Paul Dostaler:
You have to accept that when using AI, you're going to have to experiment. And when you experiment, you'll have successes, you'll have failures. But you need that flexibility to adapt and buy-in from your leadership to make sure that you can go through the experimentation and chart your path to success.
Anne-Marie Henson:
Hello and welcome to Accounting for the Future. I'm your host, Anne-Marie Henson. Today we're exploring a widely discussed topic through an industry-specific lens. So we're discussing how to scale AI, but specific to the manufacturing industry. I want to welcome our guest today, Paul Dostaler, who's a partner and our National Manufacturing and Distribution Leader at BDO Canada. Paul has over 20 years of experience leading development teams in areas of connected services and IoT, application transformation and solution architecture. His specialties include assisting in technology adoption and using data to realize business ROI. Paul, thank you so much for joining us today.
Paul Dostaler:
Thanks for having me, Anne-Marie.
Anne-Marie Henson:
Good. I know we've talked a lot about the importance of the manufacturing industry in Canada. It's undergoing so many changes and obviously is quite impacted by current events today as well, which compounds the issues about staying profitable and being able to innovate and adapt for the future. But before we talk a little bit about that and dive into it, I wanted you to provide a bit of an overview on AI. And for our viewers, we have had a couple of previous episodes on AI, more fundamentals and how it can be used to improve the finance function and governance around AI. What I'd like to hear from you, Paul, is a couple of years into this real big journey on AI and everyone's talking about it. What would you say today is the rate of AI adoption right now? And what trends are you seeing in the market?
Paul Dostaler:
Great. Yeah. So, in terms of the rate of adoption, especially in manufacturing, but also in adjacent industries, most reports are saying that Canada is about 10 years behind in its adoption of technology in general. And that's true for AI as well. That surfaces in multiple different ways. But if you've been paying attention to the media for the past couple of years, that's mostly exemplified in our lagging productivity. So all these things are connected. You're tempted to draw a link. We won't do that here explicitly today, but we can talk about how we're seeing a disproportionate amount of investment into AI across different countries. So if you look in the US, you've got huge projects like Project Stargate that are structured to put together an investment of over $500 billion into AI infrastructure as compared to some of the similar initiatives that we have in Canada that are really just a few billion dollars. And so that's sure to have an impact over time.
One of the trends that I wanted to quickly mention was that AI itself is evolving. And so traditionally, we would've talked about things like machine learning, which is still prevalent today, but we're seeing a trend where even Gartner is talking about how generative AI is the most deployed type of AI into production. We'll talk a little bit more about that during our talk today. The other trend is that in manufacturing, more and more manufacturing are choosing digital transformation as a mean to adapt to all of the current event and the changing environments. We here at BDO actually came out with a survey leadership report called Momentum here in 2025, and it gave us a few additional insights to this point. So in that survey we saw that 33% of respondents were actually already in a digital transformation journey, and presumably that has an AI component to it as well.
And of that 33%, two-thirds were already seeing a return on investment, which is very positive. Now, if you dig a little bit deeper into the survey, maybe 33% are in a digital transformation, but it is actually only 11% of respondents had no intention of doing a digital transformation at all. So we're talking about mid 80% of Canadian manufacturing organizations that are in intentional about adopting technology to help bring efficiency in general. The other last little tidbit that I thought was interesting is in a separate line of questioning, we asked how many respondents were already using AI tooling regardless of the digital transformation? And 48% of respondents actually indicated that they're already using AI within their business. So I thought that was really great.
Anne-Marie Henson:
Oh, that's really interesting. 10 years seems like a lot to catch up on, although it seems like today with the speed of technology that it is a gap we can close. But I want to ask you a little bit more about that. What do you think is causing this lag in AI adoption and investment in Canada specifically? What kinds of challenges are businesses facing in terms of their adoption?
Paul Dostaler:
Yeah. So the approach to AI isn't necessarily what we'd call a traditional approach. And so I think the businesses are struggling to understand just that, how do we adopt it? How do we jump in? How do we invest? How do we take that risk? So if you look at the major research bodies, Gartner specifically, was talking about how estimating and demonstrating business value for AI is the biggest leap, the biggest challenge in terms of justifying that investment. And so, you'll see that a lot of leaders are talking about AI, but they're a little bit scared to jump in. But recognize that AI is different than other technologies. It's not just a standalone tool, it's actually more of a muscle. And so, the more that you use AI, the better it becomes and the more potential for ROI that it has.
And so, whereas previously being a fast follower might've been a great strategy in terms of employing a new technology. We're finding that with AI, you really want to jump in as soon as you can and you jump in with the lower risk use cases so that you can kind of wet your feet and get used to using the technology. But it's important to get started. Just a little bit more on that because we're creatures of habit and so we don't want to downplay the big hurdle that is change management when it comes to adoption. And so, you have to accept that when using AI, you're going to have to experiment. And when you experiment, you'll have successes, you'll have failures. But you need that flexibility to adapt and buy-in from your leadership to make sure that you can go through the experimentation and chart your path to success.
Anne-Marie Henson:
Yeah. I'm happy you said that because a lot of people talk about the importance of AI and it being on the technology side or the digital side of things, and the importance of integrating it in your business. But at the end of the day, most every company is still run by humans who need to accept the change, not just accept it, but embrace the change and see the value in it. Which I think for some people, you don't know what you don't know until you see it in front of you and you really see the true benefit of using a tool that's AI-driven or AI-generated to be able to help in your day-to-day job. But I do think that there's still a little bit of skepticism out there in terms of real industry experts and people who've been in the business for 30 years in accepting that this tool could help make them better. Right?
Paul Dostaler:
Yeah, we see it all the time.
Anne-Marie Henson:
So I know we're here talking specifically about manufacturing today, you being one of the leaders at BDO Canada, but I know that you have a background in many other industries. So I'm curious to ask you a little bit about maybe some industries where you see AI could be really beneficial where we haven't unlocked that value yet?
Paul Dostaler:
It's funny because I am going to talk about manufacturing because we saw, for example, the financial services industry adopt AI very early on. And it makes sense, especially in the area of fraud detection. They have a lot to protect and so they're getting very, very sophisticated there. We then saw AI trickle into retail and consumer business. Really sophisticated ways to go-to-market and use techniques like sentiment analysis to gauge how markets would be reacting to your products and be able to increase sales accordingly. But it's really with manufacturing where I feel that we have the most opportunity everywhere from on the plant floor with robotics and physical machines up to scheduling and ERP. And then in the back office with finance, with legal, it's really applicable everywhere. And because of that lag, I feel like it's our biggest opportunity.
Anne-Marie Henson:
Yeah. Well, it's good to hear and good to see that you're at the forefront of it helping our clients especially. So let's talk about a business that approaches you and wants to adopt or even scale AI. They might be using it a little bit, but they don't feel like they're really maximizing the value they're getting out of it. What considerations would you say they need to take into account?
Paul Dostaler:
Yeah. I think it's important to look at the business as a whole. So we've been trained over the years to look at classic use cases for AI, and again, the predictive capabilities of AI. But now with the advent of generative AI, again, it's applicable throughout the business. So, look holistically at your business and think about the parts that are taking you the most time and start thinking about what more sophistication around automation and generation might help speed up those workloads. So we know that CEOs know that in order to be competitive with their business, they need to use generative AI to create those efficiencies and reduce those transactional workloads that can be accelerated through AI.
So again, think about which parts of the business can be assisted by AI and then improve those efficiencies. Realizing a little bit of ROI at a time and building upon it. The other consideration is that AI works on data. And so there's a high degree of centralization that's really needed to get the best out of AI. And so as you think about digitizing more and more of your business, try and reduce those silos to make sure that your corpus of data is well-connected so that AI can draw insights from across the entire organization.
Anne-Marie Henson:
Those are all really good pieces of advice for sure, and really help for our listeners, I'm sure. And I want to talk a little bit about manufacturing again specifically and how AI can be used in thinking through the pain points of that industry. Is there a possibility to see it improve the logistics, warehousing operations? And are there any examples that you can give of work that maybe we have done, or you have witnessed in that space? And how does it tie... I am a CPA by background, so I'm a bit more knowledgeable about the financial reporting impact of things. How do you tie all the importance of warehousing and logistics and supply chain to better financial reporting?
Paul Dostaler:
Good question. And I have a view that essentially everything that the business does is in support of clean and effective financial reporting. And so, really, good inputs create good outputs and getting more insight into those inputs further create better results. So maybe I can talk a little bit about typical use cases that we've worked on and that I see frequently, but I'll dive into a little bit of IP that we've created at BDO that's really relevant these days. So typically, in manufacturing, we help customers with things like when we talk about the shop floor, we'll talk about predictive maintenance, we'll talk about anomaly detection to help identify situations where we might not be producing quality in order to correct. And then we'll move up the chain, I spoke to scheduling. So optimizing scheduling. And whether that's scheduling for your plant lines themselves or scheduling for your maintenance activities, scheduling for your support staff, schedule optimization is certainly an area where AI can help.
We then get a little bit more sophisticated, and we look at the area of supply chain. And that's so relevant today with all of the tariff talks and needing the ability to be more sophisticated to how you do your scenario planning with your supply chain. How do you estimate how diversification might impact your business? And AI can help there too. Again, more traditionally in terms of your sales, use machine learning to really try and forecast demand, forecast revenue and develop systems where you can play with what if scenarios. What if I change the price of certain my products? How is that going to impact my top line, my bottom line, et cetera?
So recently at BDO, we're on our own AI journey and what we've done is we've created this product that we call BDO Boost. So BDO Boost, it's a conversational AI tool. So, to the regular user that might look like a ChatGPT interface, if you've ever played with ChatGPT. But what's interesting is that it's ChatGPT, but in a private setting for your corporation. And so, what it allows you to do is create these rooms where individuals can collaborate in terms of uploading pretty much as much documentation as they want into one of these virtual rooms and then they can start having a conversation with their data. So they can use the power of generative AI like ChatGPT to run a number of different analyses.
So just a few examples of this is if you're an organization that deals with many, many contracts, you might want to upload all of your contracts into one of these rooms and then you can start to summarize what your total obligations are as a corporation rather than trying to depict it yourself from hundreds or thousands of different contracts. You might want to look at terms and conditions over time. This is an interesting one where if you have a lot of suppliers and they tend to tweak their terms and conditions year over year, sometimes it's really difficult to understand the nuance of what those tweaks are doing in terms of your risk exposure over time. So we found that these tools are really good for identifying that and summarizing it back to you so that you can take whatever necessary action.
We've seen legal departments upload incredible amounts of documentation to help identify which pieces of different testimonials, documents, chats might support their narrative in terms of evidence that they're trying to collect to support whether those are in a litigious kind of way or just for a claims scenario. And then another one that you might find interesting is in the world of compliance. We here in tax and assurance have all kinds of huge manuals that describe exactly what compliance might need in terms of submitting, creating various documents. And so this technology is also great at instantly understanding a compliance manual and then being able to identify areas of interest in other documents where it might be able to tell you or direct you towards sections that are at risk of non-compliance. So really speeding up the idea of finding the right places to focus on to do your work.
Anne-Marie Henson:
Oh, that's incredible. I'm curious, how long did it take us to develop this tool?
Paul Dostaler:
It's certainly not been two years that it exists, and it continues to grow. And so the tool was born within BDO, but today it's at a state where we can actually offer it to our customers as well.
Anne-Marie Henson:
I think that's incredible. And I just want to pick up on something you said previously that I think is great advice for companies when they're maybe a bit overwhelmed with where to even start or where they can get the most ROI in investing a little bit in this type of technology. Your incremental improvements, let's get 10% better at something today and then we'll build on it, instead of trying to think that you need to boil the ocean and do everything at once. So it sounds like this tool could provide a lot of great advantages to companies in specific areas that start with your biggest pain point, what relatively seems like it could be a simple task and ends up taking a lot of hours and focus on that and then build on it once you get a little bit of confidence.
Paul Dostaler:
Absolutely. And it doesn't require a full digital transformation of your business in order to be used. Because yes, it could tie into your network if that's the way that we want to use it, but it also works in a standalone way where I just really want to analyze one document or maybe five documents that I have in a folder, and so once the tool is stood up, it's ready to use instantly.
Anne-Marie Henson:
Yeah. Oh, that's amazing. In thinking through some of these items and benefits it could provide to companies, there is the risk or the governance side of things as well. And it does feel like AI has maybe been an area that has lacked regulation. Things are maybe changing so fast and there are other things on regulators' minds these days, so we haven't seen as much. But just in thinking this through, are there any regulatory or compliance considerations when companies are thinking about or starting to integrate AI into their financial and operating processes?
Paul Dostaler:
Absolutely. I mean, every organization is going to have their own kind of policies that they need to put in place regardless of the technology that they're adopting. But you touched on data governance. It's important to have strong data governance within your organization. Because the thing with AI is that you need to understand where the data is coming from, where it's being processed, and whether there's any opportunity for sensitive data to essentially seep out of your organization. So I'll give you a bit of an example. You may have seen in the news or in the last multiple years of how some organizations are suing OpenAI because they have proven that the OpenAI model used proprietary data of theirs in order to generate responses to questions. But the question there is how did that data leak into the model? It's difficult to understand where that comes from.
And so, there's the idea of ‘is my model, by training it on the data that I have, is it exposing the data to the outside?’ And so in a product like BDO Boost, we hear at BDO understand that very, very clearly, and so we can help answer those questions. But depending on which tool you are adopting, whether it's third-party off the shelf or building your own, it's really important to understand where those lie. Because AI typically needs to use the power of the cloud, which may be secure for you, it may have open doors intentionally or unintentionally. And so, a lot of rigor needs to put into that side of things.
Anne-Marie Henson:
Yeah, absolutely. It's funny because a lot of the benefits of AI are the repetitions and the data and the insights it gathers from gathering information from different sources, and yet it also probably poses one of the biggest risks. So, there's two sides to the same coin that make it a little bit difficult to navigate.
Paul Dostaler:
Exactly.
Anne-Marie Henson:
So I'd like to ask you then about laying the right foundation for this transformation for a company deciding to go on this AI journey and adopting it within their organization. There must be some things that manufacturers have to get right in order to have access to the most reliable information possible in making decisions for their organization. So can you talk through what you would say the top things that companies absolutely have to do correctly in order to maximize ROI from use of AI?
Paul Dostaler:
Yeah, that's a good question because it encompasses a lot of the tidbits that we've talked about so far. So the first thing that you need to get right is to pick the right use case or the right use cases. Don't boil the ocean, choose a practical approach that has a really high probability of demonstrating some level of ROI, and then build from there. Because again, you're trying to convince your entire organization, the decision makers in your leadership group, that there is value to this kind of investment. And so you don't want to take a huge risk right out of the gate. You want to build it in stages.
Secondly, we talked about data governance. Again, understand where the data comes from, where it gets analyzed, and which doors might be open or closed to allow data to seep through. Sometimes it's a favorable thing to expose some of that data to the more public sphere. But as long as there's no chance of your sensitive data being compromised or escaping your private network. And then finally, the one that I think that is the most important is do not underestimate the need for change management. So to me, AI is as much a mindset as it is a tool, and it's effectively a change in paradigm in how you approach problem solving. And so organizations really need to understand that adopting AI means experimenting, it means having successes and failures, and it means charting a path that brings the most benefit. And you do that by being flexible and adaptable to change. And that's really going to be your key to success.
Anne-Marie Henson:
Yeah, those are great pieces of advice. And I just want to touch on that a little bit. I think some of the organizations that I've seen that have done this right are ones where the AI adoption project or whatever they're working on doesn't just sit with the technology team or the information team within the organization. They've actually built committees from operations, from sales, from finance and accounting, from HR. And there's a contingent of individuals from different areas in the company that have input into not just what could work, but what are the risks? What are the blind spots? What are the changes that need to happen from a technology perspective and a human perspective to make these successful? And so sometimes I think where it's gone wrong, I've seen these types of projects work in a silo and then all of a sudden get deployed without having had stakeholder feedback throughout the process.
Paul Dostaler:
Absolutely. So especially with the advent of generative AI, this is where we see AI for the entire organization. So there's this misconception that because AI is technology, that it's your technology department that's going to benefit from it the most. But in some of these programs that we do here at BDO as well is that we'll take people from all departments, not only for the education piece, but we'll send them back out with new knowledge of how AI could be applied. And we'll ask them to go ahead and use the tooling to just automate just little sub-processes within their workspace and then bring them back to the group so we can share and learn and help spread what AI can do more quickly across the entire organization.
Anne-Marie Henson:
Oh, that's fantastic. I have one last question for you, Paul, and I guess we wouldn't be able to have a conversation today without talking about tariffs and economic uncertainty because-
Paul Dostaler:
I knew where this was going.
Anne-Marie Henson:
... it feels like a little bit top of mind these days. But just given what's going on, I think generally there's a situation and an issue that companies need to navigate right now that will pass at some point in the future, but I think it gives companies an indication that these types of events could happen more often just given the global uncertainty. So with that in mind, the current tariffs that are ongoing and just broader, the impact it's had on economic uncertainty for Canada, how can we leverage AI to either help businesses adapt or stay informed of all the changes and ultimately remain competitive and profitable?
Paul Dostaler:
Yeah. So I have a couple of thoughts on this thing. So firstly, I think that what this situation has created is it's created a bit of awareness of areas of our business where we maybe lack a little bit of sophistication. And when I say sophistication, it's our ability to do scenario planning around what happens if I change these parameters in my business? What happens if I diversify my entire supply chain into Canada and Europe, for example? And it's taking a long time for us to assess the impacts of that when really the technology that we have in terms of the analytics and the predictive capabilities that we have, we should have the tooling where we can predict with a high degree of confidence exactly the impact that's going to have on our business.
And the speed is so important right now because, as you know, there was a time there just a month ago where twice a day it felt like the world would change on us in terms of how tariffs were being applied and at what rate. And so, bringing sophistication in your organization was true prior to the tariff scenario, but the tariffs has just exacerbated the need for bringing that level of sophistication. And just like we're being bombarded with change and data, using tools like generative AI to better understand the data quicker and summarize it for human consumption is really going to give us that edge that we need to first be reactive to change, but then at the same time work on proactive processes and changes that we can make to better reduce the risk of being on the losing end of that change.
Anne-Marie Henson:
Yeah, absolutely. If I think back the past few years, this is especially true for manufacturing, but it definitely applies to other industries as well. Think back through March 2020 when the pandemic really happened and there were lockdowns and uncertainty about whether or not they'd be able to operate and were they an essential business or not. Then you had the issue of product and goods literally being stuck on boats in the ocean for months at a time where I know some companies decided to pivot and overstock inventory right when interest rates started going up and demand started decreasing.
So you would've hoped that this year would've been a year where they could stop playing defense and stop being so reactive, and then all of a sudden there's all these issues around tariffs. So I guess it's just a lesson that no matter what's happening, there's going to be disruption and it's going to continue to probably come faster and more often than it used to. So companies should just start trying to find the tools, to your point, about learning to be proactive and not just obviously having to react really quickly to these types of situations.
Paul Dostaler:
You nailed it. That's exactly it.
Anne-Marie Henson:
Well, Paul, thank you so much. I always love having these conversations with you and I'd really like to thank you for your valuable time and your input today. I hope our audience appreciated this discussion. If you liked this episode, make sure you leave a review or comment and click the follow or subscribe button to stay tuned for new episodes. Thanks to our listeners for tuning in today and to all of our episodes. I'm Anne-Marie Henson and this has been BDO's Accounting for the Future. We'll see you next time.