On site at SAS Innovate: Deloitte Canada’s Nat D’Ercole on Viya migrations, the data governance gap, and the 80/20 flip AI might finally deliver

Robert Dutt: Hello, and welcome to In The Channel from ChannelBuzz.ca, bringing news and information to the Canadian IT channel community for the last 16 years. I’m Robert Dutt, editor of ChannelBuzz.ca, and your host for the show.

This is our third and final episode from last week’s SAS Innovate 2026 in Grapevine, Texas. And if you’ve been following along, you’ve heard the view from SAS Canada leadership – the AI maturity story, the governance urgency, what the mid-market channel opportunity looks like – and then the global channel strategy conversation with John Carey, the build-out of the indirect motion, the TD SYNNEX partnership, and where the channel goes from here. What we haven’t heard yet is what it actually looks like from inside a real enterprise engagement. That’s what this episode is.

My guest is Nat D’Ercole, data transformation leader for AI and data at Deloitte Canada. Deloitte is one of SAS’s major global systems integrator partners, and Nat works with the kind of large Canadian enterprises that are right in the middle of the AI transformation conversation – Viya migrations, data governance strategy, OSFI E-21 readiness, risk and fraud modernization. The practitioner reality, not the roadmap.

We talk about what it actually looks like to walk into a client and untangle 20 or 30 years of SAS implementation. We get into data governance maturity as the thing that most determines whether an organization is ready for AI. We talk about what Nat calls human hallucination, and why it’s not as different from the AI kind as you might think. And we close on a concept that Nat has been waiting 30 years to see become real – the 80/20 flip.

Let’s get right into it. My chat with Nat D’Ercole.

Nat, thanks for taking the time. I appreciate it.

Nat D’Ercole: Pleasure to be here.

Robert Dutt: Obviously, you guys are one of SAS’s major global partners, but for an audience that’s primarily VARs and MSPs – that kind of partner – the Deloitte AI and data practice might be a bit of a black box. Can you tell us a bit about what it looks like day to day? Who are your clients? What are they typically asking you to solve today?

Nat D’Ercole: Of course. Our clients are facing complex issues in terms of how to manage their data, manage their models, and obviously working in an age of AI and sorting all that out in terms of where they are today, what are they using today, the cost of running all that today, to where they need to get to – both from a data, tech, people, and process perspective. So being a professional services firm focused on helping our clients with both advisory, implementation, and supporting our clients’ systems are key areas that our clients look to us for support.

Robert Dutt: A little earlier, I talked with Ryan Macdonald, who leads SAS Canada. The subject of hidden SAS came up – in so much as a lot of customers end up finding they’re running SAS software, running key business functions on SAS software, and not necessarily even aware of it, because it’s just become such a part of the underpinnings. It’s just there. It’s invisible even to themselves. When you walk into a client that engages Deloitte on, say, a Viya migration, is that something that you often see? And what does that journey kind of look like?

Nat D’Ercole: Great question, Robert. And that comment from Ryan really makes sense to me. Our clients have been using SAS for many, many years – some 20, 30 years, and maybe even longer. And so SAS is used for everything from data management, modeling, analytics, reporting, data wrangling, and so on and so forth. And it’s a web of solutions that organizations across departments have implemented. And so understanding what they currently have in place is a challenge. And so we do help them with that in terms of providing them with a scan of their current environment and helping them understand what workloads are actually running their business versus workloads that haven’t been touched in years. And with that, we’re able to help them with a roadmap to address those workloads and determine what is fit for purpose in terms of moving to a future state.

Robert Dutt: You guys are dealing with big projects and pretty high-stakes stuff, and not the simplest thing – like a Viya migration at enterprise scale is clearly not a simple concept. What do you see as the real cost and complexity pressure points for customers? And how do you help clients navigate those without the project stalling out?

Nat D’Ercole: You know, I think what’s really important is to understand – just building on my previous answer – understanding what is running their business and the cost structure associated to that. So obviously there’s technology licensing, there’s training on existing solutions, target solutions, change management, upskilling, etc. in terms of some of the key cost drivers. And let’s also refer to storage as well as another area of cost. So analyzing our clients’ environments and really taking a closer look at each of those buckets to help them figure out where are they now, and what are the opportunities, what are the options for them moving forward.

Robert Dutt: Governance – obviously a big topic here – and the idea of governance and trust becoming inseparable from the AI conversation has been a big theme here and elsewhere. Curiously, what are you seeing in that, and is it changing what you’re being hired to do? Are clients coming to you with a technology problem, or are they coming to you with a governance and risk problem that has a technology component to it?

Nat D’Ercole: Yeah, so clients are hiring us to solve a business problem that is enabled by technology, enabled by change. And to address your specific question around governance – governance comes in the form of data governance, AI governance, model governance, etc. We do find that the level of preparedness in organizations around data absolutely varies from immature to mature. So those organizations that have addressed data governance are those that are most prepared for the AI age and being able to take the next step.

Now, not everything requires structured data and highly clean data. So depending on the use cases, it is quite possible to apply AI and begin to see benefit. However, more and more I do see organizations invest in things like master data management, invest in data governance, and invest in operating models. And those operating models are also AI-ready. So we’re starting to see the need for roles such as prompt engineers, AI engineers that are interrogating results of models, ensuring that there’s a continuous feedback loop – and where models are drifting or hallucinating or so on and so forth, that there’s a human loop catching that. So these are new roles that are being created and need to be part of an overall governance strategy.

Robert Dutt: What role do you see yourselves playing in leveling up those organizations who haven’t gone far enough in governance thus far to get the most out of the AI future?

Nat D’Ercole: I’m actually working with a client right now where they haven’t addressed data governance and they’re stuck with legacy solutions where very much it’s been the wild wild west – if I could use that term – in terms of accessing data, enabling analysts across the organization to wrangle that data and develop outputs that their leaders consume. And so when that happens without governance, you get things like what I refer to sometimes as human hallucination, where there’s multiple versions of the truth. Organizations do see that today. And to me, that’s the human side of these hallucinations that we’re seeing with AI.

So for those organizations, in terms of leveling up, it is certainly approaching it from a people perspective first – ensuring leadership is in place, necessary roles around domain ownership, necessary standards and policies are in place. And really, what is the motivation for elevating data governance in the organization, ensuring that that messaging is clear from the executive level down.

Robert Dutt: So if human-in-the-loop is the solution to AI hallucinations, is AI-in-the-loop the solution to human hallucinations? Just kidding. Moving on to the regulatory environment – first thing that comes to mind, especially because SAS is so big in regulated industries, is finance and OSFI E-21 in particular. When you’re working with organizations that have to meet that bar, do you see it creating real urgency in the conversations you’re having? Or are clients still finding ways to buy time or building out how they respond to some of the regulations that we see?

Nat D’Ercole: Well, there’s nothing like having a catalyst in place to motivate – exactly. So yeah, I think that’s where regulation provides guidance, direction, standards. These are areas that organizations can look to in order to inform how they need to move forward as well. So that’s very much welcome, I would say, in terms of helping organizations steer their investments so that obviously they comply – and no one wants to be facing penalties.

Robert Dutt: Sticking with financial services – risk and fraud is highlighted as an area of strength for the Deloitte/SAS partnership. Where are you seeing the most active investment and I guess the most interesting use cases right now?

Nat D’Ercole: I would say in terms of risk and fraud, procurement integrity are areas that are horizontal across organizations. You can go from a fraud perspective – not just procurement, but other types of fraud within organizations. And then from a risk perspective, there are areas around financial risk where organizations need to ensure that they have proper scenario modeling in place to understand what stresses they need to address from an organization and modeling perspective. So I would say those are common use cases – asset liability management, treasury – just being more versatile, more accelerated in terms of running these scenarios. So solutions like SAS do provide capabilities to address that speed of process.

Robert Dutt: In general terms, as you’ve been here this week at the event – whether it’s a specific announcement, whether it’s an area of conversation, whether it’s what the leadership at SAS is thinking about – what’s caught your eye, caught your ear, and made you think, “Oh, I need to learn more about that”? What’s been your headline of the event?

Nat D’Ercole: The keynote – the interview that Jen Chase did with Mel Robins really hit home for me, and how she applied it to AI. And for me, ensuring that leaders are leaning in and providing the change that they want – or being the change that they want to see in the organization, living the change – and also helping organizations from a leadership perspective, executive perspective, to be comfortable. Many employees, I would say, across industries and organizations – some as Mel referred to – are afraid of what AI’s potential can do to their jobs. That’s a real human reaction. And so from a leadership perspective, creating the right environment for people to begin to lean in. I’ve said many times that, “Will your job be replaced?” – and oftentimes the answer to that is, “Yes, it’ll be replaced by those folks that are embracing AI.” So now is the time to lean in and begin to learn how to use it. So Mel’s comments definitely resonated. I looked around a large room – over probably 300 tables – and many people nodded with some of those remarks. So for me, that really resonated.

Robert Dutt: Pulling on that leadership thread a little bit – from where you’re sitting, what does good leadership look like in terms of guiding that AI discussion? Because that can be everything from really understanding it, making the case for it, making clear communications – not pushing, but being behind the organization’s efforts – to the kind of stereotypical thou-shalt-from-on-high, “The board tells me I have to do AI. Everyone’s talking about AI, make it happen.”

Nat D’Ercole: I think from an executive perspective, beginning to make investments in AI and ensuring that there’s a path forward for the organization – as individuals, departments, and then the enterprise. So that path forward, typically when we work with clients, we look to understand where the low-hanging fruit might be, both from an efficiency perspective and effectiveness. By effectiveness, being able to get insights faster, being able to run through processes faster, but at the same time ensuring – back to our previous comment – ensuring that the human is in the loop. Executives are also looking for ROI in use cases. And I would say that ROI should be looked at most definitely, but be somewhat lenient in terms of the payback timeframe. Some may be one year, some may be two years. The important thing is to start and begin to learn from the experiences, and have a set of – or journey roadmap of – use cases that will enable the organization to be more efficiently effective as a whole.

Robert Dutt: One of the bigger announcements here – and certainly the ones that got a lot of the attention and a lot of stage time – was SAS AI Navigator, built around governing AI use cases, models, and agents all at scale. Does a tool like that change what you guys deliver, or does it slot into something you’ve already been building? Does it kind of augment manual processes for you?

Nat D’Ercole: Yes, I would say it complements our trusted AI framework. I really like the visuals around the AI Navigator, and it really showed how AI could be green, could be yellow, and then could be red – and then ensuring that there’s a human loop addressing those red drift areas. So it certainly complements. And knowing how to bring the two together is, I would say, areas where clients will need help, and certainly what to prioritize first.

Robert Dutt: In talking to Ryan, the idea of clients increasingly looking at engagements that involve the scale of a GSI such as yourselves alongside niche industry-specific partners in the same engagement – and kind of creating that ecosystem approach. Curious if that’s something that you’re seeing and building for, or still more of an exception than rule in Canada.

Nat D’Ercole: I would say, going back to a previous question, we do lead from a business perspective and clients are coming to us to ensure that the technology investments that they are making make sense from an overall business perspective. And so how those investments are realized, we will often be an orchestrator of our alliances – both technology alliances and potentially industry-specific – where there’s expertise that we need to pull in as part of solutioning for our clients. So not abnormal, I would say. Where it’s justified, certainly our ecosystems and alliances are a key value driver for our success.

Robert Dutt: What’s the common genesis of that? I’m curious how often it’s you guys pulling in another party because they add something to the engagement, versus customers having an incumbent or someone they want to work with alongside you. How does that start, basically?

Nat D’Ercole: It really starts with having the conversation with the client – what are they thinking, and how can we help them best, bringing the best resources and capabilities to their problems. Clients may also have biases in terms of what they’re comfortable with. So it’s understanding that and advising them on whether that makes sense or doesn’t, and why.

Robert Dutt: Let’s get meta with AI a little bit here. There’s a lot of conversation in consulting about using AI to deliver AI projects faster. Is that something that you guys are doing in this practice? And what does it look like if it is?

Nat D’Ercole: Oh, absolutely, Rob. These are demands that our clients are requesting – that whenever there’s any engineering in place, whether it’s custom engineering or custom build solutions, custom build models, what have you, or migrations for that matter – migrating from legacy code, legacy reporting solutions, legacy SAS to SAS Viya, etc. – leveraging AI to accelerate time to value, lower the cost of delivering. And so to that end, we have developed accelerators. We do leverage AI and AI-assisted development engineering – AI-augmented engineering, if you will – to deliver overall lower total cost of implementation.

Robert Dutt: What does the team that you’re building to do this work in Canada look like? I’m curious especially what the skills you’re most looking for are, and what are the skills that are hardest to find or most need to be developed because they’re brand new.

Nat D’Ercole: Certainly data scientists, engineers, domain expertise in an industry that understands the business problems, understands the business language, change management – these are core consulting skills. I would say it just gets further augmented in the area of AI, and ensuring that resources have or are building experience or getting upskilled in the areas of AI to solution our clients’ problems. So I would say those are the key areas. And the last one is that trusted AI area as well – where our risk practice is focused on that. So from overall servicing a client, being able to pull from all facets of our multidisciplinary capabilities across the firm are key aspects in terms of why clients are coming to us to support them, because it’s not a technology problem.

Robert Dutt: Last one for me – what does success look like for a Canadian organization that’s, let’s say, 18 months into this kind of a transformation? And what’s the one thing that most often determines whether they get to success or not with an AI project?

Nat D’Ercole: I would say having clearly defined upfront business rationale – what does the future state look like from a business economics perspective? I’m not just talking about financial return. I’m talking about what does it mean for their people, and being able to sell that. Having that vision in place and actively working to chip away at building that out with the organization, within the organization – upskilling them so that they have the necessary skill sets to move forward, take on more themselves, et cetera.

So you definitely need to have the persistence, the top-level leadership to continue to drive, and I would say celebrate successes, advocate for better ways of working, and the benefits that it’s driving for the organization. So just continuing to sell the benefits, continuing to provide that vision for employees so that they understand what this means for them as they move forward. Those use cases where AI is replacing just the redundant tasks that employees are working on to get a report out – these are all areas where AI can improve the efficiencies, improve the quality, improve the trust, so that employees can focus on those higher-order, higher-value areas, strategic thinking – things that they’ve been hired to do.

I’ve been in this business for over 30 years and there’s always been that 80% of the time people are pushing data around, preparing data, and 20% is being spent on value-added activities. So AI really provides now the opportunity to flip that – finally. But obviously it does require safeguards, it does require executive support and leadership. So yeah, it’s a great time to be in, to be consulting, and to be working with clients to help them realize better ways of working.

Robert Dutt: All right. Well, good luck in making that flip. It is a long time coming, as you say. I hope Innovate finishes strong for you, and thanks again for taking the time.

Nat D’Ercole: Thank you, Robert.

Robert Dutt: There you have it – Nat D’Ercole from Deloitte Canada. I’d like to thank Nat for his time, and that wraps up our three-episode run from SAS Innovate 2026. Thanks for listening.

Few things I’m taking away from this one. First, the human hallucination concept. When organizations haven’t addressed data governance, you end up with multiple versions of the truth – different teams, different numbers, different answers to the same question. Nat’s point is that this is the human-side equivalent of what we’re trying to prevent with AI governance, and that the organizations that have already solved the data governance problem are the ones that are actually ready for AI. Not the ones with the best AI strategy – the ones with the cleanest data foundation.

Second, the 80/20 flip. Nat’s been in this business for over 30 years. For most of that time, people have spent 80% of their time pushing data around and 20% actually doing value-added work. AI has the potential to flip that. That’s not a new observation, but hearing it from someone who’s been watching it not happen for three decades really gives it some weight.

And third, Deloitte positioning as the orchestrator. They’re not just the big GSI anchor in these deals. They’re the ones pulling in niche specialists, aligning technology alliances, and making sure the business case holds together across all of it. That ecosystem John Carey described from the vendor side – this is what it looks like from the delivery side.

Hope you enjoyed this special coverage from SAS Innovate 2026. As fate would have it, we’ll have a new series starting later this week – more on that to come, but safe to say I’m currently on my way to Las Vegas.

If you found this one useful, follow or subscribe to the ChannelBuzz.ca podcast. We’re on Apple Podcasts, Spotify, YouTube, and most of the major directories. Ratings and reviews are greatly appreciated and really help others in the channel find the show.

Until next time, I’m Robert Dutt for ChannelBuzz.ca, and I’ll see you in the channel.