PodChats for FutureCFO: AI in finance as a compliance imperative

Speaker A: Foreign. Report reveals that a striking 83% of APAC AH CFOs see AI adoption as a key force reshaping finance, while 72% believe its impact will be significant within three years. As understanding what AI can do for finance teams, we are starting to see the adoption narrative move from discretionary innovation project to a necessary part of the governance and controls framework, a AH language CFOs and compliance officers understand understand all too well to be clear. Anxieties about moving fast in the adoption journey remains persistent, as is maintaining trust, a theme echoed in the Deloitte survey which found CFOs reinforcing fundamentals and cost discipline even as they invest in AI. In this Podcast for Future cfo, Nikhil Parambat, Regional Vice President for Asia at Blackline, offers some insights into how CFOs and finance leadership can fine tune their adoption strategies as AI moves from a nice to have to a compliance imperative. Nikhil, welcome to Podcast for Future cfo.

Speaker B: Thank you for having me Alan, and good to see you again.

Speaker A: It's a pleasure to have you back on program, Nick. Now, across Asia, CFOs are being asked to close faster and support real time decisions. Yet many closed processes remain, stitched together with spreadsheets and manual workloads. Where are finance teams in the region still most vulnerable to this spreadsheet dependence? And what specific risk does this create for a CFO's ability to trust the numbers in a volatile environment?

Speaker B: Great question, Alan, and I think it's a great reflection of what is actually happening on the ground. Right. The greatest exposure when you look at this kind of a scenario is with high volume time sensitive areas of the close. When reconciliations, late journals and intercompany mismatches are, uh, managed through isolated spreadsheets and emails, you end up creating data silos and blind spots that hide the emerging risks. So that is definitely one thing that really stems out from this use of spreadsheets and, and manual workarounds. The second element, when you look at the exposure as well as you look at the environment around us, right. The volatile conditions today demand continuous visibility and not just month and hindsight. So even a process is stitched together manually, finance is trapped in a transactional scramble and the CFO cannot trust the numbers or act as strategic advisors when decisions need to happen faster than what can be manually done. And in finance, eventually trust is the ultimate currency. So finance leaders, I mean, like even, you know, they cannot clearly see what is automated, what is manually registered, and where human judgment is applied, then actually the whole thing breaks, right? So relying on opaque manual workarounds fundamentally breaks the trust where it matters most. So this is where you see this, the areas where the teams are most exposed and therefore the underlying reasons why CFOs at times will struggle to trust the numbers that these uh, spreadsheets typically generate.

Speaker A: It's a scary thought indeed for a CFO if he or she is unable to trust the numbers that are coming out from the system. There's a term that Blackline uses, self driving close. Now in business terms, what does this operating model look like in practice? And what are the biggest misconceptions finance leaders in Southeast Asia and perhaps even Northeast Asia have about it?

Speaker B: It's a unique, uh, or a new concept if you were to look at it very pragmatically. It is an orchestrated operating model which is powered by what we call as agentic Financial Operations or afo. Now to make Agentic Close a reality, uh, we rely on Blackline's foundational elements. And what are these elements? There are three. A, uh, unified governed data layer, uh, ensuring very distinctly accurate data is driven by exceptional efficiency with end to end workflow orchestration, and finally financial intelligence with real time insights and embedded governance. So this is the foundation on which our uh, agentic financial operations is built, which allows you to kind of essentially walk through the entire closed. Right, and a series of activities and tasks that are required in pretty much a very seamless fashion where we kind of actually have AI agents, automation and humans all coexisting as part of a seamless process. So when you look at this, on this specific foundation, our AI agents like Verity prepare Verity Match. They allow AI to continuously prepare, analyze and execute accounting workflows on a single control platform which sits above the erp. And all this while the finance professionals remain firmly in control. So that's in a nutshell, the concept that we have with respect to the self driving close right now, where the biggest misconception is more about the fear of losing control. You know, the same hindsight as what we have when we first heard about self driving car. And like, you know, all of us are used to driving a car, our uh, biggest fear is about losing control. What if the self driving car just goes and does something that typically you and I would not do, having exercise our judgment, right. That is still the same fear and misconception in a way about losing control. But self driving does not mean removing the driver. We are just simply shifting from manual controls to algorithmic controls as part of that. And because of that element Human in the loop concept is non negotiable, right? The system autonomously handles the routine tasks, but the human remains firmly in the loop for complex exceptions, final approvals and accountability.

Speaker A: I don't doubt that we will go back and revisit some of the key points you just mentioned. At any rate, as agentic AI, uh, evolves from pilots or co pilots to autonomous actors, which specific close activities, perhaps reconciliation, journal entries, intercompany matching or variance analysis, which of these are the safest and most impactful to automate first, giving finance teams the fastest return in confidence and efficiency.

Speaker B: So the best way to kind of look at this is to focus on an outcome based approach and typically start with either your biggest pain point when it comes to the close activities, or maybe the greatest volume of data that is involved with respect to the close activities. Because either one of these are uh, actually going to solve uh, problems that currently you have with respect to your speed and efficiency and productivity in the context of getting to a reliable financial close. So we've seen customers where basically they take high volume reconciliations, like bank reconciliations and so on as being the most pragmatic starting point because they typically tend to be either manual or semi automated. And so it's a great place to start and reap the benefits and showcase the rest of the function, the efficiency benefits that you get as much as governance and control that comes along with that. Right? In fact, when you look at some of the early adopters of Blackline's Verity Prepare as well, where they use Verity Prepare for performing account reconciliations using this AI agent, customers are actually reporting a 94% reduction in manual preparation time. So that is again a very significant reduction and efficiency gain when you look at how we can use these tools to actually make things a lot more better and efficient. So using AI based tools to analyze and provide insights on existing data is also another quick way to determine what controls need to be in place and where existing gaps are. So, so we could start, like I said, with either the pain points, a uh, big pain or a high volume area. But you could also use AI based tools or AI analyzers to kind of actually analyze existing data. Which means that I'm not doing anything else outside the process, but I'm actually using AI to help me analyze what I'm currently doing. Right? It helps you figure out where your gaps are. Uh, it helps you figure out where you do not have suffic controls. Again, Blackline's customers use one of our tools called Journal Swiss Analyzer to analyze their entire universe of journals and detect anomalies way before they become a pattern. And in that fashion you have been proactive about how you kind of catch things before it actually gets caught by your auditor or sometimes you just totally miss that out. So, so again, another way in which you can kind of start on without actually doing something dramatically different to your workflow and your existing processes, but actually using AI from the perspective of your play analysis and in subsequent corrective action once you've done these fundamentals, then you can obviously follow with what we believe are a bit more complex in terms of workflows, which is primarily around uh, the entire journals workflow or could be around your intercompany reconciliation workflow. So Here again, AgentIQ AI can propose entries and manage the back and forth confirmations while keeping a perfect audit trail. But for most customers we see that as being something that is slightly further down the journey because you start with the high volume or the greater pain point activities first and then you kind of move there. And one of the best practices that we would recommend is to actually go on a sequenced approach rather uh, than trying to do everything on a big bang. Because this way you can at least get fast measurable roi. And throughout that journey at least you also ensuring that there is transparency and governance mandates are totally met as well. So that's how I would recommend having looked at how our customers are uh, actually approaching the journey with respect to AI adoption.

Speaker A: As CFOs gain greater experience using artificial intelligence, it's becoming clear that agentic AI itself is only as good as the data it operates on. Before a CFO can confidently let parts of the financial close go on autopilot, what critical data process and control foundations need to be in place to ensure governance, security and an auditable chain of trust.

Speaker B: So Alan, I think the answer is in the question itself, right? In the sense that uh, essentially foundational data readiness is everything from the time that we have had systems in place. And it is even more true now as we kind of uh, fully embrace the age of AI. Because you cannot scale AI on top of bad data. AI is just going to churn out even more bad outcomes for you on that bad data at a much faster clip than you and I can even imagine and monitor and really correct. So fixing data silos and unifying data governance across legacy systems must happen first. I mean this has always been true in a way, even previously prior to AI. But I think with AI it is really a sacrosant, it is foundational and this needs to be Done without doing this, there is honestly no tangible benefit at an enterprise level, uh, with respect to actually doing AI projects. The second element in terms of when you look at these, uh, the control element is to actually embed governance into the workflow. Because when you look at AI, I mean it's easy for us to kind of start trying to do many, many things with it. But what happens or typically with these smaller projects that people, you know, we' seen the certain finance teams embark on, is that while they are um, great from the perspective of showcasing the power of AI, they are totally lacking the governance and controls that is required with respect to anything that we do from a finance and accounting standpoint. Balancing aggressive AI adoption with strict governance by having configurable threshold risk scoring and mandatory human checkpoints. That is again next thing that we really need to kind of make sure we are addressing as a core foundation before we embark on full autopilot mode. Eventually the goal is to actually have what we would call as controllable autonomy. We want AI that is entirely auditable and explainable at every step, backed by enterprise grade governance and standards like what we have ISO 42001 certification, which basically ensures that the CFO office maintains the absolute authority. Now Alan, the key here is this, right? You see, the more AI you deploy in finance, the more a centralized system of finance controls becomes crit. And that's important because you can keep on deploying AI in silos or in whatever fashion you might want, but the more you deploy it, the greater is the need to actually have a system of finance controls around it. And Blackline is again in a prime position to actually be that platform as most of our customers use it currently. But eventually what you're having is you have a workflow that has a governance embedded M which runs off trusted data and has human in the loop oversight where required.

Speaker A: And um, thank you for that, Nick. Now here at future CFO, when we speak to CFOs, many of them these days are looking for that controllable autonomy direction that they would like to achieve anyway. As the system starts handling the heavy lifting of routine tasks. The finance professional's role is said to be shifting from processor to validator. In these days, uh, strategies, how do you see the role and skill set of finance teams evolving over the next two to three years? And how should CFOs prepare their people for this transition to an advisory role?

Speaker B: Spot on, Alan. I mean there will be transition and there is an ongoing transition and that will keep on evolving. So finance professionals will Become supervisors of digital intelligence. Uh, that's the way we would look at it, right, because the era of pure transactional number crunching is over. I mean that was the case previously as well, and that's more so the case today. So what we need purely from a skill set perspective therefore, is the finance team, in addition to actually obviously being the experts when it comes to finance and accounting domain, now must also have detailed knowledge of the underlying business operations. This becomes fairly critical and expanding the horizons out to actually understand how the business operates is key for them to be actually able to drive strategic insights for the business. And when you look at it from that perspective, it's also a very critical talent retention strategy. I think we did speak about this previously as well. And uh, you've spoken about this in terms of the talent shortage when it comes to finance and accounting professionals. And you know, the top talent doesn't want to come in and do manual data entries, right? So elevating their roles through AI directly impacts our ability to attract, retain and appropriately compensate high value strategic thinkers. Uh, so that's all one rolled into one. Then the second element is more the technology side of things in the sense that now that you are embracing more and more AI, what does that do to the people that you have? You need people now in the team who are fundamentally very good at data analytics. So upscaling teams with foundational data analytics knowledge is critical so that they understand how AI reaches its conclusions and evaluate the underlying data with respect to what AI is generating. So that's in the area that we definitely foresee that CFOs should be preparing their teams on. I mean outside of the realm of that and depending upon what other AI CFOs intend to evaluate, obviously an element of prompt engineering is also critical because that, that allows you to actually ask the right questions from whatever AI models you might be using outside of standardized solutions like Blackline. So again, we foresee that as well as another skill set that needs to kind of come more into the mainstream because once you're good at that, then you actually are asking the right questions. Uh, coupled with the business knowledge that you already have, uh, uh, you know, all of those things help you really, really take advantage of the power of AI and leverage the insights on a more real time basis.

Speaker A: Alan, now since you brought it up in a region marked by rapid digitalization, yet persistent skills gap, what are the key implementation challenges for CFOs in places like Singapore or Hong Kong, Malaysia or the rest of Southeast Asia and even Japan, who are Trying to scale AI LED finance.

Speaker B: So uh, that you know, the wide gamma of countries out there in our region. So the simple biggest thing that has always been an issue when it comes to scaling up is the very complex regulatory and accounting compliance landscapes. Right. So in these markets it's not about just scaling AI from a technology challenge perspective. It is actually about ensuring that the AI's algorithmic controls meet the rigorous localized compliance requirements of distinct finance regulators. So that's fairly critical cycle because while I uh, might be an organization that's operating across the region, it is not going to be one size fits all. So I need to kind of just make sure that when I'm scaling it up I'm also scaling up in line with the local regulation. So that is definitely one key challenge that we find when it comes to kind of scaling up. The second element, and we touched upon that before, is more about legacy ERPs and very complex multi entity structures that characterized a lot of large organizations in our region. So again, integrating AI without first resolving the data silos, they would create a massive integration risk and again will not let you scale up the way you should ideally scale up. And uh, I won't draw a distinction. You can still scale it up by using other technology solutions. But the reality is you will then miss out on the governance and control element which is fairly critical when it comes to finance data. So it's important that you kind of resolve these data silos. Right. And then there's the governance gap as well. Whenever I speak to CFOs and leaders in our customers and also with our process aspects, they are all eager to want to transform, but they are hesitant to let AI touch higher risk activities because they lack the framework for AI governance and explainability at this point in time. So again, uh, that is a challenge, uh, and we are slowly starting to see organizations actually establishing AI governance frameworks, leveraging of frameworks that exist either with the ISO 9000 framework, ISO 42001 framework I mentioned, or even in Singapore for instance there's a government framework that's been introduced by imd. So they are leveraging of those frameworks, but that is still work in progress. Of course one simple way to kind of do this and overcome everything is to actually have a tailor made platform for financial operations like blackline. But these are the challenges uh, that we are seeing with respect to scaling up across the board.

Speaker A: As AI agents become more autonomous, questions of accountability arise. How can CFOs adapt their internal controls and compliance frameworks to effectively govern AI ensuring That autonomous close meets regulatory standards.

Speaker B: Absolutely. Alan, before we get there, I just want to kind of call out one thing here. See, I think we all get so much caught up with the AI hype in a way. I think it's sometimes useful to just take a step back and remember that not all operations need to have AI. When you look at all of finance operations, right, they are typically rule based and for anything that is rule based, typically basic, or the automation tools that we have always had prior to AI works very well in addressing these requirements. So it's important that we kind of decide judiciously as to where do we need AI and where do we need automation. You can deploy AI very selectively, let's say in areas where there are very high data volumes. You can deploy them very selectively, let's say when it comes to judgment based decision making based on past patterns of a human in the loop, had done previously and learned from it, and then apply that decision making the next time around. So that's again a great place to kind of basically use AI. Uh, you can use them, use AI, ah, for real time insights, uh, in the context of the larger domain of the business, business. So there are elements where AI is a good fit and, and makes very sense, great sense to use it and there are areas where automation is a preferred alternative. And that's exactly the approach that we have taken at Blackline as well. So I just want to kind of put that out there because it's important to kind of just draw that distinction. Especially because of the fact that when it comes to finance and finance data, uh, it is important to understand that, you know, finance is not a probabilistic function, finance is a deterministic function and AI is a uh, probabilistic function. And so that you important to kind of therefore figure out where AI makes most sense in the context of what we're talking about. Now going back to the question itself, uh, Alan, obviously in terms of controls, the biggest shift is from manual to algorithmic controls, right? So now that we can actually have an engine or a layer which kind of is a lot more intelligent, we can kind of move from having manual controls to actually algorithmic controls. Um, so areas like let's say post close sampling is no longer enough, controls must evolve to continuous monitoring and real time risk scoring. So and that is feasible and possible today. So at least that is one area where we can kind of look at it. We spoke about this previously. It's about designing compliance directly into the workflow. So automated segregation of duties, comprehensive audit trails and 100% traceability of every action that agentic AI takes. I think that is extremely critical. We keep on talking about how it is important that AI, uh, actually is more of a glass box where it actually explains all the actions that it has taken so that you as a user, you as a business stakeholder, understand why a certain decision has been made. An auditor comes in, he or she can also look at that same data and from uh, of the, of the AI explainability and then understand, okay, why AI made that decision. So it's very, very important to actually have that and finally again reinforce a human in the loop. I mean, at the end of the day automation should actually make accountability more visible. So to that extent it is not about removing humans totally and have an end to end autonomous process. It is about making sure that at the right levels in that workflow, we actually have humans in the loop. So that any exception handling specific area where human oversight is still required continues to be there and eventually approvals in most cases, unless they are below certain threshold of predefined rules of materiality actually handled by humans. That is fundamentally the third element that I would want to bring out in the context of this particular question. Alan.

Speaker A: Thank you, Nick. It's interesting how you are one of those few executives who've actually told me that not everything needs to be AI'd anyway. Finally, how does achieving a self driving close free the CFO's office to focus on what matters most, such as strategic analysis, partnering with the business and steering the organization through volatility?

Speaker B: I agree. I guess every day you wake up and you're like looking at something and figuring out, oh, okay, I didn't anticipate that. And it's so tough. You can't, you can't anticipate everything. But having said that, Alan, see, think about it this way, right? When you look at a traditional close, it is a month in, month out activity that the finance and accounting team undertakes. It is a rigorous activity because it is eventually an activity that results in your financial documents or statements. Those statements are what eventually your organization stand behind. That's what, what your stakeholders, your investors, your regulators look at and say, okay, that, that is how well you're performing as a business. So it is all about trust. And which is why it actually takes time. Uh, which is why it is not something that is taken very lightly because that is fundamentally the core, a core of how an organization reports its performance to the larger audience outside. So when you get to the point where you can now automate a large of that closed process, you can actually take elements of what used to be a human decision making and based on the understanding of what the human did previously, now automatically apply that going forward to similar instances in the future. You are actually taking away a lot of the time that goes in terms of actually executing the close. And what does that mean now? Uh, that means that fundamentally I am now in a position to be actually slightly more more strategic driver in terms of the business. By automating all this heavy lift we by eliminating the month and fire drills, we are giving the CFO's office now a lot more time back for them to do what they were originally supposed to do but were never able to do because of the constraints of the systems per se. And essentially you can now drive a much broader business impact. In fact, one of our customer enterprise customers noted that finance team want a partner partner who has their back. So and so with a trusted AI foundation, now partner finance actually has somebody who can have their back and that means that therefore now they can effectively partner with other departments like sales for instance, analyze cost structures and assess business impact and drive growth strategies. And while all that is great, the biggest benefit would be the fact that you will now have faster visibility. You know that a lot of organizations actually have time targets towards how quickly can I do I want to close my books now? Why do they want to do that? They want to do that because the faster I close my book, the earlier I have visibility on what is working and what is not working. With autonomous close or self driving close, you now have almost a real time visibility with respect to how your business is performing and when you have real time insights into how into cash, into risk, into performance. It allows you to shift from reacting proactive, you know, just reacting to past data to now being more proactive about managing, managing market volatility. And in that fashion you actually become a uh, true and trusted business orchestrator, helping guide the organization's strategic direction.

Speaker A: Alan, Nick, as always, it was a pleasure having you on podcast for future CFO Alan.

Speaker B: The pleasure is all mine. Thank you for having me.

Speaker A: That was Nikhil Parambatt, Enterprise Application Sales leader and Regional Vice President for Asia at Blackline on the topic of AI and finance as a compliance imperative. You are listening in the podcast for future cfo. As always, if you have a topic you'd like us to cover on this channel, simply email us at. We'd also like to invite you to sign up for our free weekly newsletter so you won't miss an episode of podcast for Future cfo. In the meantime, stay safe, have a great day, and see you on the next episode of Podcast for Future cfo. Bye for now.