SPEAKERS:
Ben Finlay, Partner, Global Healthcare Advisory, Stratt
Angela Genco, Transformational Marketing Head, Bayer Italy
Cyril Mandry, International Head of Omnichannel Customer Experience Oncology, MSD
Itziar Canamasas, Global Head of Oncology, Boehringer Ingelheim
Phil Krzyzek, Head of Regional Business Excellence Leads, Merck- KgA
KEY TAKEAWAYS:
• LLMs are already replacing pharma's core informational value to HCPs
• Cross-industry data shows limited correlation between click-through rates and sales impact
• Competitive differentiation will hinge on personalized, and first and foremost personal customer engagement
• Organizational architecture may outlast technology as pharma's most durable competitive moat
The Blind Spot No One Could Explain Away
The most arresting moment at Pharma 2026 wasn't a keynote claim or a polished case study. It was an admission. Phil Krzyzek, Head of Regional Business Excellence Leads at Merck, described asking his own organization a single diagnostic question: "Do you know what LLMs are actually saying about our brands at the moment?" The answer from his colleagues revealed that this is somewhat of a blind spot and an emerging topic that requires urgent attention.
Ben Finlay, Partner at Stratt and moderator of the final Day 2 panel, articulated the key structural condition that makes that silence so costly: "The front door of healthcare has fundamentally changed with the rise of, pick any language model that you wish to choose." With 72% of US physicians now reporting AI use in clinical practice, the shift isn't directional. It's operational. HCPs are accessing clinical information through LLMs today, and pharma's engagement infrastructure was not built to see that happening.
Against that backdrop, Angela Genco offered the session's quieter counterweight. "I'm sitting in a role that two years ago didn't exist in Bayer," she noted. Bayer Italy's experience rests on a deliberate choice: to change its organizational model so it could move closer to customers' real needs. The disruption demanded a structural response, not a strategic memo—and the distance between organizations that merely observe the shift and those that reorganize around the customer defined everything that followed.
When Optimization Becomes the Wrong Problem
The existential question Krzyzekposed had no comfortable answer: "In a world where doctors and patients can access quality information that precisely answers their question, why do they want to see a sales rep from industry? Like, what actually is the value?" He wasn't performing provocation. He was naming a gap that no panelist could close on stage. He pushed further, asking whether the industry's current organizing concept might simply be obsolete: "Is omnichannel Blockbuster Video? Is it just going to be completely superseded? I don't know. But these are important questions."
The danger Finlay identified is precisely that these questions get deferred: "Are our language models just another channel to optimize for?" That reframing with treating LLMs as a placement surface rather than a structural challenge, is how an industry can invest heavily in AI while remaining strategically exposed. If pharma's historical value proposition to HCPs was clinical information access, and LLMs now deliver that more precisely and on demand, then optimizing the delivery of that same value through any channel is an exercise in diminishing returns.
Cyril Mandry, offered the most durable reframe. "Clicks don't cure, humans do. You really have to focus on the interactions, on the relationship you build with the customers." His argument was structural: the value that survives AI commoditization is relational rather than informational. The problem is that pharma's measurement infrastructure was built for the model that's being displaced. Reach, frequency, and click-through metrics cannot capture relational value. That means the measurement system itself continues to allocate investment toward the traditional capability, not the one worth building.
Mandry extended this with data that should be uncomfortable in any commercial planning conversation: "If you look at data across industries, there's no correlation whatsoever between click-through rates and actual sales impact. There's none." Companies can achieve excellent funnel metrics while producing no commercial effect. The metrics confirm activity; they don't confirm value.
Discover more on this topic at Pharma Commercial Data & Tech Europe 2026 (4-5 November, London) Europe's collaborative home for data and tech pioneers. Visit the website here.
The Architecture Answer
If the engagement model requires reinvention, the enabling condition is organizational structure. Bayer Italy's Dynamic Shared Ownership implementation provides the session's most detailed evidence that structural redesign, ahead of and independent of specific technology deployment, can produce measurable commercial acceleration.
Genco described the foundational act plainly: "The first decision was to destroy the business units. There are no more business units in Bayer Italy. And we rebuilt all the organization around two different teams: the customer teams and the capability teams." Customer teams are non-hierarchical, cross-functional, and segment-focused, with a virtual seat reserved for the customer in strategic decisions. Capability teams, marketing, medical, commercial excellence, market access, sales, maintain functional expertise and serve customer teams through what Bayer calls a "span of coaching" rather than a span of control. The hierarchy didn't disappear; it was redirected.
The commercial result was specific. "One of the achievements we had thanks to this cross-functionality was to achieve the reimbursement of a new indication within 6 months when the benchmark, the internal benchmark, was 12 months in the past." A six-month reimbursement timeline in Italy, half the prior benchmark, came not from a new AI tool but from removing the organizational friction that had been slowing decisions. Bayer also mandated that all staff, including data scientists and analysts, spend 30% of their time with external customers. The internal/external customer distinction was formally abolished.
Itziar Canamasas, Global Head of Oncology at Boehringer Ingelheim, described a directionally similar model at a higher level of abstraction: "What we want is to create the brain of an organization that will be composed by human-only work, or by human-augmented work by agents, and/or by agent-only work, maybe supervised by humans." The framing matters because it positions AI not as a deployment target but as a component of organizational architecture. One layer in a system that must be designed deliberately, not accumulated incrementally.
Krzyzek connected structure to the engagement question he had opened: when an organization is labeled by function rather than organized around customers, the function label immediately creates separation. Teams oriented around medical, commercial, and digital will optimize for functional metrics. Teams organized around customer segments will optimize for customer outcomes. The organizing principle determines what gets measured, what gets resourced, and ultimately what value the organization delivers.
Data That Decides, Not Data That Accumulates
Finlay drew a distinction that clarifies where most AI investment in pharma is currently landing: "Data accumulation alone is not sufficient. We have to go beyond accumulation towards decision systems." The industry has spent a decade building data infrastructure. The competitive gap is now about what that infrastructure produces — and most of it still produces reports, not decisions.
Canamasas grounded the stakes in precision oncology's clinical arithmetic. "In precision oncology, the first thing we need to achieve is finding those patients that may only be 2 to 4% and have a mutation." At those target rates, data architecture isn't a commercial efficiency question. It is the determinant of whether the right patient receives the right therapy.
Mandry offered an analogy that reframes where pharma actually sits in the data-to-insights hierarchy. He described a conversation with a former Uber CMO: "Uber talks about our companies as data rich, insights poor, whereas we, the legacy companies, are insights rich, data poor." That framing was from several years ago, and the gap has narrowed. But the insight points in a counterintuitive direction for oncology specifically. A smaller, more defined target population doesn't produce less intelligence; it can produce more actionable intelligence, because the clinical stakes and the specificity of need are both higher. Volume of data is not a prerequisite for quality of insight. The prerequisite is an organizational structure capable of turning signal into decision.
The Value That Survives
The panel didn't resolve the paradox it named. No one could. But across four companies at different stages of the same structural challenge, a coherent direction emerged.
Mandry's distinction between personalized and personal experience points toward what that direction looks like operationally. Personalization is algorithmic, it tailors content based on inferred preferences. Personal engagement addresses a customer need, which requires understanding what the customer is actually trying to accomplish. In Oncology, a physician's need isn't only information about a drug. It's confidence that the right patient gets the right therapy at the right time. The engagement that serves that need looks different from the engagement that delivers product messages through preferred channels.
Canamasas reframed the entire disruption from the patient's vantage point: "The opportunity that this transformation gives us is the empowerment of patients. The fact that the first point of contact for many of us is going to be these artificial intelligence platforms gives an opportunity for patients to be much more empowered than they've ever been before."
Finlay's closing provocation, whether the missing voice was the patient or the family caregiver who should have occupied an additional chair on stage, points to the structural implication most companies haven't yet absorbed. Borderless engagement isn't a channel architecture. It's a decision about whose outcomes the organization is actually accountable for. Bayer Italy's DSO model made that accountability explicit by building it into the operating structure.
The companies that haven't made that choice yet aren't behind on technology. They're behind on a prior question: what value do we exist to deliver, and to whom, in a world where information asymmetry is no longer a defensible advantage?
To get you highlights of Pharma 2026 faster, we are using generative AI technology to summarise the transcripts of the sessions. If you have any feedback about the summary, please contact lucy.fisher@thomsonreuters.com.
Discover more on this topic at Pharma Commercial Data & Tech Europe 2026 (4-5 November, London) Europe’s collaborative home for data and tech pioneers. Visit the website here.
