Key takeaways:

• AI has moved from experimentation to a core driver of growth, with leaders now accountable for measurable value, not pilots.
• The biggest returns from AI come not from technology alone, but from changing processes, behaviors and ways of working.
• Agentic AI is reshaping how work gets done by enabling humans and AI agents to collaborate across the value chain.
• Shared context across patient journeys is essential to unlocking personalized, compliant engagement at scale.
• Organizations that focus on clear use cases, progress over perfection and activation over analytics will capture value faster.

For life sciences, AI is no longer about proofs of concept. It’s about value creation, speed to impact and changing how work gets done across R&D, commercial and patient services. Maximizing the value from AI initiatives means thinking beyond adding more technology. Companies need to change processes. They need to change ways of working.

Gains are likely to be uneven. Companies can extract value much faster in areas like commercial and supply chain than areas like research and development—which requires a whole set of process changes.

Agentic AI adds a new dimension to the puzzle. Humans and agents working together requires the right data and ecosystem. Agentic AI brings a new way of working, and the conversations leaders are having now are about how they can best enact change management across their workforce.

ZS Principals Srihari Sarangan and Mahmood Majeed recently had a conversation with Salesforce’s Kyrsten Musich (VP, Healthcare and Life Sciences Industries and GTM) in which they discussed how AI is moving from experimentation to a true driver of growth across life sciences.

Mahmood Majeed: Kyrsten, we’ll start with you. What are you most excited about with AI in life sciences?

Kyrsten Musich: What’s most exciting for me is seeing how clinical trials are now being designed and submitted with AI embedded into some facet of what those trials will look like. That’s very different from even two years ago. My husband is a medical oncologist. I’ve been watching his career and seeing the industry's evolution first-hand for the 13 years we’ve been together.

We’re also seeing new ecosystems coming to market. Whether it’s AI imaging coming into clinical trials around biomarkers or companies like GoodRx announcing a partnership with Surescripts and thinking about how to reinvent the e‑prescribing ecosystem with AI at the center, these changes aren’t just at the technology level. They’re about how patient ecosystems are going to work.

Srihari Sarangan: I’ll build on that. For years, the industry has talked about how operating models, organizational structures and behaviors limit transformation. What’s exciting about this era is that technology is acting as the driver of change.

In clinical trials, acceleration depends heavily on how technology enables the right behaviors and engagement models. Some of the agents and workflows being built are creating cycles of data and intelligence back into the ecosystem. As a technologist, this is validating. Technology can be a driver of change.

MM: Let’s talk about engagement. Personalized engagement has been a big focus for pharma for many years. But when it comes to different kinds of customers—patients, consumers, physicians and accounts—we’re looking at different ways of engagement. In an agentic AI world, what does truly differentiated engagement look like?

SS: Engagement across customer types is beginning to converge, even with compliance guardrails. There are three things we’re observing.

First, there has to be shared context. Most organizations are anchoring that context around patient journeys. That shared narrative dictates what different stakeholders need to hear.

Second, shared context does not mean shared moments. Every role has its own workflows and moments of truth. What matters is orchestrated moments, where the right roles come together when barriers arise.

Third is orchestration. We’ve talked about orchestration for many years, but those models are going to evolve significantly because they now need to account for many more variables across ecosystems.

MM: That looks very different across therapies—GLP‑1s versus gene therapy versus oncology and immunology. If you’re in these very specialized ultra-therapeutics, what does that modality look like?

SS: The growth dynamics are very different. In rare disease and cell and gene therapy, pharma is involved from diagnosis to fulfillment. In GLP‑1s, pharma acts more like a high‑volume consumer business.

Historically, organizations used different platforms for marketing, media, field engagement and services. The future is a common data foundation that brings these federated ecosystems together through a shared translation layer.

KM: Shared context is critical. For a long time, organizations assumed compliance prevented shared context. But when you think about the shared context around therapeutic journeys, that conversation changes.

We’ve seen how lack of shared context destroys value. Account managers often had to manually reconstruct what had happened across teams. That data exists, but it’s fragmented and unusable. That’s lost value.

MM: You’ve seen the healthcare industry broadly. Are there examples of organizations doing this well?

KM: Some provider organizations are doing this well, particularly those that embraced progress over perfection early. By managing expectations and staying focused on progress, they’ve been able to transform more quickly.

The other key factor is being very clear on use cases and value creation and managing expectations around that.

SS: I’d add that emerging pharma organizations often do better than large incumbents. They’re forced to get this right early, particularly around launches, and that shows up in results.

MM: As we think about growth, Kyrsten, what do you see as the most important frontiers for pharma?

KM: In the near term, it’s about outcomes. Commercial models haven’t changed much in a long time, and AI creates an opportunity to rethink flexibility, productivity and how time is used.

Longer term, I’m focused on ecosystem redesign, especially direct‑to‑patient models. What we’re seeing in GLP‑1s will extend into other therapeutic areas. I also believe payer and PBM models in the U.S. will change significantly as AI increases transparency.

MM: As a final question, what is one thing organizations should stop doing, and one thing they should start doing, to drive growth with AI?

KM: Organizations should stop searching for perfection. AI is not a traditional application. It’s more like onboarding a new employee. It requires ongoing training and management.

SS: Organizations should stop doing analytics for analytics’ sake. Analytics without activation don’t create value. Insights need to be tied to action.

I think the start part’s a little tricky. I think you’re going to have to diversify the types of things you do. You might want to do something in Malaysia, something in Spain, something in U.S.—and do something in individual therapies across these different countries to understand truly what’s effective and what’s not.

KM: One that I would start as you're doing these projects is having an expectation baseline of your end users. I have watched so many people walk in with assumptions like, “Our reps don't know AI” and “Our reps just want the answer.” And we have gotten a lot of those expectations wrong. There are a lot of reps who pride themselves on being analytical and knowing how to run a deep analysis on their business.

I think really having a clear understanding and baseline of whatever persona it is that we’re trying to solve these challenges for, having a really clear understanding of their expectation baseline around these technologies before we make assumptions on what the experience should be is really important for the success of the project long term.

MM: Thank you, Kyrsten. Thank you, Sri. This was a fascinating discussion.