of documents or information that's coming together. One is helping to actually make recommendations or deploying journal entries on their behalf. And three is dynamic close. So those three aspects are really focusing on the senior accountant level. And so that's the area that we're really going to make sure that as we meet with our senior accountants, what it's going to allow us to do is to upscale. We talk about upscaling advisory work and we're going to help them support and actually move up a layer to provide more review, more understanding of what's happening and helping to spend more time with our clients.
Dan Hood (02:56):
Gotcha. You mentioned the growth of CAS practice. I talked a little about your role. Maybe just give a brief snapshot of the size of the practice. You said it's grown enormously over the last three
Kane Polakoff (03:06):
Years or so. Yeah. So the three year journey has been fun. When I started, CAS did not exist within CohnReznick. Today we have over a hundred people, over a hundred clients, and really have grown globally. So we have an operations here in some of our 30 offices around the country. We have a large facility in our Chennai operations in India. And then we're just starting to build out our Philippines operations. So we've kind of gone through all the basic foundation and that's why we thought it's a good time to really get into AI because we have the core tech stack. We have the clients and we're ready to rock and roll.
Dan Hood (03:41):
All right. Well, at an organization, there's a lot of room for applying AI to help solve some of the frictions that may come along. All right. So that gives us a sense of what's going on at CohnReznick, what you all have planned. Maybe take it a little broader. Are you seeing other organizations look at AI? How are other firms thinking about it?
Kane Polakoff (03:58):
Yeah. So I'm part of some different organizations. I know you've been through some of them with us. And when you look at all the CAS leaders in the industry, and I talk quite often with them, all of them are at either the very beginning phases of the journey, or some of them have been doing this for two or three years. Now, some of the questions that I have with the other CAS leaders are, do we build it internally? Do we do it through the general ledger system that we're working with, or do we use a provider that's going to integrate? So I think if you go and do a survey for maybe the top 50 CAS leaders, we would all have a little bit of a different answer to that question. However, everybody is now trying to figure out then how do we deploy it?
(04:36):
How's that's going to benefit our organization and how do we train our staff to support that going forward?
Dan Hood (04:41):
Yeah, because that's going to be a crucial issue. Staffing and CAS itself has always been a little different from staffing and in other major practice areas of the profession in terms of the talent mix and the focus. Do you need CPAs? Do you not? What kind of skill sets are you looking for there? And then to add AI onto it, as you said, it's an issue for accounting firms in general. But in CAS, are there elements of parts of your staff where that's the right place for them? Those are the right people to be using AI or is it sort of blanket across the team?
Kane Polakoff (05:10):
Yeah. So we have five key verticals that support our CAS practice today. And our goal is to deploy it across all of the five verticals. And we're looking at it holistically. So it's not, "Hey, it's going to be only deployed for this client or not for that client or for that team or not for that team." So we are, use your term as we're going to create a holistic approach that we're going to deploy across our practice. So it's almost going to become a role and an individual almost when you think about it. You think about artificial intelligence, we talk about human intelligence, but AI will be a processor as part of that process. So it's kind of a different way of thinking about it now. So we need to ensure as when we deploy this AI to our different clients, how are we reviewing it?
(05:55):
How are we validating that the information's correct? Is it secure? What type of information is being provided? How much power do we need to run it across many different general ledger systems? How is it going to integrate? So there's been a lot of thoughts and we've been kind of spending the last couple years kind of defining how we're going to go to market with that. And it's a pretty comprehensive plan and we're going to take it slow, but we feel confident that we'll be able to get there very quickly.
Dan Hood (06:21):
Gotcha. I want to take a different facet of CAS because this is one that before AI was even a ... Well, not that AI was ever a glimmer in somebody's eye, but before AI even entered the picture, within CAS, there's sort of a balancing between the compliance element of CAS and the advisory element of CAS. And it's one of those things you said, every firm has a slightly different balance within their practice, whether you're focusing on the basic back office work and getting that done or whether you're focusing more on advising your clients on how to better grow their business. And most firms are doing some level of both, but then they got to balance it. When you look at AI, where do you see that applying most? And certainly at least to start, is it more on a compliance focus on getting the grunt work done or are you looking at elements of ways to bring it into the advisory aspect of things?
Kane Polakoff (07:10):
So I think it's going to be an evolution. Excuse me. So I think that the very beginning will be at the transactional level. And that foundation, when you think about doing a bank reconciliation or credit card reconciliation or having data from many different systems coming together. So we're going to use the tool to help facilitate that process. So you think about the vision of CAS, we talk about CAS 2.0 or 3.0 is we want to give more with less. We want the tools that we have to allow us to spend more time with our clients. So what most of our time spent today is on accounts payable, doing reconciliations, creating financial statements. And if we're able to enable AI to help reconcile that or to eliminate some of those steps, I'm looking at it improving my capacity by 50% or even more. And when you think about, we have the global capacity, I think you've seen the statistics you shared with through AICPA where cash practices are growing 15, 20%, 30% top line year over year.
(08:10):
And in order to continue to build your capacity, we're going to leverage AI as an augmentation to that. So we have our global operations, we're going to deploy an AI. And if we can focus on the reconciliation and that transactional layer to get to almost a point where we have proper review, we still need to do QC quality insurance and making sure that's all in place. But I mean, a perfect world, you think about accounts payable, straight through processing, invoices come, they go straight in, they get GL coded, they get approved, they get paid. So you think about it from a closed management perspective to be able to get the information, reconcile the information, to post journal entries on behalf of an individual, have an individual review it. During the closed process, you have dynamic closed management. Here's step one, here's step two, step three, here's step four.
(08:56):
And then also it automatically generates the work papers for you. So think about that. Then at the end of the day, you have a financial statement. So most clients today or clients expect 10 to 12 days to close the books. So you think about what that potentially can mean. But also what that means is instead of the staff is spending so much time getting all this information together, they're spending time working on, not in the business, understanding the information to talk about and tell the story to our clients of, here's what we're looking at. We're pulling information, benchmark data, we're pulling all that information together and we're actually having consulting advisory discussions with our clients. And that's where I think things are going.
Dan Hood (09:32):
Gotcha. So it's for the moment at least, it's freeing you up from a lot of the day-to-day compliance work. It's doing that work faster, getting it to you faster as you can then turn around and use it as the basis of your sort of advisory conversations with the clients. In the long term, do you see a role for AI on that side of the fence, on the advisory aspect?
Kane Polakoff (09:50):
Yeah. So I think about the reporting aspects of taking data from the general ledger system, from the payroll system, from operational data, and you're creating the financial statement through the transactional and compliance work on a month to month basis using AI to do that. Now, once you have all that data, you're using AI to run certain logarithms to determine if or else statements are helping do projections or helping to provide information specifically on an industry itself. And you have that industry data that the AI is able to review or pull that information in, bringing it all together, and it provides a script of information, which is advisory to a client. And think about it as an actor having a script, imagine if you're an accountant then having a discussion with a client and you have a script of here's all the key things which you share with the client, and they look at it and they go, "Oh my God, I didn't even know about that.
(10:46):
" And then you're helping to partner with them and to drive better performance for our client's organization. That's the powerful. Now that may not happen in the next six months or in the next year, but we are going to get to that point very quickly. And as I see the evolution of AI, even in the last six months, I've seen a lot of demos and seen a lot of use cases that are being done, we're going to get there very quickly. And it's important for all of us as we're listening here to jump on to the bandwagon to be part of that solution. Well,
Dan Hood (11:14):
I think for a lot of accountants, a lot of accountants are probably worried about AI and what it's going to do and how it's going to take over their jobs. And a lot of accountants are also worried about having those kinds of conversations like, "What does that conversation even look like? " So knowing that it's supporting, it's enabling the conversation, it's not taking it over, but also that it's giving you that script to think a lot of accountants would be desperate for that script and be like, "This is great. Now I know what to say." Because it's not a thing that they've been trained on in many cases, right? The sort of soft skills of that kind of client interaction are not ... I don't think a lot of accountants get trained in them, certainly don't get educated in them at college. I don't know why they would, but that's good stuff.
Kane Polakoff (11:52):
Yeah. And I think to that point is we've been talking to some of the colleges and universities around the country and trying to bring awareness to CAS. I know that assurance and tax has been out there, but going back to specifically for CAS of the soft skills, the analytical skills, the communication skills. So as new members or folks come into our profession, that's that expectation where 10, 15 years ago, you would be working in a computer doing information and just providing information to somebody, but we're expecting folks to come in, understand, to be able to articulate the value and helping support. So there is a shift in upskilling and training that's required. And a question, are you scared that we're going to lose our jobs? But you think about 20 years ago, you had a bunch of data entry operators in these huge rooms, keen information.
(12:41):
Now that you don't see that anymore. So I think it's a refinement of what we're doing today. And yes, it is going to push us to be more analytical, to be more as an extrovert, and to not be afraid of having those conversations with your clients. But I think that's a good thing for the profession.
Dan Hood (12:59):
Sure, sure. It certainly moves it up in terms of value added and in client perception. I think the client perceives the value of that a lot more than they do the information that gets tossed into their desk once a month and they're like, "I don't know how to use it.
Kane Polakoff (13:12):
" Well, I mean, you hear the thing is, here's a financial statement, throw it over the fence. All right, next one, throw it
Dan Hood (13:17):
Over the fence. There's a pile of unread financial statements on
Kane Polakoff (13:20):
The other side of that fence. Yeah, it's ironic that over the last years, we spent all this time putting in financial statements, we send it to the client, we try to talk to them, and then you find out they didn't even look at it. So we want clients that want to understand that and that helps them make decisions. And I think it's going to be a learning curve for the client too. So there's going to be an education with them and what the expectations should be. And I know a lot of clients are just struggling today to get financial statements, but my hope is that's only the beginning of what we can provide to them going forward.
Dan Hood (13:49):
Right. Excellent. You mentioned among other things in terms of in terms of AI, well, AI in CAS, one of the aspects of AI is making sure that you're double checking, you're making sure the number's correct, making sure that the AI is functioning the way you want it to function. And that's obviously for a lot of people a big concern with AI and always talk about having a human in the loop to keep an eye on it. Are there any other concerns with AI as you go about implementing it, starting up your pilot programs and looking at them? Are you saying, "Oh, we also got to make sure this doesn't happen or this doesn't happen."
Kane Polakoff (14:18):
Well, I think if you're going to use an AI provider, you just want to go pick somebody off the street and say, how I'm going to use them. So I think as we put them through our process internally where you look at security, you look at privacy, you look at data integrity, you look at many different things and you evaluate that. And when you think of the infrastructure that's going to be built to support what we're creating here, and you look at all these data centers that are being built here in the US and you hear about the new ones we're going to build in space, which I'm looking forward to that day when that happens. You
Dan Hood (14:47):
Want a site visit?
Kane Polakoff (14:48):
Yeah. Well, I'm ready for that one. I don't know when that'll be, but I think that it's security, it's validation. I think it's going to even be more important to have a quality assurance program to be able to look at either stratified audits or focus audits to make sure that the information that's being provided is correct because we are still responsible for the data and for the information that's then put into the financial statement. And the last thing you want to do is you're using AI and you think everything is good. You go to the beach and you come back and you find all the data's wrong. So I think it's a partnership just as you have a team of people that's doing the work, you're going to add a new member of the team and the person's going to be an AI agent. And so I think that's the difference, but the process we want to keep similar, same kind of controls and security is very important, especially when you look in PCI or HIPAA compliance and all that.
(15:35):
So you want to make sure you look at for a SOC 2, go through the normal processes and that you have the rigor that's in place to do that too.
Dan Hood (15:42):
Gotcha. Makes a ton of sense. I want to talk a lot more about the subject because it's, like I said, these are two of the biggest acronyms in the profession right at the moment, AI and CAS, but we're going to take a quick break.
All right. We're back talking with Kane Polakoff of CohnReznick, a CAS leader kicking CAS.
Kane Polakoff (16:03):
We're kicking some major CAS.
Dan Hood (16:04):
I got to get the catchphrase in there. It's one of my favorites. But so we've been talking a lot about AI and how it's going to impact how it's being worked into CAS. And I want to dive a little bit more deeply into, you talked a little bit about how thinking about working AI into your CAS workflows. We'll dive a little bit more into that. When you see, how do firms want to start thinking about integrating AI into their workflows? In some ways, it's greenfield. They can put it anywhere they want. How do they start thinking about the best places to put it in the best way to do that?
Kane Polakoff (16:29):
Yeah. So I think like for us, we wanted to look at some of the key use cases that would make the biggest impact. How can we reduce a lot of the manual work that's required? Are there situations where you're matching different information or you're reconciling something? So we identified three or four different use cases as a starting point and we didn't want to do 10 or 20 or 30 and our hope is by these one or two, you take the top two, which is impacting 80% of the work, kind of looking at the Pareto chart from that perspective. And so that's kind of how we went about doing that. Now we're, like I said earlier, we're deploying that across different verticals that have different technologies. So each of our verticals have a different general ledger system, have a different AP automation tool, have a closed management tool, expense management.
(17:17):
So by using these two or three use cases and going across all those verticals, we're going to get a pretty good idea of how it works with QuickBooks Online or Sage Intacct or NetSuite or Acumatica or Yardi. And by starting that way, we will start learning how that is going to change the way a senior accountant does work. So that's how we went about and kind of evaluating that. And we also reached out to a lot of the different AI providers and asked them, "What use cases should you use?" And all of those that looked at me and says, "Well, you tell me what you want. " Those are the ones we're not using. So the ones that actually came and have done a lot of research and have met with a lot of us, the public accounting firms and CAS practice leaders that have had ideas of how to do it.
(18:01):
And what the evolution is, is kind of how we went about doing that. But also more importantly, those two or three use cases that we'll implement are great, but we also know what four use cases, the fifth use case and sixth case, and to get to the point over a journey to get us to where we want to be. We don't want to go big bang because we feel that the change of management will be very hard to govern. So we wanted to start slow and in a contained environment so that we're mitigating any risk for any issues or if something happens that we can address it upfront and we're using that as a way to move forward in the proper way. So we have created a governance process. So we have identified a project manager who actually is in the audience today helping to oversee this with their project manager.
(18:45):
We have a team of how we're going to go together. We have sponsors and champions that are in each of these verticals that are part of that process. And then we're going to go and actually meet with folks in person. So we're talking about today is how when we start this, our provider's going to come and meet at some of our offices. We're actually going to go out to India, spend time in our Chinai office and the same thing in the Philippines. So it's a, "Hey, we're going to start crawling first and we'll start walking and we'll move from there." But we want to make sure that people are embracing it. They see the value in it and they're not scared of it. If they become the champions for the work, if these initial seven or eight clients, then everybody else will hopefully fall in line and see the value.
(19:25):
So that's the approach that we've taken.
Dan Hood (19:26):
Right. Gotcha. And a lot of those are just standard, good principles of change management, right? Just building buy-in, et cetera, et
Kane Polakoff (19:34):
Cetera. Normally, but you do see some cases that people want to just shove something in and feel pressure to do something, but we're trying to do it on our time. So from a timetable perspective, it's, hey, we don't have to get it done just say by the end of March or end of July. It's let's do it right first and then we'll go from there.
Dan Hood (19:51):
Right. Avoiding the shiny object syndrome, we got to get it in now because everyone else has it.
Kane Polakoff (19:55):
That's right.
Dan Hood (19:55):
Makes sense. We got very deep and detailed there. I want to take a step back, a little 10,000 foot view on two timescales. How do you see AI impacting CAS over the next few years? I think you gave us some sense of your roadmap. Do you think that's what's going to apply across the profession? Sort of a lot of compliance focus, a lot of clearing up a lot of that daily grind to free accountants up to other work. Are there other ways you think it's going to impact them over the next few years? Well,
Kane Polakoff (20:23):
I think the capacity, freeing up the capacity is one which we talked about. I think also being able to use it for not only for CAS, but for tax and for some other cross-selling. So when you think about CAS, we're talking about capacity improving processes, but the other part is implementing technology and solutions. So as part of the onboarding process, as an example, how do you, for some of these general ledger systems and tech stacks, it could take two to three months to implement. Now, by creating scripts or leveraging AI to help deploy and expedite that, can you imagine implementing that in days? So I think that is something that, as I kind of step back is, yes, getting the hard work, it's going to take a couple years to just handle the compliance and the work that we talked about, but getting into implementation of technology, getting into then the advisory of the dashboards and the reporting, integrating payroll and HR systems with general ledger systems and really becoming ... The way I look at the analogy is you have the brain which becomes AI.
(21:26):
Then you have a nervous system that comes down and connects with everything around you, which reminds me like Terminator. It kind of scares me a little
Dan Hood (21:33):
Bit. Wait, if the AI is the brain and then it's the nervous, where are the people? Are
Kane Polakoff (21:37):
We the
Dan Hood (21:37):
Shoes?
Kane Polakoff (21:37):
Yeah, I don't know. I'm getting scared here. No Skynet here. But I do think that that's what you're going to see is that you're going to have open architectures of general ledger systems. You see a lot of that in the industry today. All the different tools have just become part of the ecosystem. And I think that's fascinating.
Dan Hood (21:56):
Well, that's interesting. One of the things I have not heard people talk about in terms of AI is its capacity, all technology systems, all software systems, regardless of how advanced they are, there's friction in all of them. As soon as you have more than one system in your practice, there's friction getting data back and forth and bringing everything together, getting a complete picture of whatever you're looking at, whether it's internal or external or a single client or your entire client base. And the notion of AI as something that can help that is interesting. It's really- One aspect. I haven't heard people talk about it.
Kane Polakoff (22:29):
Think about this, for example. I am a senior accountant. I'm working on a vertical on three clients and I'm reviewing the AI and I'm finding the issues. I'm doing some things right. I'm doing something wrong. That information gets pulled into a central location and then gets posted into a performance management system, which then is used to evaluate somebody's performance. That's what I'm talking about. It's able to integrate with different systems that take data almost like a traffic cop and bring it from one place and move it to another. That's what I'm talking about.
Dan Hood (23:00):
Yeah. Well, and then judging us on it. So that's also terrifying. Again, Aki, you're not selling this hard. I feel ... Dog, but it's all fascinating stuff. And I mean, that's a great example, right? If it can bring all those together and get a sense of, "Hey, person X seems to always have issues with this. " Maybe that's something to train them on or something to bring up for them, but also just that bringing all that information together in one place and being able to sift through it intelligently, which is almost impossible for a human being. There's too much data. Well,
Kane Polakoff (23:30):
And everybody says, "What's the source of truth? You think the source of truth is becoming the brain of the AI engine working with the information to bring it all together. Just like if you use Copilot today and write me an email on this, Bootman comes back and spits it out right for you, but it's like the same thing. You can find the data, provide the information. And the one thing that we didn't talk about that was interesting is that AI that I was reading in the paper where it was writing an article on behalf of somebody else and it got mad at the author because he said the author was wrong. It actually was in the Wall Street Journal. And I thought that was just fascinating where you actually thinking about, no, that's not right. And it's actually not attacking. I won't use that word to scare like a Skynet, but is actually providing, "Hey, are you sure you want to be right about this?
(24:10):
I don't think what you're saying is correct." And that is thinking in that way, which is kind of interesting.
Dan Hood (24:14):
Very cool. Very cool. Well, so obviously I think I was going to say, let's talk a few years out, but I think we've gotten there. Obviously the end result of SkyNut in between is cleaning up a lot of compliance work, but in between is there's huge scope for AI to be thinking about a lot of the things that accountants also need to think about. How in your view of this sort of the midterm future, like I said, between planes full of human skulls and Skynet machines crushing them and the current version, what's that vision in between of accountants working with AI to boost their practices, not just by taking care of the grunt work, but by getting those insights and managing those and being enabled by those?
Kane Polakoff (24:55):
Yeah. And I think, yeah, I think it's having most of the transactional work being done by them, by the data. And of course, there's a quality assurance layer that's there and creating these amazing dashboards that are dynamic and conditional that'll help forecast the future more where as we look at accountants are more historical speaking. So I think that's cool. The other thing that I see is that you may have different types of AI. So a general ledger system, they may be working on Claude versus somebody may be working on open AI versus something else, where you're going to see some of these ecosystems of AI actually coming together from different types of technology too, and to be able to pull that information in. So I think that as we look at the big investment that a lot of the general ledger systems, publicly, you see all of them that are partnering with that, it's how does that then integrate into potentially the workflow, the dynamic close process, the things that we're using from a CAS perspective and how the new technology and their AI comes together.
(25:55):
And I think that's going to become very interesting and how that all moves and how that shakes in the future.
Dan Hood (26:00):
Yeah, because it's going to be very, no matter how it ends up, whether it's Armageddon or Paradise, it's going to be fascinating to watch. Kane Polakoff of CohnReznick, thank you for coming in and sharing all this.
Kane Polakoff (26:09):
Yeah. Yeah. Thanks for your time, Dan.
Dan Hood (26:11):
All right. Thank you all for listening. This episode of On the Air was produced by Accounting Today with audio production by Adnan Khan. Rate or review us on your favorite podcast platform and see the rest of our contact on
