PwC's Chief AI Officer on the SaaS-pocalypse, Agent Governance, and What's Real

0:00Now we're seeing business teams show up and say, hey, I vibe coded this new application. I'd like to get it into production. And the tech team, they're going nuts because they're like, wait a second. This got to integrate with our enterprise architecture, our data layer. It's got to be secure. There's a whole lot we got to do to get industrialized and into production.

0:19Dan Priest:Welcome to the Talking AI podcast, where we talk AI with both experts in the field and early adopters. I'm your host, Matt Page, and we're here to demystify AI for you so you can get some value from it. Let's talk some AI. In the past few months, we've watched Anthropic almost single-handedly cause many market panics across multiple sectors. They've put out security scanning and code review capabilities and cybersecurity stocks tanked. They showed how cloud code can do COBOL modernization work and IBM dropped 13%. And with each new feature release, They're driving some to call it a SaaS-pocalypse.

0:55Dan Priest:And the market's trying to figure out what AI is about to replace next. But what actually is happening inside the company is doing the work. And that's what this episode's about. I'm sitting down with Dan Priest, the chief AI officer at PwC. He oversees their massive AI investment, works with the vast majority of Fortune 500, and has a front row seat to what's actually happening behind that lines. And the reality sometimes looks very different from the narrative. And we're getting into what companies are actually doing to get ROI, where they're wasting money, why most organizations are measuring the wrong things, and what leaders are pulling ahead and what are they doing.

1:31Dan Priest:This is your AI reality check. Let's get into it. Welcome to the show, Dan. Thanks, Zach. Glad to be here. Yeah, this is going to be a fun conversation because we're getting beneath all the crazy headlines that are going on. And like I said, in the intro, in the last few months, we've watched this anthropic almost single handedly caused these mini market panics across all kinds of sectors. Uh, and it seems like each new feature they put out, it causes shockwaves through the markets, but what's actually going on here is the market correctly pricing in this disruption, or are we watching an overreaction play out?

2:06Dan Priest:Or is it something in the middle? What's, what's your take on what we're witnessing? Yes. Yes. And probably yes.

2:16It's, you know, the disruption we're seeing in the SaaS market is in some ways rational, right? It's that the investors are looking out and say, hey, we've got to price future earnings. And the clarity on those future earnings around SaaS companies has gotten a little less clear because of these announcements you just referenced. And so classic investor logic, what's the discount rate? What's the appropriate discount rate to reflect certainty or uncertainty, the level of risk they're signing up for around those earnings? And the discount rates have gone up. So the valuations have come down. And I would just sort of summarize by saying there is still hype out there that factors in.

3:08There's a lot of reality, though, a lot of reality to build with and just a ton of confusion. And so investors are trying to make sense of this complicated world. Management teams are too. And we're starting to see patterns emerge where leaders are doing certain things around business a strategy that factor in AI. And they're starting to develop approaches to rebuilding critical capabilities to compete and win with AI. And they're getting ahead. And then others are trying to make sense of all those market gyrations that we're seeing. And there's a little paralysis around how do I get started? Where do I get started?

3:54What should the returns be? How much should I invest? And they're just trying to make sense of that. And some are doing a better job than others.

4:01Dan Priest:Yeah. And this is not a finance podcast by any means. So you're still in the right place if you're here for the AI, but you hit such a core point because when people are pricing equities and how the market's pricing them, it's based on expectation of future cash flows. And that uncertainty is driving a lot of this change because it used to be very predictable And now it's just not as much. And the interesting thing is too, executive teams are seeing all of this. So they're getting pressured by the board by what are we doing in AI? Because it's causing all of this panic and hype and demand and all of that.

4:36Dan Priest:Quick break in the pod. Our State of AI 2026 report just dropped and it breaks down what actually is changing in AI, what's hype and what leaders need to be paying attention to this year. You can grab it right now on our show notes or at hatchworks.com. I'm curious from your perspective, like what are you seeing in the organizations are you seeing you know CISOs at fortune 500 companies actually ripping out their security tools and like hey we got we got Claude uh here now to do all of our security or is it a bit more nuanced than that like it's the actual real world nuanced yeah yeah you know on the the security front um they've actually done a pretty good job anybody who's building with these capabilities, one of the first hurdles they had to cross was what is the appropriate security architecture for AI?

5:26And you don't want your data assets leaking into these large language models. You don't want inappropriate use of customer data that compromises the integrity of those data. And so to be able to build industrial strength solutions, we had to solve those security problems very early on. And I think we're in pretty good shape, not universally. You can see some missteps in the market happen. But for the most part, I would say the CISOs have figured this thing out. And then to your point, yeah, I've spent a lot of time with board members. I've spent a lot of time with C-suites. Two years ago, the most popular client conversation I was having was with chief technology officers, chief information officers, chief data officers.

6:21That has fundamentally changed. And now these days, I spend the majority of my time with CEOs and their leadership teams. And it sort of signals the shift that's happened, that the technology is good. And we can talk more about that because it's really good. It's not good at everything. And the leaders are figuring out where is it industrial strength and where is it still maturing. And there's some flat spots still. but CEOs are also to the point about the pressure they're getting from boards and the investor community are saying, figure it out. We should see a demonstrable investment being made in AI.

7:08We want to move beyond experimentation. It's not that the age of experimentation is done. Like we live in a digital world that has to digitize a whole lot more. And these code generators are the best path towards doing that. So it's certainly about the tech, but it's much more about the business discussions on what's going on in my industry. How much are my peers investing? Where are they getting a return on that investment? And then they start to look inward. They say, okay, irrespective of what's happening in the industry, in this age of AI, the number one success factor is human talent. Like, where do I have great engineering talent that I can combine with deep domain expertise, that I could combine with really good change leaders that I'm going to oversee as a leadership team?

8:04Not in a delegated model where I say, hey, tech team, go figure it out. But instead, where can I lean in and create certainty around the results I'm investing in? That's what's happening right now. And the market gyrations are forcing a more serious conversation at the management level about what do I do with AI and how do I secure my future? And it's fascinating to see because the advice I give to leadership teams is if you leave that space out addressed, meaning if you don't have an affirmative strategy that contemplates AI, your board and investors will start to worry. And so the leaders, they're going through a process of saying, okay, where do I need to be a leader?

8:56Where can I tolerate a lag? And where might I exit certain products and services or certain parts of my business? Because the gap is growing so quickly. It's a minor part of my business. And I'm not willing to invest to do what it takes to become a leader in that area. So this lead lag exit analysis creates the byproduct is a strategy that they can go and talk to investors about. They can assure board members that they've got a plan.

9:25Dan Priest:Yeah. You just gave me so many rabbit holes. I want to go down there, but it's such an important point up front because I do agree early on, this was like, oh, this is technology, IT, CTO, engineering. You got this. But we've seen that shift too. It very much is transitioning into the business function. I think in large part, a lot of leaders are realizing the potential because they're starting to use these tools, but it's nuanced in terms of how you integrate it into the business. And there's a great AI leader, his name's escaping me right now, but he talks about This is more change management than it is technology.

10:00And he gives the analogy of, you know, I'm not going to cure heart disease by putting

10:04Dan Priest:a treadmill in every home in America, right? It's about changing the habit. And that's the hard part there. And so let me ask you this. I think one interesting thing I've noticed is a lot of this gets down to the tasks and the processes within an organization, right? A group of tasks effectively make up a job function. And a lot of people I've noticed immediately go to, okay, let's just, let's automate these tasks. But a lot of times I feel like you got to step back and say, should, should this process even still work this way or exist now that we have this new tool in our tool belt, that is this probabilistic system that can do all of these great things.

10:48Dan Priest:So it's almost a bit of process reinvention is the first step, which in many ways is just kind of mind-blowing, I think, for a lot of leaders, because that's completely reorienting how you work. yeah so to your first point uh i i have observed that there's an 80 20 rule in this uh this world of ai and the 80 20 rule is this 80 about of this is about uh business transformation 20 is about the tech yeah so business transformation is about how where can i incrementally improve a process to get benefit? Where do I fully reimagine a function or an impact zone that I care about? How do I get people to adopt it?

11:43So innovation is far outstripping adoption right now. And so like figuring out your people strategy to drive adoption, figuring out the new partnerships that are required to create value for the enterprise, figuring out how the operating model and maybe even the business model, like how do you compete and win and make money in the market? All of those things are up for discovery. That's all business. I haven't talked about tech yet, right? Tech is the 20%, which is, okay, I've got a dirty data state. It's fractured. It's hard to deal with. It's hard to get value from. I've got a complicated application architecture.

12:26Tech is really important. and I've got this new layer, this new agentic layer. What do I do with that? Tech is really important, but it is 20%, yeah, roughly, of the equation. So much more about the business transformation. That's why ROI has been a bit elusive. It's like the most common mistake I see is companies flip those. They make the conversation 80 % about the tech and 20 % about the business transformation. And to your question about reimagination, listen, if I do that lead lag exit analysis and I say I want to be a leader in a certain area, customer care, sales effectiveness, growth oriented initiatives, that's a reimagination.

13:16Like you should not be incremental there. And that is really hard for companies to do right now because for the last couple of decades, maybe longer, we've been not investing in innovation cultures that lead towards that re-imagination. We've been investing in standards cultures, meaning I'm going to standardize this process, everybody adopt it, get more efficient, drop those dollars to the bottom line and do it again and again and again. Now it's reimagined it all. And they're like, okay, that sounds great. I just don't have the skills to do. So it is an enormous leadership challenge facing most companies right now.

13:57Dan Priest:Yeah. It's so funny how it just gets back to this, the innovators dilemma in a sense. And you have these incumbent organizations and a lot of them do struggle with it, but it's not like some new thing. uh, you know, AI, generative AI is new, but the, the concept of the innovators dilemma is still very, you know, reoccurring in a sense. And, you know, you have these startups and newer organizations and they, they don't have all the baggage in the history. And they also have natural constraints where they kind of have to work in this new way. So they're leading with that. And you mentioned, I love the analogy you just gave in terms of a lot of organizations were oriented around optimization and standards and all of those things, which is what they should have been doing.

14:45Dan Priest:But now you've hit this major inflection point, this transformation. The people side of it that you touched on, I think it's just so critical. And it's extremely elusive in terms of, okay, well, how do you actually drive adoption in the organization with all of your people? I've seen a lot of different approaches to it. Curiously, Like what have you seen work or be successful? And I'm sure there's multiple approaches and angles you can do to start to enable your team while also giving them direction. There's almost the top down and the bottoms up. Like what's your thoughts on tackling that piece of it?

15:26It's exactly what you just said. It's top down and bottom up. So we're saying the leaders do is two track transformations. and track one is that re-imagined track that you referenced earlier. Track one is I've picked my spots. I know where I want to lead. I know what the shifting table stakes look like. I have to be successful in these areas and track one gets a lot more management attention, a lot more investment, your best engineering talent, your deepest domain experts, your best change leaders. A track one is I'm going to make this thing successful because the future of my business depends on it.

16:10Track two is where everybody started, frankly, which is I don't know exactly what to do with these tools. So I'm going to give it to the tech team and I'm going to give it to the citizenry, the corporate citizenry and ask them, go figure this out. Right. Tell me where there's some good opportunities to apply AI. and that track is important still, right? Because you want people to get comfortable with the tools. You want people to start incorporating them into their ways of working. It just doesn't produce the big transformational outcomes that management teams care most about. And you might be pleasantly surprised with some bright spots that emerge in that track too, like some pockets of innovation that flow.

16:58but that's not where we're seeing the outsized impact. We're seeing it top down, much more strategy driven, much more leadership led, much more intentional. And that pivot has made a big difference. And just being able to organize your thinking around, I've got both tracks activated. I'm managing them in very different ways to get the right results.

17:27Dan Priest:Yeah, it's almost like this, you know, you almost have to give permission plus provide the direction I've noticed because a lot of folks, especially early on, they were very restrictive in terms of the use of AI, which logically that makes sense. But what I see on the ground is a lot of people are just scared to use the tools in a sense. So they're not adopting them. I think that's one interesting piece about it. But how do you think about that? Because literally I can go a week and there's five new amazing tools that have come out. You know, OpenAI has done this, Anthropics done that. How do you think about more specifically what tools you're leveraging while also giving the autonomy to use the best tool for the specific job function or role?

18:17Dan Priest:Any nuance or interesting approaches you've seen within larger organizations that may have more bureaucracy on that side? Yeah, a couple of things. One is in the beginning, there were hard choices made about which tool you're going to use. Now we're seeing companies choose not to choose too soon. So you see things like model gardens emerging where, you know, depending on the architecture, you're getting the best of all of the tools or all of the models. And they're good at different things, right? And you want to have access to the best. And to your point, it's been really fascinating to watch the arms race around the tooling, the level of investment these model providers are making, the blistering piece at which they're improving and the rankings just keep bouncing around, right?

19:28Like the leader today isn't for sure going to be the leader tomorrow. In fact, if we look at the last couple of years, you can bet that there'll probably be a new tool, a new release that's whiz-bang and you're going to want to take advantage of. so you know making choices that don't lock you in but give you the best and then lining up those solution sets so you like the stable of solutions is growing and then lining up those solution sets against business problems you're trying to solve some some solutions are super expensive and you might be willing to make that investment because they're solving a 50 million dollar problem I'm going to spend$25 million because I'm solving a$50 million problem, right?

20:16That the benefits of that problem are significant or solving it. And then there's this value curve where you start moving across the enterprise and like, well, not all problems produce$50 million in benefits. And so I need a solution that's fit for purpose, capability-wise, and the economics have to make sense too. So I fill out that solution set. So I start mapping it to different business problems going after. That's one point on the tooling. The other is it just ain't about the tooling, right? If you believe what I said before, you're AI pundit. It's mostly about the non-technical skills that have to become AI powered.

21:06And that is much more about how should this process run? with AI? That's not a tooling question. That's a business question. Or, you know, what should the configuration of my partnerships look like? Or how do I get people motivated to change and adopt? These are all business skills that are now AI infused. And it's interesting to see companies, they continue to invest in the tooling layer, but give short shrift to all of the other non-technical AI skills that have to get developed to be successful and get an outcome. And so it's sort of get the right tooling in place. Be careful about lock-in or creating solutions that limit your optionality.

21:57That's important. But then also recognize that like every other big change in business history, there's so much more than just the tech. We keep learning that lesson over and over again. And you've got to invest in a modern apprenticeship model where people who use the tool rethought, they reimagined the process, reimagined how work will get done, reimagined the people models, all those non-technical skills, pair them up with the rest of the teams that are taking on similar problems and help them be successful and create these feedback loops so that you can get the outcomes you want. Yeah.

22:37Dan Priest:I think to a certain extent, you almost just have to keep your blinders on because once something great comes out with one tool, give it a week. The next one will probably have the same thing. But there's this interesting, I want to get your take on this. So I've noticed this in my own team. I covered strategy. We also have marketing as a function in my team. but I've noticed this like bleeding over and blending of roles in a sense to where my analyst and strategist and writer and designer they all can do more and they're also bleeding into the functions of their counterparts but it's almost been a good thing to where they can take a concept or whatever thing they're working on further than they used to be able to in the past and there's a sense of, you know, you can now do a whole lot more.

23:31Dan Priest:And there's this narrative of, you know, AI is going to take all of our jobs. And I think there, we were seeing some of that play out. You saw block do layoffs. You're seeing all of this happen, but in a sense though, and just curious your take on this. I feel like the bottleneck within a system is always going to shift to wherever the bottleneck is. There's always going to be a bottleneck that exists. Right. And I'm not just doing the same amount of things. I'm doing way more. And my team is way more capable. I think the key thing is them being enabled and leveraging AI. But I don't sit around twiddling my thumbs going, man, we don't got anything to do.

24:13I feel so much busier now than I did a year ago.

24:18Dan Priest:Uh, so I just feel like just humans in general are just naturally always going to focus on progression and abundance and all of these things. Um, but this is a very meandering topic, not quite of a question, but I'm just curious your thoughts on that. And it gets, I think, to the nuance of the debate of does AI take all our jobs or does it create this new emergence of, you know, human capabilities with AI and all of that? I think AI promises to be disruptive. Jobs will change. That blurring that you're talking about is a very real thing. We see that a lot in the relationship between traditional IT teams and business stakeholders who are getting a lot more tech savvy.

25:09and yet that relationship and that interaction model is shifting quite a bit, right? In the past, the business would show up at the front doorstep of IT and say, hey, I got a requirement, can you help me? And the IT team would be forced to say, oh, get in line, right? Let's like, we'll work you through the process. Create your ticket. Now businesses, right, here's your ticket. Now we're seeing business teams show up and say, hey, I vibe coded this new application. I'd like to get it into production. And the tech team, they're going nuts because they're like, wait a second. This got to integrate with our enterprise architecture, our data layer.

25:50It's got to be secure. There's a whole lot we got to do to get industrialized and into production.

25:59Dan Priest:But you know, secretly they're thinking, man, this thing's pretty good at the same time too, but there's all of those. repercussions you also mentioned shadow it is just a whole new uh whole new realm today but it's it's to your underlying question is that it will be disruptive jobs and roles will change but we live in a digital world that is not terribly digitized right now and you think about how much manual work we're still doing. I would love for my smartphone to be smarter. I'd love for my enterprise systems to be more intelligent and more capable and more touchless. I would love to have more of a boost working across that system's landscape.