Merck’s GenAI Journey: From Experiments to Enterprise Value

Generative AI is reshaping expectations across the life sciences industry, yet many organizations still sit between big ambitions and the reality of scattered pilots, uneven outcomes, and uncertain adoption. Yet many organizations remain caught between bold visions and the reality of scattered pilots, inconsistent results, and hesitant users.

At the Indegene Digital Summit 2025 Virtual edition, a session featuring Keyur Brahmbhatt (Head of Data Digital and IT (DDIT) Business Partnering for RQS at Merck Healthcare R&D) in conversation with Nikesh Shah (VP, Safety and GenAI Solutions at Indegene) offered a rare moment of clarity. It unpacked how Merck moved past the early wave of GenAI excitement and built the foundations for scalable, enterprise-level impact. As Merck’s progress stands out among emerging AI success stories in life sciences, its journey provided a practical look at what it truly takes to move from experimentation to sustained value in complex, regulated, and knowledge-driven environments.

The discussion placed GenAI’s evolution in context, acknowledging how the industry has moved from peak hype into the “plateau of productivity as per Gartner Hype Cycle. The initial excitement that followed the launch of large language models pushed many organizations into rapid experimentation. But the momentum also created pressure—teams rushed to try new tools without aligning them to real needs, mature workflows, or long-term goals. This shift makes a clear AI implementation strategy essential, especially in regulated, knowledge-driven environments.

Merck framed this moment as a pivotal transition: the industry is moving past the peak of excitement and into a more serious phase where productivity, governance, and measurable outcomes matter more than novelty. The session emphasized that a successful AI implementation is a dual journey. On one side sits the evolution of model capabilities. On the other sits the equally important—but often overlooked—people-centric change. This includes upskilling teams, building comfort with AI-assisted workflows, and fostering an environment where experimentation is encouraged rather than resisted. Without user confidence, even the most advanced tools struggle to drive sustained impact. GenAI succeeds only when people trust what it delivers and understand how to work with it.

Unlike many AI success stories that highlight only early wins, this journey focuses on the strategic levers and operating model required for long-term value. Merck approached its early GenAI initiatives differently. Instead of viewing pilots as isolated tests, the organization treated them as building blocks of a broader transformation.

The team focused on areas where GenAI already demonstrates strong potential, such as medical and regulatory content creation. But the intent was not to automate individual deliverables. Instead, the goal was to reimagine how entire workflows could be redesigned to reduce complexity, shorten timelines, and elevate quality.

This strategic framing helped in avoiding a common pitfall—pilots that generate interest but fail to integrate into real operations. By treating pilots as part of an end-to-end redesign, the organization created clarity on where GenAI fits and how it could unlock cumulative efficiencies.

One of the most compelling insights from the session was how Merck approached the transition from pilot activity to production deployment. This shift rarely happens organically. It requires intentional design, governance, and alignment.

Merck’s internal program, known as the “GAME Plan” (Generative AI for Medical), started with a focused set of deliverables in the medical domain. It helped with a structured AI implementation strategy that balanced experimentation with governance and prioritization. Its AI pilot program began with a narrow set of medical deliverables to validate feasibility before scaling. Smaller, well-defined outputs made it easier to test feasibility, refine prompts, and standardize processes.

A key enabler was access to an internal sandbox environment that acted as a safe space for the AI pilot program. This made experimentation accessible to business teams and lowered the technical barrier to trying new workflows. Users could explore ideas without relying heavily on engineering resources, creating faster cycles of learning.

Equally important was the balance between bottom-up energy and top-down direction. Pilot teams explored what worked. Leadership introduced a value-based framework to determine which use cases deserved scaling. This combination ensured innovation was encouraged, but decisions about investment and roll-out were grounded in business relevance.

The discussion stressed that scaling GenAI requires more than technology—it required shifting mindsets from passive adoption to active piloting. This involved not only upskilling teams but also nurturing a culture where AI was seen as a collaborator, not a black box. A recurring takeaway was the importance of choosing the right early use cases. Merck used a value-versus-complexity framework to prioritize opportunities. Low-complexity, high-impact tasks were selected first, not because they were easy, but because they built confidence. Insights from the AI pilot program informed the value–complexity grid, helping teams identify where scale made the most sense.

This approach helped reduce early friction. Teams could see tangible results quickly, which encouraged participation and created a strong foundation for broader adoption. It also allowed the organization to refine its governance, quality processes, and operating models before scaling to more complex use cases.

The discussion reinforced that early momentum matters. When early use cases deliver clear value, they become internal proof points that accelerate broader transformation.

Another important dimension of Merck’s journey was its emphasis on quality over volume. Many organizations fall into the trap of producing more GenAI output without ensuring it is usable. This often leads to what the session described as "workslop"—content that still demands heavy manual review because it lacks the necessary structure, depth, or accuracy.

To avoid this, Merck built a purpose-built platform called MyCo-Creator, tailored to narrow but high-value applications. This ensured that output met the expectations of reviewers and aligned with regulatory requirements. The platform delivered a level of consistency and accuracy that encouraged adoption rather than skepticism.

High-quality, domain-specific outputs proved essential for successful AI implementation, especially when scaling across regulated teams. This decision played a pivotal role in enabling a shift from experimentation to reliable, repeatable execution.

The narrative from the session makes one point especially clear: scaling GenAI is not a technical exercise. It is an organizational shift. A disciplined AI implementation strategy helps organizations ensure that technology, people, and processes mature in parallel.

For leaders seeking practical AI success stories grounded in real-world execution, Merck’s experience offers a blueprint worth studying:

As the industry continues navigating the shift from hype to meaningful impact, this journey offers a blueprint for moving forward with confidence. Successful AI implementation ultimately requires alignment across technology, governance, and culture—an insight reinforced throughout the session.