1. AI is now embedded in nearly every pharma strategy. Where do you see the biggest disconnect between strategic intent and operational execution?
The intent is no longer the challenge. Most Pharma organisations are clear they want faster, more confident, and data-backed decisions. The real disconnect begins when AI is introduced as a layer of intelligence, but the underlying decision-making process remains with individuals.
Today, while over 70 per cent of pharma companies are investing in AI, a large percentage of initiatives struggle to scale beyond pilots. The reason is simple, AI generates insights, but the decision-making process remains fragmented.
In practice, teams still rely on multiple dashboards, offline discussions, and manual reconciliation. I’ve seen cases where strong predictive insights existed, yet decisions were delayed because stakeholders were aligning on different versions of the same data.
AI is being embedded into systems, but not into decision workflows. Until organisations bring structure, consistency, and real-time visibility into how decisions are made, AI will continue to sit beside the process, not drive it. The real transformation lies in integrating AI into decision-making itself.
2. Despite heavy investments, why do most AI initiatives in pharma remain confined to pilot programs rather than scaling enterprise-wide?
I wouldn’t say most initiatives are confined to pilots. What we’re seeing is a phase of cautious expansion, which is natural in a regulated industry like pharma. However, scaling AI is fundamentally different from piloting it. Pilots succeed in controlled environments. Scaling exposes enterprise complexity.
I’ve seen a portfolio prioritization model work extremely well within a single function. But when extended across regions, inconsistencies in data definitions and program structures made the outputs difficult to trust.
The model didn’t fail, the operating environment wasn’t ready. Scaling AI requires consistency in data, clarity in ownership, and alignment across teams. These are not technical challenges. They are organisational ones. However, scaling AI is fundamentally different from piloting it. Pilots succeed in controlled environments. Scaling exposes enterprise complexity.
3. How critical is data readiness in accelerating AI-driven decision-making, and what are the most common gaps you encounter?
Data readiness is the single biggest determinant of whether AI actually influences decisions. Having data is not the same as being ready to use it for decisions. I’ve seen organisations with extensive datasets struggle to answer simple questions because definitions varied across teams. In another case, leadership reverted to manual judgment because they couldn’t trace how inputs were derived.
The gaps are rarely complex, they are basic: consistency, context, and traceability. When these are missing, confidence drops. And when confidence drops, decision-making slows. AI depends on trust in data. Without a strong data foundation, AI can generate insights, but it cannot drive decisions. And in a regulated industry like pharma, trust in data is not optional - it is essential.
4. To what extent do fragmented data ecosystems and legacy systems continue to slow down AI adoption across pharma organisations?
Fragmentation remains one of the most significant structural barriers to AI adoption in pharma. Most pharma organisations have evolved their systems function by function, leading to disconnected data environments. I’ve seen situations where clinical and commercial teams were both using advanced analytics, yet operating on completely separate data views.
Each function was optimising locally, but there was no unified understanding at the portfolio level. Fragmentation doesn’t just limit insight, it limits alignment. Without a connected view, decisions remain siloed, and AI cannot bridge that gap. True value comes when data is integrated across functions, enabling decisions that reflect the full enterprise context. The shift pharma needs to make is not just towards data integration, but towards decision integration where insights, context, and actions are aligned across the organisation.
5. What organisational or cultural barriers are preventing AI from being trusted in high-stakes decision-making processes?
Trust in AI is not built through models, it is built through clarity. I’ve seen AI-driven recommendations rejected simply because teams couldn’t explain how they were derived. In contrast, similar insights were accepted when assumptions were transparent and traceable.
The difference is not accuracy, it is understandability. There is also an accountability challenge. When AI is involved, ownership of decisions can become unclear, creating hesitation. Many leaders rely on years of experience, and they are open to AI, but not to opacity. If AI is a black box, it will be questioned. If it provides structured support, it will be trusted. Ultimately, trust is not a technology outcome; it is an organisational outcome. AI will be trusted when it is embedded into structured decision processes with clear ownership, transparency, and governance.
6. How can pharma companies transition from experimentation to embedding AI into core business workflows and daily decisions?
The transition happens when AI becomes part of routine decision-making, not a parallel initiative. I’ve seen organisations generate valuable insights through standalone dashboards that were rarely used in actual decision forums.
The shift occurred when those insights were embedded directly into workflows, linked to specific decisions, with clear ownership and outcomes. That integration was simple but powerful.
Embedding AI requires clarity: what decision it informs, who owns it, and how outcomes are tracked. When AI is connected to real decisions, adoption follows naturally. Without that connection, it remains an isolated experiment. The organisations that will scale successfully are the ones that redesign workflows so that AI, data, and human judgment operate together at the point of decision.
7. In your experience, is the challenge more about technology limitations or about aligning people, processes, and governance? Why?
At this stage, the challenge is far less about technology and far more about alignment. Technology has matured significantly, most organisations already have access to capable tools. What varies is how well people, processes, and governance are aligned around them.
I’ve seen organisations with strong analytics capabilities still struggle because different functions operate on different planning cycles and data interpretations. This leads to multiple versions of the same reality.
Technology can enable capability, but it cannot create alignment. Without shared definitions, governance, and coordination, even the best tools fail to deliver consistent outcomes. The organisations that will lead are not the ones with the best tools, but the ones that align their people, processes, and governance to act on insights consistently at scale.
8. How do regulatory requirements and compliance concerns impact the speed and scale of AI-driven decisions in pharma?
Regulation does not limit AI adoption in pharma, it defines how it must be implemented. In a highly regulated industry, decisions need to be not only accurate, but also traceable, explainable, and defensible. organisations must demonstrate what decision was made, how it was derived, and whether it can be consistently reproduced.
This naturally slows early adoption. However, when systems are designed with compliance in mind, ensuring clear data lineage and explainability, regulation becomes an enabler.
Over time, these compliant systems create consistency and trust across the organisation. And in pharma, trust is what ultimately determines whether AI can scale in high-stakes decision environments.
9. What role does crossfunctional collaboration play in unlocking real value from AI?
An essential one. Decisions in pharma rarely sit within a single function, R&D, regulatory, commercial, and supply all influence outcomes. When AI is implemented in silos, it produces fragmented insights.
I’ve seen teams achieve strong outcomes within their own functions, yet struggle at the portfolio level because those insights were not aligned across teams. The result is fragmented decision-making, where each function moves forward with its own version of priorities.
The real value of AI emerges when these perspectives are connected. A shared data context enables better coordination and more informed decisions. AI is most powerful when it integrates functions, not when it optimizes them in isolation. AI is most powerful not when it optimizes individual functions, but when it enables the organisation to act as a coordinated system.
10. Are current KPIs and ROI frameworks sufficient to measure the success of AI initiatives, or do they need to evolve?
Most current frameworks are not sufficient - they need to evolve.. They focus on cost savings and efficiency, which are important but incomplete. In pharma, the real value often lies in decision quality and speed - faster prioritisation, earlier risk detection, and better resource allocation.
These outcomes are harder to quantify but far more impactful.
Organisations need to evolve their metrics to reflect how decisions improve, not just how processes become efficient. AI’s true value is not just operational - it is strategic. The organisations that will lead are the ones that move beyond measuring efficiency and start measuring how effectively they make decisions at scale.
11. How can pharma leaders ensure that AI insights are not just generated but actually acted upon in decision-making cycles?
AI insights create value only when they are directly embedded into decision-making cycles.. I’ve seen organisations generate extensive AI-driven insights that remain underutilized because they are not integrated into decision forums.
The turning point comes when insights are linked to specific decision points, with clear ownership and accountability.
Adoption does not come from availability — it comes from relevance. When teams see how an insight influences a decision, usage becomes natural. Without that connection, even the best insights remain unused. Ultimately, the goal is not to generate more insights, but to ensure that every critical decision is supported by the right insight at the right time.
12. What strategies can help bridge the gap between data science teams and business stakeholders to improve adoption?
The gap between data science and business teams is less about capability and more about priority alignment. Data science teams focus on models and accuracy, while business teams focus on decision impact, outcomes and timelines.
Bridging this requires early collaboration —- defining problems together and ensuring outputs are interpretable and actionable.
It also requires shared metrics. Not just model performance, but decision impact.
When both sides align on outcomes rather than outputs, adoption improves significantly. The goal is not better models alone, but better decisions.
13. Looking ahead, what structural or technological shifts are necessary for AI to become a true decision-making engine in pharma?
Structurally, organisations need to move toward integrated operating models. Siloed functions limit cohesive decision-making. Shared accountability will become increasingly important.
Technologically, the shift is toward platforms that combine data, context, and decision workflows.
AI will move from standalone models to systems that support decisions end-to-end. The organisations that make this shift will not treat AI as a tool - they will embed it into how they operate. For AI to become a true decision-making engine in pharma, organisations will need to fundamentally shift both how they operate and how their technology is structured.
14. If you had to prioritise one change to accelerate AI impact across pharma organisations, what would it be and why?
One change, it would be to bring structure to decision-making. In many organisations, decisions are still made through discussions that are not fully captured. Inputs are scattered, and the rationale is often implicit.
I’ve seen cases where introducing a structured approach to decision capture created immediate clarity - and over time, revealed patterns across programs.
That structure becomes the foundation for AI to contribute meaningfully. Without it, AI remains peripheral. With it, AI becomes part of the core operating system.
“Closing Quote - AI cannot transform decisions until decisions themselves are structured.”
Read the full article — it's free
Register with Pharma Focus Asia to unlock expert insights, research articles and in-depth industry analysis.
