Vinayak Acharekar coauthored this article.

Pharma’s AI focus is shifting from isolated tools to coordinated execution. While early use cases in forecasting, analytics and generative content delivered value, most intelligence has remained siloed. Agentic AI changes that by deploying autonomous agents that work across systems to streamline entire workflows, not just tasks.

The 2025 ZS CDIO outlook shows this change is already underway, with 45% of enterprise IT and 41% of R&D discovery leaders aiming to streamline end-to-end workflows with AI agents. Meanwhile, in commercial, 36% plan full workflow transformation and 46% are automating specific activities in the next year.

The direction is clear. AI is moving from experimentation to enterprise orchestration. As this shift evolves, new questions arise about where intelligence should be designed, owned and scaled. Increasingly, the answer is global capability centers (GCCs).

Why GCCs are a natural place to start scaling agentic AI

No longer considered centers for cost savings, pharma GCCs now sit at the intersection of data platforms, digital infrastructure, talent and governance—making them natural environments to pilot, prove and scale agentic solutions:

• Centralized data and analytics enable reuse across markets
• End‑to‑end visibility supports multiagent orchestration
• Integrated domain and engineering talent enables effective supervision
• Mature risk and compliance frameworks embed guardrails from the outset.

Together, these strengths position GCCs as test environments for codifying learnings and scaling intelligence enterprisewide.

Agentic AI in GCCs follows a maturity journey, not a switch

As ambition around agentic AI accelerates, many organizations and their GCCs are still navigating the organizational change required to move from today’s AI applications to a truly agentic cognitive enterprise.

This journey requires rethinking not just where AI is applied, but how intelligence is designed, embedded and scaled across the organization. In a truly agentic enterprise, intelligence becomes ambient by continuously sensing context, coordinating across systems and acting with increasing autonomy within policy-driven guardrails.

Work shifts from linear execution to dynamic orchestration, with humans and agents operating as integrated teams. Decisions are embedded directly into workflows rather than routed through static handoffs.

Progressing across 3 levers of agentic AI maturity in GCCs

Achieving agentic AI maturity typically involves progressing across three levers for a pharma GCC:

• Reengineering your GCC operating model
• Scaling agentic AI with the right roadmap and governance
• Ensuring an agentic AI-driven roadmap allows for human-AI collaboration

Lever 1: Reengineering your GCC operating model

Reengineering for agentic AI typically starts by building an AI-native operating model in which agents handle most execution and humans focus on context, judgment and key decisions. One way this evolution takes shape is through redesigning workflow archetypes so that execution-heavy activities—such as data ingestion, KPI generation, driver analysis, scenario exploration and narrative drafting— move toward autonomous agents.

Human contribution increasingly centers on shaping the right questions, validating assumptions, reviewing outputs and managing edge cases. Again, the goal is to create an AI native operating model where agents take on most execution, while human expertise anchors contextual decision heavy activities.

Lever 2: Scaling agentic AI in GCCs with the right roadmap and governance

As agents become more autonomous and take on more responsibility, governance must evolve with them. It should shift from a compliance function to being an embedded system that ensures quality, trust and controlled scaling.

Behavioral shifts are key to a smooth transition. Organizations increasingly build trust through measurable accuracy gains, reshape habits by making copilots the default entry point, and hardwire capability through role‑based upskilling.

Two elements typically anchor this evolution:

• A layered technical architecture that brings together context‑rich data products, reusable analytics engines, an agent layer and NLQ‑driven copilots
• The responsible expansion of agentic workflows through reviewing exceptions, auditing outputs and managing escalations

Together, these two shifts enable autonomy without sacrificing rigor.

Lever 3: Ensuring an agentic AI-driven roadmap allows for human-AI collaboration

As GCCs become more mature, they reorganize into lean, cross‑functional teams that blend domain expertise, analytics and engineering. And it’s vital they’re supported by a culture that values experimentation, accountability and collaboration between humans and agents.

What does this look like in practice? Analysts expand into AI oversight, stewardship and prompt engineering roles, while business users increasingly rely on NLQ-driven assistants for self‑service intelligence. New specialties—including data engineering leads, AI configurators and automation orchestrators—begin operating alongside redefined analysts who frame questions, stress test AI outputs and shape decisions with business partners.

This shift requires more than new skills. It demands a cultural transition from task execution to ownership, judgment and continuous learning. Changing your mindset from one of risk avoidance to one of experimentation is also important.

Work is standardized into “skill packs”—reusable bundles of domain and functional skills—in mature GCCs. As humans and agents operate together, headcount alone becomes an insufficient planning metric. Instead, digital tech equivalents (DTEs) help leaders measure the combined capacity of people, agents and platforms, enabling more accurate scenario modeling, productivity targets and value tracking across reengineered workflows.

Enterprise platforms help accelerate the GCC journey to agentic AI maturity

Scaling agentic AI reliably requires a strong enterprise platform, one with context‑rich data products, reusable analytics engines and a governed agent layer embedded into the tools teams already use.

An enterprise platform mindset is what moves organizations from isolated assistants to coordinated agent ecosystems, making intelligence repeatable, observable and easier to manage at scale. Forward‑leaning pharma GCCs are already achieving success with this approach, as platform‑led design reshapes how analytics work is executed and how teams organize around it.

Case study: How AI is transforming a pharma GCC into an analytics engine

ZS is working with a global pharma company to bring together strong data foundations, AI-native analytics and new ways of working to drive faster, smarter decision-making. Generative AI agents and context-rich data products will enable the company’s GCC ecosystem, automate repetitive production work and strengthen the overall analytics structure.

The initial pilot with the company’s U.S. team delivered 40%-45% savings in analytics execution within nine to 12 months. It also doubled speed to insight and freed analysts to focus on scenario planning and strategic problem solving, leading to 60%-75% savings longer term.

ZS’s work will help this pharma organization and its GCC build strong data foundations and establish a clear analytics structure that doubles its speed to insight.

From early pilots to enterprise scale: Closing the gap in agentic AI adoption

The industry is increasingly aware of what’s possible when agentic AI is applied deliberately across workflows, platforms and roles. But companies like the one detailed in the case study represent the leading edge—not the mainstream.

Most pharma organizations today still sit between the early stages of adoption. Their pilots are active, but scaling is constrained by legacy operating models, fragmented ownership and talent structures optimized for execution rather than orchestration.

Even in other industries, GCCs have embedded agentic assistants into knowledge and service intensive workflows:

• Thomson Reuters’GCC led augmentations of unified knowledge repositories and workflow automation
• Walmart has gone even further by positioning their GCC in India as a global owner of agentic capabilities, where reusable agents are embedded directly into day to day developer, operations and business workflows

These two examples show that agentic AI adoption often unfolds incrementally, with organizations simultaneously experimenting, scaling and laying foundations across different parts of the enterprise.

The path forward for agentic AI in GCCs

How can your pharma GCC realize the full value of agentic AI? Success starts with rethinking a few core foundations. We recommend a pragmatic approach focused on near‑term horizons:

• First 30 days: Select one high‑volume, low‑risk workflow with clear ownership and success metrics
• First 60 days: Reengineer the workflow around humans and agents working together; stand up core data products and governance
• First 90 days: Scale what works, retire manual steps and introduce shared productivity metrics to guide the next wave

Agentic AI will not simply make GCCs more efficient. It will redefine how intelligence is created, governed and scaled across the enterprise. The organizations that succeed will start small but think end to end by deliberately choosing the right workflows, redesigning them for human-agent collaboration and scaling with discipline.

Pharma GCCs are no longer support centers. They’re the architects of intelligent, orchestrated organizations.

Chaitali Chaitali, Nitesh Toshniwal, Dhvani Maheshwari and Ritika Mahadevan contributed to this article.