How Zoetis Turns Broad AI Adoption Into Business Value
Zoetis is treating AI less as a lab experiment and more as a portfolio discipline: every meaningful use case needs an expected P&L effect, a measurement plan and a post-launch review. That is a useful operating model because broad access creates enthusiasm, but value only materialises when leaders can separate adoption from impact. The management implication is clear: AI governance cannot sit only with technology teams; finance, business owners and operational leaders must jointly own benefit definition, tracking and course correction.
The companyโs willingness to scale back initiatives is especially important. In many organisations, sunk-cost pressure keeps weak pilots alive, while successful tools are not curated or reused. Zoetisโ agent curation role and change champion network suggest a practical answer: manage AI as a managed product ecosystem, not a collection of one-off experiments. That raises a portfolio question for leaders elsewhere: which capabilities merit enterprise support, which should remain local, and who decides when an agent is retired, standardised or expanded?
Its experience also shows that adoption is not a communication problem alone; it is a design-and-ownership problem. Involving users early increases the likelihood that workflows change, rather than just adding another tool on top of old habits. The trade-off is slower start-up and more stakeholder effort, but the payoff is better usage and fewer abandoned investments.
The next questions for executives are practical: what metrics prove AI is improving cycle time, cost, quality or customer outcomes; what controls prevent uncontrolled sprawl; and how will the organisation balance model choice, security and cost without fragmenting accountability?
At Zoetis, broad access to artificial intelligence has been accompanied by a requirement to demonstrate what significant investments deliver. The animal health company, which generated approximately $9.5 billion in revenue in 2025, is applying AI to research, customer service and everyday employee work. Keith Sarbaugh, its executive vice president and chief digital and technology officer, sees adoption and financial accountability as complementary responsibilities.
Keith joined as CIO in 2023 and subsequently assumed a broader mandate spanning technology, digital, AI, data and cybersecurity, alongside global customer experience. He connects investment decisions with the needs of veterinary professionals, animal producers and pet owners. โMy job is less about running the IT machine,โ he explained, โand itโs more about making sure weโre making the right investments in tech, digital, and data to enable business outcomes.โ
Defining Value Before Investing
Zoetis uses a business value realization framework for material technology investments. Before work begins, technology leaders collaborate with business and finance partners to define the expected benefit, its potential impact on the profit and loss statement and how it will be measured. Checkpoints during implementation allow teams to adjust direction, followed by an examination of actual usage after deployment.
โWe use it not just to measure the value, but to optimize the solution and continue to drive incremental value once weโre live,โ Keith said. The approach also creates room to acknowledge failure. Some initiatives have been substantially scaled back, and he favors getting solutions into use so teams can evaluate them, improve them or retreat when necessary.
Making AI Accessible Across the Company
Zoetis initially introduced enterprise ChatGPT, but questions about model flexibility and the economics of broad deployment prompted a different approach. Its Zen AI platform gives employees access to roughly 10 or 11 large language models through a common interface. Keith reported that 95% of colleagues use the platform regularly and estimated that employees have created approximately 10,000 agents.
Maintaining choice matters because models improve at different speeds and have different strengths. โI think having the flexibility of choice and also price point has been key to us extracting the right kind of value from it,โ he noted. The volume of employee-created agents has also introduced a practical responsibility: Zoetis now has someone curating them so useful capabilities receive attention.
One popular agent helps employees prepare individual development plans by asking relevant questions. Keith highlighted it as an example of an idea that had not emerged through conventional planning. Adoption has been supported by an AI Change Champion Network of about 350 colleagues worldwide, giving employees nearby peers who can answer questions and help them apply the tools.
Connecting AI to Animal Care and Customer Experience
AI is also changing how Zoetis develops products and supports care. Keith described applications in target identification, computational therapeutic design and identifying potential failures earlier. In some cases, AI has halved the time required to prepare study reports, saving months per submission. A clinical data management system with embedded AI is intended to improve data quality and reduce manual validation.
โBoth of those things translate to real value for the animals and the caretakers we serve,โ Keith emphasized, referring to higher-quality research and shorter timelines. AI also supports diagnostic sample analysis in veterinary practices, helping improve speed and confidence. His visits to clinics and cattle producers help him understand how technology performs in the settings where customers actually work.
That customer focus extends to a system combining information from sales interactions, digital channels and service calls. Keith described roughly ten times the interaction coverage of the companyโs earlier Net Promoter Score approach. AI analysis exposed recurring problems that had been obscured across separate channels, giving the customer experience team an initial set of 10 to 12 improvement opportunities.
Building Adoption Through Early Involvement
Keith acknowledged that Zoetis initially underestimated change management. The apparent strength of a business case did not ensure people would use the resulting solution. โI completely undersold or underappreciated the value of change management in driving some of the adoption of these solutions,โ he reflected. The lesson emerged from early use cases where user involvement had received insufficient attention.
The company now involves stakeholders in defining problems, envisioning solutions and shaping development. โIt just leads to a different level of ownership, and with the different level of ownership comes a different type of adoption,โ Keith underscored. For Zoetis, the discipline of measuring value depends on that participation: employees must help shape the work before technology can meaningfully improve it.
Enjoyed this article? Sign up for our newsletter to receive regular insights and stay connected.