95% AI Adoption: How Zoetis Built a Model-Agnostic Platform That Stuck

0:00We're using it to automate study reports on the back end. We're cutting the time of that process in like half with AI in some cases. That saves months and months per submission. That's a big deal. It allows us to get innovative treatment to the market faster.

0:15Peter High: Welcome to Technovation. I'm Peter High, president of Metastrategy. My guest today is Keith Sarbaugh. Keith is the executive vice president and chief digital and technology officer of Zoetis, the world's leading animal health company with approximately 14 ,000 employees and$9.5 billion in 2025 revenue. The company develops medicines, vaccines, diagnostics, biopharmaceuticals, and digital solutions that help veterinarians, livestock producers, and pet owners. Keith joined Zoetis as Chief Information Officer in 2023, and his role has since expanded to encompass IT, digital, artificial intelligence, data, cybersecurity, and global customer experience.

0:51Peter High: He and his team enable business outcomes and improve the experiences of veterinary professionals, livestock producers, and animal owners. Prior to joining Zoetis, Keith was an executive at Biogen and spent nearly 20 years at E.U.I. Lilly. In this conversation, Keith and I will discuss how direct exposure to veterinary practices informs Zoetis' technology investments, how AI is changing drug discovery, and how the company achieved broad adoption of its model-agnostic generative AI platform. We'll also explore Zoetis' framework for measuring the value of technology investments, its approach to agentic AI, and the importance of change management as AI becomes part of everyday work.

1:27Peter High: Keith, welcome to Technovation. It's great to speak with you today. Peter, I'm happy to be here. I'm finally, I feel like I finally made the tier that makes it to the podcast. Let's quote it. Thank you for the invitation. Today's episode is brought to you by Tynes 3B. Organizations are pushing teams to move faster with AI. That's creating a wave of AI-generated code and unsanctioned automations, leading to rising costs, security risks, and huge IT burdens. Meet Tynes 3B, a single, secure environment that enables employees to build with AI while giving IT and security complete control to run and govern that work.

2:00Peter High: 3B enables effortless building. Describe what you need in plain English and 3B builds a fully functioning workflow. It uses IT-authorized connectors, meaning credentials, state-protected, and non-technical teams can innovate without taking on unseen risk. Also, with full visibility across every workflow, IT gets total governance, while features like Auto-Tune and Auto-Fix ensure workflows improve and self-heal continuously over time. Employees get the freedom to build, while IT gets the control to govern. No shadow AI, no data leaks. Tynes 3B moves IT and security from blockers to enablers, giving teams the speed to innovate without compromising control.

2:37Peter High: Visit the Tines website at tines.com. That's T-I-N-E-S dot com to learn more and try 3B for yourself. And now on to the interview. That's not the way I look at it. You certainly have been there for some time, but I'm glad we are finally getting on the record together, having known each other for quite some time as well, Keith. Well, one of the things I know about you is that you're a leader, an executive within Zoetis. And for some who are listening or watching, they may be less familiar with Zoetis. And I thought maybe you could take a moment here at the outset to describe the business that you help lead.

3:09Yeah, thanks. So Zoetis, we're the world's leading animal health company. We're a Fortune 500 company. We have about nine and a half billion in revenue, about 14 ,000 colleagues. And we operate in over 100 countries around the world. Our purpose is pretty simple. It's to nurture our world and humankind through the care of animals. In the businesses we're in, we predict, prevent, detect and treat illness across all sorts of animals. We sort of look at our business in two pieces. There's a piece around companion animals, so cats and dogs and the pets we all know and love, but also feed animals for, you know, growing protein demands around the world.

3:47So it's an important remit. We care deeply about the care of animals and have been in this business for some time. We have 12 potential blockbusters in our pipeline and blockbusters. We're looking at, you know, big deal health products, animal health products sitting in the next several years. And Zoetis has a long history of innovation. So it's not a lot of me too, but a lot of new establishing new therapeutic areas and sort of new treatments to really change the course of animal health.

4:18Peter High: Fascinating. And you joined the company in 2023 as the chief information officer. A couple of years ago, you became the chief digital and technology officer. Talk a bit about your purview, if you would. Yeah. Yeah. So exactly. I joined as a CIO and in that role for about a year, I had the remit to cover a lot of the business partnering that IT does and some of the core services that IT provides. About a year into it, I was promoted to chief digital and technology officer. So I report to our CEO, Kristen Peck, and I'm on the executive team. My responsibility really spans information technology, tech and digital, AI, data solution, cybersecurity, sort of anything we're doing in the tech and digital space.

4:59And then separate to that, I also have responsibility for our global customer experience program. So how are we connecting with customers via different channels around the world? I've been here now, I guess it's been over three years, and it's an amazing team. It's an amazing company, high value on innovation, high value on colleague and colleague experience, and a lot of people who care deeply about our mission and our purpose. And for me, that makes the job very, very exciting to work with professionals who care as much as my colleagues here at Zoletta's do.

5:30Peter High: That makes a lot of sense, Keith. And I wondered at a time where there's still debate about whether a broad purview like yours is the right one versus, for example, pulling out digital, pulling out AI, having different leaders who maybe even peers leading these different functions. I wonder, not that you've got a counterfactual to this, but I wonder what you see as the advantage of having all of those responsibilities under one executive. Yeah, it's interesting, Peter. It's something I think a lot about. I see a lot of commonality across those domains. And at Zoetis, I think where we're winning is where we can tie technology to business outcomes and the business outcomes drive a particular customer experience.

6:17And there's actually more of a thread there than you might see on the surface. I would say these days, my job is less about running the IT machine, and it's more about making sure we're making the right investments in tech and digital and data to enable business outcomes. And it's much more a hybrid of technology and business than maybe the world would have been five or ten years ago. And the customer experience thing, I've had a career in IT and commercial and sort of between IT and commercial. That appealed to me tremendously because it allows me to stay close to our customers and commercial, a piece of the business I care deeply about.

6:55But the role that technology plays in enabling and driving differentiated customer experiences is pretty profound and something, you know, maybe we'll get into later today. But there is definitely a tight connection. I feel it's a privilege to have the responsibility I do and wake up every day pretty excited to do it.

7:12Peter High: Can you find the customer? Actually, it's an interesting thing in animal health, a little bit different from, you know, your past at places like Eli Lilly and Biogen, where it would be somebody like me or you. The ultimate customer is our companions or as you defined it earlier before. How is customer defined within a setting like Zoys? Yeah, it's interesting. So we look at it a couple of different ways. There's business to business and who that customer is, and then there's business to customer, actually. So a lot of the times when we talk about customer, we're talking about veterinary professionals, both either in a clinic, you know, seeing our companion animals or working on a farm, whatever.

7:48But sometimes we're also talking about the end customer and the customer, you know, being either the pet owner or in some cases, I guess you could stretch it and say the animal themselves. But at Zoetis, we need to pay attention to all of that because certainly the veterinarian has a key role to play in the delivery of care to the animal and in the relationship with the pet owner or the animal owner. But also we have unique challenges in how we connect with the pet owners. And I think the expectations are rising there with the consumerization of everything, the prevalence of AI and everybody having access to great information.

8:23There's multiple channels we have to play in and ensure we're sort of looking at those customer touch points holistically.

8:29Peter High: Talk a bit about how you and your team get involved with some of those customer touch points. As I understand it, I think from our past conversations, Keith, you've actually spent time, for example, in veterinary practices to understand how those customers work and their needs and so on. I'd love to understand a bit more about those kinds of interactions and what you glean from them. Yeah, for me, Peter, that's probably one of the more exciting parts of my job, I think. I try to spend a lot of time in the field with customers, whether that's veterinarians, it's veterinary health teams who are using our diagnostics products.

9:04I was out last month with some cattle producers all around Kansas. The customers are very diverse. Their challenges are very diverse. And I think spending time in the field is just an awesome way to kind of see how our customer service teams engage with our customers, the challenges, the pain points. You know, it's easy to sit in a corporate office and come up with technology solutions that you think, you know, are going to change the world. But until you get out there and you see how it lands and how it's being used, man, I take away pages of notes every time I've done one of those visits. And one thing I'll say, you know, I grew up in health and life sciences.

9:43I spend a lot of time in pharmaceuticals. When I look at the difference between human pharma and animal health, the amount of time and attention and the relationships we have with our customers, it's just pretty awesome. You know, there's no shortage of time and their willingness to tell you about their challenges and look to us to help them solve challenges, whether it's running their practice or delivering care to animals. It's just really inspiring to me to see that close connection to the purpose of the company and how the things we're doing in tech and digital really truly impact customers that impact the care of animals.

10:18And that, for me, having that tight connection is just, just makes the job even more exciting.

10:25Peter High: Fascinating, fascinating insights all. I mean, since you mentioned your past at companies like Biogen and Eli Lilly, a human centered pharmaceuticals versus the purview that you've noted at the outset, and we continue to discuss here at Zoetis. How much overlap is there, would you say, between the roles that you've had in the past and the role you have now? It's such an interesting, on the one hand, you've really worked in one industry, one might add, but might decipher. But as you just pointed out, there's some really fascinating differences as well. Yeah. Peter, it's really awesome. When I was contemplating to move to Zoetis, I remember thinking about how much I value my purpose and what we were doing in human health and human pharma.

11:14And that was a big deal to me and something I had developed over 20 plus years in human pharma. And I thought, I wonder when I moved to animal health, how I'm going to feel about the purpose and the mission. And I think I might've shared this with you once before. I don't know what it says about me. I feel even more compelled in animal health than I do in human health. The connection people have with their pets, the role that animals play in food and protein production, you know, around the world. It's just an awesome, it's an awesome opportunity to have an impact on humanity. I will say the businesses are remarkably similar in some ways, but different than others.

11:49So kind of the value chain of the business, it's about innovation and research and development, and then scaling that into manufacturing and supply chain and commercialization. Like the value delivery process is very similar. The processes are similar. The technology, the opportunities, the vendors are very similar, but there are some differences. And I think a couple that pop out to me, you know, in human health, you're designing medicine for one species. In animal health, the diversity of species is like just incredible. And therefore the complexity of the business is sort of magnified. And I mentioned at the offset, like we're involved in the prevention, and the detection and the treatment, like that takes us into technology solutions that you don't often get into in human health.

12:34So we have a diagnostics business where we help detect disease. We have vaccines that are in prevention. We have treatments, you know, for diseases that are, that are, you know, diagnosed and treatable. We have devices. It's just a different kind of business. And the diversity of the business drives complexity and it drives different demands for tech and digital solutions than I've ever seen before. And it's just really pretty cool. So I liked it because I was able to leverage the skills and the experience I had built in human health for 20, I guess, 23 years when I reached here. But also it's new enough and it's different enough that it's a new business to learn while leveraging what I've already done in human health.

13:17So for me, it was a really cool transition and something I'm pretty excited to be a part of.

13:21Peter High: Yeah, that's fantastic. I want to talk a little bit with you about artificial intelligence, if I may, Keith. I know Zoetis is using AI in drug discovery and development, where the company's talked about making R &D faster, more precise, even less biased. I wonder how significantly could AI change the economics and speed of bringing new animal health products to market from your perspective? Yeah, it's a huge opportunity for us. I mean, AI and R &D is, of course, underpinned by fur review and quality control. But we've used it and we are using it really in target identification, designing new therapeutics computationally, and being able to figure out when something's going to fail or we're going to hit an issue, hitting that point quicker.

14:09You know, all in, we're using it to automate study reports on the back end. We're cutting the time of that process in like half with AI in some cases. That saves months and months per submission. That's a big deal. It allows us to get innovative treatment to the market faster. We've got a new clinical data management system. It's got an AI embedded. It improves data quality. It reduces manual validation. All of this with the goal of it's higher quality R &D and done in a shorter timeline. And both of those things translate to real value for the animals and the caretakers we serve. I mentioned diagnostics, AI.

14:49We have AI powering our diagnostics. So when you go take your animal to the veterinary practice and they read a sample from the animal, we have AI that helps improve the confidence and speed in which that sample is read. So it's really, yes, it's a part of R &D. It's a part of the way care is delivered. I'm really proud that we're using AI, not just in how we run the company and how do we make Keith and Peter more efficient and how do we do cool things inside the company, but how we show up in the marketplace and what we're able to do for the care of animals, I think is really pretty profound.

15:23Peter High: Fascinating, Keith. I know from our past conversations, you've emphasized the necessity to define value of an AI investment before building it. and then continuing to measure the value after implementation as well. I wonder, do you have a framework that you use? Are there metrics that matter most as you contemplate this? I think it's such an interesting idea, but still so new to so many. I'd love to hear what you thought. This is something we've worked quite a bit on. I know there's a ton of skepticism out there. You know, the MIT report, every week there's something different, questioning the value of AI.

15:55I would say we've been very intentional. We use something called the business value realization framework. So it's something we use not just in AI, but across our IT investments or anytime we're going to make a significant tech investment. And it's a pretty simple premise. It's like before we make the investment, what's the value we intend to receive? Is it incremental revenue? Is it cost savings? Is it productivity? What's the impact we predict on the P &L? And how are we going to measure it? So before we start, that's what we do. We have different checkpoints along the way of, you know, are we on track?

16:25Do we need to pivot? and then it sort of culminates in the system is up live and in production. How is the actual use aligned or not with what we anticipated? And a lot of times you find it's deviated some, and that often is a good cue of how we want to enhance it going forward. So we use it not just to measure the value, but sort of to use to kind of optimize the solution and continue to drive improvement, incremental value once we're live. And I would say we've had a pretty good track record. Not everything has hit. We've had a couple that we walked back in a significant way. But, you know, my philosophy is I would rather get it live, know it, have the opportunity to influence and improve it.

17:09And if not, you know, make a decision to retreat. And, you know, occasionally that's the answer. But I think on the whole, AI has been incredibly valuable for us. I think this value realization framework and having the discipline, this is not something we do in tech and digital by ourselves. It's with our business partners and with the support of our finance partners. Just having a regimented discipline process around making those investments and ensuring we're achieving the value, I think, is the key going forward. We can't do it on everything, but we certainly do it on any significant or material investments we make.

17:40Peter High: I also know you have a model agnostic generative AI platform that's now used by almost everyone. I think it's 95 % of Zoetis employees use it. I wonder how, how'd you achieve such a broad level of adoption? And what have you learned about moving AI from the, the sort of experiment used by enthusiasts to something that's used everywhere every day? Yeah. So we started like, I think many of your listeners will have with like an enterprise chat GPT and got early adopters. So a lot of enthusiasm around that. The problem when we looked at the kind of the model flexibility and really the cost and the efficiency of that, we couldn't see that on a large scale.

18:21And we had made a strategic choice. We wanted to make AI accessible to all colleagues at Zoetis. So as you mentioned, we've invested in something we call Zen AI, the degenerative AI Zoetis platform, where we expose, I think now we're up to 10 or 11 LLMs, you know, basically be a dropdown. and we've got 95 % of our colleagues using it on a regular basis. I think they've created something like 10 ,000 agents or something. It's like an obscene number of agents. So much so we now have somebody whose job is to curate the agents and make sure the ones that are most valuable are kind of getting the prominence and the attention they deserve.

19:01I think key to driving the adoption, we built an AI change champion network. We have about 350 folks around the world. And they're part of this community and their role is to help evangelize and help at a grassroots level, help their teammates adopt and leverage AI. And if I were to put my finger on sort of any one thing that's got us to the broad adoption, it was that getting those champions set globally, equipping them with the resources to be able to help evangelize AI. So they're sitting there next to colleagues who can turn their chair around and ask them a question. that's just a more effective way, I think, to drive utilization and interest in the platform.

19:44So I think that's been great. And then some of the use cases we're seeing and the agents we're seeing, they're things that we would have never dreamed to do from a top-down exercise, but like really, really creative solutions. One of the ones that I think is most interesting, somebody developed an agent to help build an individual development plan. So as Zoeta's colleague? What are the things I want to work on? An agent that sort of prompts relevant questions and helps build the development plan. Again, had we like gone the normal process, I don't think that's something that would, it's something that never surfaced in the processes we followed, but it's one of the most widely used agents on the platform.

20:23And I think one that generates a lot of excitement. So it just goes to show like the ingenuity, the creativity that exists in the organization. If we could put the platform in place, given the tools to do that in a safe and responsible way, like things beyond our expectations are happening. And it's really cool. We definitely have the challenge of, you know, I hear from a lot of my peers around token consumption. We started this year, I think, with a very ambitious, lofty goal on what we would want to see the organization consume in tokens. And as of June, we were about double that. So halfway through the year, we had doubled what I thought was a stretch to do for the year.

21:01And, you know, Frankly, I'm okay with that. It's a good indication of the platform's providing value. Of course, we're having to do things to make sure it's consumed in a responsible way, but I'm just thrilled with the uptake we've got and the energy I see. It's not often that you go through a day or half of a day of meetings and not hear about what people are doing in Zen AI on this platform. It's really pretty cool.

21:26Peter High: Yeah, super interesting. Out of curiosity, why has remaining model agnostic been important to you? Yeah, I mean, like, it's a little bit of an arms race, right? And like what the leading model is today and the most interesting model today is going to be different, not in six months, but it's going to be different in six weeks. And I think having the flexibility to select models that are more fit for purpose, depending on the type of work you're doing, do that in a safe and responsible way in a platform like we have with Zen AI. I think it's just it's critical that we not get too married to one thing, one, because they're going to leapfrog each other.

22:02And two, I think some models are just different or better at doing different things than others. And I think having the flexibility of choice and, you know, also price point has been key to us extracting the right kind of value from it.

22:16Peter High: Yeah. You mentioned earlier the use of agentic AI and the introduction of that to some extent. I'd love to hear more about that. But also, I wonder what needs to be true before you're comfortable allowing an agent to move from recommending an action to actually taking one. Yeah. Look, I think the biggest opportunities are the things that are transformative and use rule-based decision-making. So in our business, things like supply chain planning, triaging customer service issues, preparing documents for regulatory submission, optimizing manufacturing processes, the things where you can anchor on a clear set of rules and sort of codify that.

22:59but also are not mundane and routine. They're transformative things. I think that's where the sweet spot is for us on our agentic journey. So in supply chain, as an example, we're using agents almost like as a planning co-pilot to improve demand forecasting. So where will our products, where will the demand come from for our products in which part of the country, as an example, in the United States? and how do we make sure we're positioning inventory in the right parts of the country so we can meet demand on short notice and with minimal freight costs. It seems like basic and not sexy. That's actually a really difficult thing to do.

23:41We're using AI and we think we're going to save a significant number of millions of dollars just on freight charges, let alone the improvement in customer experience that that drives when we can deliver products that much quicker to our customers. We're doing the same things in software development that I'm sure a lot of your listeners are using. We're using AI to assist developers all steps along the way. We're seeing 30, 35-ish percent productivity improvement with our developers using AI as an assistant. I'm really curious to see kind of how that evolves as you look at AI not being an assistant to the developer, but the developer and the team itself.

24:19We're really curious about where that's going to go. So it's these things, you know, the agent opportunity for me is it's rules based, but it's still transformative. And I think those are the opportunities where we're seeing the most value.

24:31Peter High: Fascinating. Returning to customers for a moment more, you've used AI to expand customer feedback and an ability to draw an extraordinary amount of data from them and analyze that data for improvements or new ideas. Talk a bit about that process as well, if you would, Keith. Yeah. So I think like a lot of companies, we were heavily reliant on NPS net promoter score to sort of understand how we're doing with customer service. And we've got a system in now where we've integrated it with all of our customer touch points. So whether that's a sales rep in an office with a customer entering call notes in the CRM, a customer coming to one of our digital properties, a customer phoning into our customer service team.

25:13All these different touchpoints are integrated and aggregated in a system, and we're able to use AI to analyze and look for trends and patterns and anything concerning or opportunistic across those touchpoints in a real-time way. So not only is it more integrated and relevant and current, it's at a scale without AI, we couldn't do it. If we had 100 ,000 touch points before an NPS we were measuring, it's 10x that today in the platform we use and being able to use AI to sort of weave it together and understand where there's opportunity. we've uncovered some things that like frankly are a little bit embarrassing like when you start to see gosh this thing recurs over and over again but because it was happening you know in different channels we didn't have the the visibility to see it as a you know a collective opportunity our team our customer experience team has come up we've got a dashboard i think our first 10 or 12 sort of really obvious opportunities to improve customer service and we're just knocking those things out right and left.

26:16And it's cool. None of it's like particularly sexy, but it's all stuff that like, if we do one of those things, it's going to drive a dramatically different customer experience. If we can do 10 or 12 of those things and continue to do this at scale and continually improve. Again, it's, it's something we couldn't have contemplated. You can't do that with humans and it's, it's only enabled by technology. So that's obvious for obvious reasons. That's one I'm particularly excited about.

26:40Peter High: Yeah. In the past, you've also spoken to me about how change management is a catalyst for AI success. It's even a precursor for it in many cases. I wonder, can you talk a bit about the evolution of your own change management processes to ensure that you are in fact making the change necessary in order to more fully take advantage of artificial intelligence, even in some of the ways you've already described? Yeah. Gosh, this seems obvious and it's a little embarrassing to say it this way, but I completely undersold or underappreciated the value of change management and driving some of the adoption of these solutions.

27:15And, you know, you can look at it in a business case or in a project charter and you say, oh my gosh, the value of this is so obvious. As soon as we build it, everybody's going to be knocking the door down to come use it. Surprise, surprise, that's not the case in most situations. So we learned some painful lessons there early on. And I would say right now, my philosophy on this is involving the stakeholders and the users of the solution as early and as frequently as you possibly can. So they're part of the definition of what the problem is. They're part of envisioning what the solution could be.

27:52They're sort of involved along the way and get some skin in the game. And it just leads a different level of ownership. and with the different level of ownership comes a different type of adoption. And I would say we missed that in the first probably six or eight use cases we developed. That's an intentional check we make now and an intentional part of our process now. And I think the results on adoption, just from doing simple common sense things about involvement and communication and seeking broader input, just leads to a completely different level of adoption. So that's a must do going forward and something we won't shortchange again.

28:30Peter High: That makes a lot of sense. Important to have learned that lesson, certainly, especially before expanding it even more dramatically, as you mentioned. Probably shouldn't have had to learn it. I probably should have known that after 20 years in IT. But hey, you know what? Better late than never, I guess. You can't teach an old dog new tricks, apparently. Yeah, there you go. So before I let you go, I wanted to see if there's anything you've recently read or listened to that you'd recommend to others, especially at this time where it's necessary to, you know, read, listen to, think about new ideas in light of all the things we described?

29:05Peter High: What's caught your eye or caught your ear? Yeah. So I think one of the topics or subtopics I'm particularly interested in is sort of the effects on humans of AI. And I care a lot about how is AI going to impact early career professionals? I have three children, one of which entering the workforce next year. I think it's interesting to think about. I don't believe it's doom and gloom. So I read and I listen and I participate in forums that are all about how is AI going to impact early career? And then I think just the topic of AI on humanity is interesting. There's a researcher and an author at Duke, his name is Dan Ariely.

29:48He's a behavioral economist. And he put something out this week even in the premise of what he was arguing was about, like, if we continue to use AI, we opt to use AI in experiences where we would have otherwise maybe had a little bit of pain and suffering to learn. are we going to sort of not dumb ourselves down, but over time, do we start to lose our edge and lose the things that make humans great because we're so reliant on AI to deliver the things that otherwise would have built our skills? So I've been reading a lot sort of in that area, trying to formulate my thoughts and think about what does that mean for how we deploy AI at Zoetis?

30:29What does it mean for, you know, how I relate to my family and friends? And we think about things of careers 5, 10, 15 years from now. That topic in general is really interesting to me. And then a little less serious, one of the podcasts somebody in our Metis group got me on was the Acquired podcast. And I just listened to, they do a multi-port series on Rolex, which is right up my eye. I love watches. I love that podcast. And it was just an awesome, awesome story about the development of that brand and in that business model. So I don't know. Those are two things I'd said in the last few days that I've been paying attention to.

31:06Peter High: Great, great suggestions, all. And so interesting, also, this topic of the impact of AI on people who are earlier in their careers. As a parent as well, I've been thinking a lot about the same thing, certainly, Keith. Well, hey, Keith, thank you so much for taking time with me today, sharing a bit about your experience at Zoetis and your career, more generally speaking. Fascinating to hear about the work that you and your team are doing and a lot of the innovation that you're driving as well. I appreciate you taking time with me today. Yeah, thank you, Peter. Thanks for the opportunity. It was a lot of fun.

31:34Peter High: Thanks for listening to Technovation. If you enjoyed this conversation, subscribe and leave a rating that helps more leaders find our podcast. Thank you. And I look forward to seeing you next time.