Tax Department of the Future: Reimagining Tax Technology
Experts stress research and evaluation when diving into new solutions

print this article

On the first full day of TEI’s 75th Anniversary Celebration, a panel of four professionals got out their crystal balls to predict how tax technology will impact the tax department of the future. The panel included Benjamin Alarie, professor and Osler Chair in Business Law at the University of Toronto and cofounder and CEO of Blue J, a leading legal technology company that leverages artificial intelligence to improve tax research; Tracey Grant-Castleman, a partner in the tax group at Crowe, who specializes in tax transformation, standardizing and optimizing tax processes, and building global teams; Ray Grove, who leads the Onesource team at Thomson Reuters as head of corporate tax and trade and advises clients on solutions to tax and trade challenges; and Mark Nadel, a director in PwC’s tax reporting and strategy practice, who specializes in process and data automation and frequently advises organizations on global transformation initiatives. Daniel Smith, head of TEI’s Silicon Valley Chapter, moderated the discussion.

Nadel noted a recent and increasing focus on data used within tax departments. “I think there are a few main drivers for this, the first being upgrades to cloud-based ERP [enterprise resource planning] systems,” he said. “This is really giving tax the opportunity to take a step back, think about where they may have current gaps or issues with how the current ERP was configured, and identify opportunities to get their requirements built in as they’re doing the next release. Additionally, the increase in transparency reporting, additional data required for transactional reporting, and more pressing now is our new global tax rules like Pillar Two. This increase in demands on the tax department really means that tax needs to rethink the role that data plays within their function.” He recommended that tax departments consider the various data sources they use now and may use in the future and to locate gaps or areas where team members must spend time doing data reconciliations, among other challenges. They should also consider the outputs their tax departments must produce. “And really the goal here,” Nadel said, “is to have a single source of truth that tax can rely upon for all their reporting and planning needs. From a tax standpoint, obviously, budgets are not increasing; if anything, they are decreasing or staying the same.”

Panelists (from left): moderator Daniel Smith, head of TEI’s Silicon Valley Chapter; Benjamin Alarie, professor and Osler Chair in Business Law at the University of Toronto and cofounder and CEO of Blue J; Tracey Grant-Castleman, partner in the tax group at Crowe; Ray Grove, head of corporate tax and trade and leader of the ONESOURCE team at Thomson Reuters; and Mark Nadel, a director in PwC’s tax reporting and strategy practice.

Grant-Castleman agreed about the increased complexity, volume, and speed of reporting. She added, “I think the other thing obviously that we’re all feeling, seeing, reading about is the workforce. Somehow there aren’t as many people who want to become CPAs and work on tax accounting. What do you do about that? How do you use tools, automation? How do you supplement that? What does a team look like? Additionally, post-pandemic, we’re all working differently than we used to. The combination of all these things really underscores the need for efficiencies in your processes and in your technology to be able to execute, do what we need to do, but also balance out all of those dynamics.”

Grove called the overall situation in tax technology “a bit of a mess.” Specifically, he said, “To the point you were just making, [with a show of hands] anybody here start their scholastic career thinking they were going into the world of tax? Notice I didn’t raise my hand—the world’s changing, and it’s changing fast. I say a lot of times taxologists need to become technologists. Who here has a meeting with their IT department at least once a week?” Surveying the room, he noted, “I’m surprised it’s that low; more of you should be raising your hands. The change that I’m seeing is it’s not just businesses, not just tax departments that need to get better at technology.”

Alarie cited another big trend: the rapid evolution of many of the new artificial intelligence systems coming online. “This is in the media—you see this all the time,” he said. “ChatGPT came onto the scene less than a year ago; it’s already grabbed the attention of so many of us in so many different ways. That’s just one example. You mentioned things are moving extremely quickly. They are. Google/Alphabet is going to be releasing this Gemini model soon, already has Bard. Anthropic has Claude 2; I’m sure Claude 3 is on the way. GPT 3.5 has already yielded to 4; I’m sure it’s going to yield to 4.5. All of the foundational models, these AI large language models, are improving really, really rapidly.”

Technologies and System Solutions

Nadel agreed that generative AI was at the forefront of new technologies. “I mean, you can’t go on LinkedIn and scroll for a couple seconds without hitting an article talking about GenAI and the impact it can have on organizations,” Nadel said. “So, what is GenAI? GenAI is a subset of deep learning where you’re training a model to generate new data or new content similar to the data that was used for the initial training. Examples of this would include visual, which would be [generating] images or videos; auditory, generating sounds; or text generations of text or code. And while it may seem like this technology has really appeared overnight, it has been around for a number of years now. The main difference is now there are chat interfaces that [make] the technology much more user-friendly and allow the end user—it could be tax professionals or developers—to be interacting and using it, versus previously where that would have been limited to someone that was either a data scientist or had more specialized experience with machine learning. GenAI models can be very powerful and ingest very large data sets. This could be financial statements, tax regulations, tax return data, to generate new insights. Tax departments now can take that information, take those prebuilt models and load their own organizations’ data into it, thereby making the results and the outputs that are generated much more customizable, specifically for your organization. A couple of use cases that I’m seeing are in more high demand at this point, the most common being chat. Think of this more as your typical Q&A, having your own personal chat research assistant. Document interrogations, this would be the ability to read and ingest data from PDFs such as tax notices. And lastly content generation, so, the ability to produce new reports or outputs, ultimately, like Pillar Two disclosures. Right now I’d say the technology is currently at the level of a foundational tax professional. So, the work would still need to be reviewed. It’s not going to be 100 percent perfect, but it’s a very strong starting point, and you’re able to produce content at a very, very quick rate and really accelerating the time that it takes to produce outputs.”

Grove cited AI as a difference maker as well, noting, “We’re learning a lot. Fundamentally, like every good AI model, it comes down to the content and data. It comes down to the approach that you’re taking, and that’s an important piece, especially in the world of tax, where accuracy matters. Being ‘kind of correct’ isn’t really good enough. I think there’s a lot of focus on that.”

Human Impact

Grant-Castleman noted that tax departments have tried for quite a while to figure out how to integrate technology into their processes, how to change their teams’ skill sets so people know how to use that technology, and how to think about data. “Depending on your organization’s size, complexity, and global footprint, you’re probably somewhere on that journey already—starting to look at your team, building technology skills, using data differently, and thinking about how to allow the data to flow between systems faster with less manipulation,” she said. “Depending on where you are in that journey, it’s going to get faster for sure, but I think it’s a really exciting opportunity. For me, as a tax person, and I’ve got to believe for a lot of you in the room, it’s been frustrating at times in my career to spend so much time looking backward. You’re gathering last year’s data, you’re manipulating last year’s data, organizing, calculating. Now you’re reporting. It’s important—we have to do it—but depending on how much manual work there is to get that done, you have less time to focus on the current year and next year and the strategy for beyond that.”

She added, “The other thing I’m thinking about is the anatomy of a team. So, an engagement team or a project team, what does that look like now and in the future? You’re going to be delegating tasks to AI as a team member. It’s thinking about who, what, where you’re doing the work. Global teams—across time zones, leveraging technology to really get that twenty-four-hour work clock going and get the job done. It’s really an opportunity. I think it’s a really exciting time to just empower your people, yourselves, and then just take a step back and go, ‘OK, I know I need to develop some skill sets. How am I going to find the time and the space and the capacity for that?’”

Potential for Missed Opportunities

Grove said that all businesses are experiencing financial transformations right now. “The systems and operations, the financial transformations, moving to cloud ERPs like you were mentioning earlier, these things are huge opportunities for the world of tax,” he noted. “I’ve stated, ‘You’re becoming operational to your business.’ You are operational to your business. Tax is becoming operational to how you do business. If you think about it, every business has commerce that they are transacting, right? Every product, every service, every asset that you transfer—all of that activity in the process of just existing as a business has a tax implication. You have all of these regimes saying, ‘I want more information. I want more data. I want it sooner.’ You look at those transactions, and those transactions ultimately end up being what goes on a provision or what ends up getting filed or winds up being part of that tax process. So, your organization is going through a financial transformation, and [if] you’re not part of that in the very beginning, that’s a massive miss for your organization. Chances are, you’re going to get saddled with something that somebody thought was going to be really helpful for you and really great and help you do your job, and it’s not going to be good enough for you to even manage the basic risk and responsibilities of your business and those basic obligations. You have to get that seat at the table at the very beginning.”

One challenge, Alarie said, is that the new technology feels somehow different. “It seems alien,” he began. “I’m actually heartened to see how many of you have actually been playing with ChatGPT and other models. Half of you have. Half of you raised your hands; that means half of you have not. I think this is absolutely happening. The capability of these systems is improving very, very quickly. Virtually nobody is an expert in these systems in tax at the moment, because it’s all so brand-new. The risk is that you don’t do the work to familiarize yourself with what these systems can do. It’s a completely solvable problem. It’s very easy to get up to speed on what’s happening here. As I mentioned earlier, the work is of folks like us on the panel to bring this technology to you and to make it easy to understand what it can do, what the advantages are. I don’t think we’re going to see a simplification in tax in our lifetimes from a legal perspective. I go to academic conferences; I’m a tax academic. I do a lot of tax writing, a lot of speaking from an academic perspective. There are tax law professors who keep banging this drum that, ‘Oh, we need tax simplification. We need dramatic simplification of policy.’”

The underlying tax system is complicated now but, Alarie asserted, will become orders of magnitude more complex in coming decades. That said, he continued, “the technology is going to be much better. So, the paradox is the underlying systems are going to be way, way more complicated, but as operators of the tax function, your job is going to become easier, and it’s going to become a judgment job.” The risk, he added, is not understanding this truth about the function.

Emphasizing the heightened complexity of tax in recent years, Smith, the moderator, noted, “There’s more data involved. Audits require more data. The type of new laws coming into effect don’t make compliance easy or simplified, as we’ve been hearing for a very long time, with people saying we should simplify the system. That’s not what happens. We have a lot more complexity in new laws coming out. The technology is just going to let us bridge the gap.”

Tasks for Tax Department Leaders

Smith asked the panel how tax department leaders and tax professionals should think differently about data, given these trends.

Grant-Castleman noted that tax departments will not simplify the tax system, but could simplify both how tax professionals report and what they need to report. “I think there’s still a need to look at your data structure, have governance controls, standards, and centralize as much as possible so that you can effectively leverage the technology available to work with that data. I think it’s foundational, once you do that work, to be able to adopt faster,” she said.

Concerning the steps tax departments should take, she cited the importance of knowing what problems need solving. “GenAI is super flashy; everybody’s talking about it,” Grant-Castleman said. “I’ve had conversations very recently with people who are like, ‘OK, tell me where to use it. I want to use it. I want to know I’m on board. I want to be an adopter.’ So, that’s great, right? It’s cool to think about that, probably spark some creativity, but if you don’t know what you’re actually trying to solve, you could end up with a costly problem on your hands anyway. For me, the advice is that you’ve got to take a step back and make sure you’ve really thought about the problems you need to solve in your tax function.”

Nadel noted that tax leaders may not see immediate success. “You may fail a bit. Factor in lessons learned and do better the next time. I would say do a four- to six-week sprint; that way you can just kind of jump in, get experience within your teams, within your organization. Focus initially on just general learning, identify use cases, focus on value and complexity, especially for the initial round. Don’t do anything too complex; do something that’s a bit easier, more of a chat-based solution. And then learn. As you learn and as you iterate and you start moving on to more complex use cases, you’re going to have a much stronger grasp. You’re going to have awareness within the organization, and you’re going to use that to build the platform up,” he said.

Assessing the larger picture, Grove said, “Tax isn’t about collecting revenues; it’s about socioeconomic engineering. That’s why it’s complicated if you really think about it. You need to look at the trends of where that’s going and wallow in that problem, of what that means for your organization. I think you need to ask for help [from] other parts of your organization.”

As promising and exciting as advanced technology such as generative AI seems to be, Grove added, it is only as good as the data and the content it works with. Addressing the tendency of AI to “hallucinate,” that is, to produce logical-sounding but incorrect outputs, Grove noted that “there’s a lot of things that you have to do to ground [generative AI], to responsibly incorporate it into processes and applications. I think that would be my advice: wallow in the problem, ask for help, and find those quick wins, but really get an understanding of your situation.”

Finally, Alarie noted that tax professionals should be careful about embarking on capital-intensive and resource-intensive solutions. “You want to do your homework. You want to really, really make sure, before you splash out a big amount of capital on some kind of initiative, that you really understand what the prospective ROI is,” he said. “I want to suggest something else, which is an interesting counterpoint to that at the other end of the spectrum: developing a lot of these tools is very capital-intensive, but the folks making the capital-intensive investments are the vendors, right? I have dozens of people working flat out on building this GenAI tax research platform. It becomes very easy for you to start trying it, . . . and I see how it assists in the tax research, tax planning, information gathering that you need to do in your organizations.”

Leave a Reply

Your email address will not be published. Required fields are marked *

XHTML: You can use these tags <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>