Why Pharma's Digital Transformation Needs Human Judgment More Than Ever

By Phanindra Venkata Gottipati CEO, VirtuNx | Business Head, Seosaph Infotech

Over the last few years, I have spent a lot of time thinking about digital transformation from a product perspective.

More recently, conversations with leaders and teams across generics, API manufacturing and pharmaceutical development have challenged some of my earlier assumptions.

One pattern keeps surfacing. The hardest part of digital transformation is rarely introducing another technology.

It is changing how people trust information, collaborate across functions and make decisions from it.

That becomes even more important as AI enters the picture.

Today, almost every technology conversation eventually reaches similar questions:

• Can AI predict project delays?

• Can it identify attractive molecules?

• Can we automate portfolio reviews?

• Can AI monitor regulatory and competitive signals?

• Can it recommend which projects deserve investment?

The technology is becoming increasingly capable. But I believe the more important question for pharma executives is:

What decisions are we comfortable allowing technology to influence, and where must experienced human judgment remain central?

Pharma Doesn't Have a Data Shortage

Consider how a generics company evaluates a molecule.

Strategy may look at market attractiveness, loss of exclusivity, competitive intensity and expected economics, R&D considers development feasibility, Regulatory teams assess the pathway and filing implications, API teams evaluate sourcing or manufacturing complexity, Commercial teams understand customers and pricing while Finance evaluates investment and expected returns.

The information exists across functions, external sources, spreadsheets, systems and, critically, the experience of people.

The problem is connecting it. This is why I increasingly believe that digital transformation in pharma should not simply be about digitizing information. It should be about improving the decisions made from that information.

A Three-Month Delay Isn't Just a Red Milestone

The same principle applies after a molecule enters development.

Imagine an ANDA development program where a critical milestone is expected to slip by three months.

A traditional project-management system may turn the milestone red. Useful, but incomplete.

A portfolio leader needs to know:

Why is it delayed?

Does the delay affect the critical path?

Does it move the expected filing date?

Is a scarce resource creating the constraint?

Can leadership intervene?

And most importantly:

What does three months mean commercially?

If the delay affects an important market-entry window, what looks like an operational project issue can quickly become a business problem.

That connection, from execution signal to business consequence, is where technology becomes far more valuable.

AI Can Connect Signals Humans Cannot Continuously Monitor

This is where I see significant potential for AI.

A portfolio may contain dozens of molecules and hundreds or thousands of underlying activities, dependencies, risks and decisions.

At the same time, the external environment continues changing.

Competitors progress, Regulatory information changes, Patent situations evolve, Customer signals emerge, Market assumptions move, No executive or project team can continuously connect every signal manually.

AI can help identify patterns, summarize evidence, detect emerging risks, challenge assumptions and surface information requiring attention.

But there is an important distinction - AI can tell us that something deserves attention, Experienced people determine what it means.

That distinction is especially important in an industry where scientific uncertainty, regulatory requirements, manufacturing realities and commercial economics intersect.

What Customer Conversations Are Changing in How We Build

This thinking has increasingly influenced our work at VirtuNx.

When we think about PortiVix, for example, it is easy to describe the problem as pharmaceutical portfolio and project management.

But customer conversations have pushed us toward a more interesting question:

How do we connect what teams experience during execution with what leadership needs to know to make portfolio decisions?

A project being green, amber or red is only the beginning. Leadership needs context.

Similarly, while working on MorViac around molecule intelligence, we could theoretically build increasingly sophisticated scoring models that tell strategy teams which molecules appear attractive.

But a score without explainability can create false confidence.

The more important capability is helping leaders understand the evidence, assumptions, uncertainty and organizational context behind the recommendation.

That is a very different philosophy from asking AI to make the decision.

Adoption Begins With Trust

This may also explain why some digital-transformation initiatives struggle with adoption.

We often treat adoption as a training or change-management problem. Sometimes it is.

But people also resist systems when the system doesn't reflect how they actually make decisions.

If an experienced formulation leader believes an algorithm has ignored an important technical constraint, they will return to the spreadsheet. If a strategy leader cannot understand why AI recommends one molecule over another, they may ignore the recommendation.

If project teams spend significant time maintaining a system but leadership continues making decisions from manually prepared presentations, the system becomes administrative overhead rather than transformation.

People adopt technology when it becomes useful to their judgment, not simply because the organization deployed it.

From Automation to Augmentation

This has changed how I think about enterprise AI.

The first wave of digital transformation focused heavily on digitization. The next focused on automation. AI now gives us an opportunity to move toward something more powerful: Augmentation.

Give the portfolio leader earlier warning.

Give the project leader better context.

Give the strategy team continuously updated evidence.

Give leadership the ability to test assumptions before committing capital.

And give everyone a more connected version of reality from which to make decisions.

The FDA's evolving approach to AI is instructive here as well. Its frameworks increasingly emphasize the credibility and context of AI use rather than treating an algorithm as inherently trustworthy simply because it performs well technically.

For pharma organizations, I believe the same principle should apply internally.

The question shouldn't only be: “How intelligent is our AI?”

It should be: “Does it help our people make better decisions?”

After years of building products, and now learning directly from pharmaceutical customers, that distinction has become increasingly important to me.

The future of AI in pharma shouldn't be about removing humans from important decisions. It should be about giving experienced people something they have never had before: the right evidence, the right context and the right signal at the moment a decision needs to be made.

That is where I believe digital transformation becomes genuinely valuable.

I’d be interested to hear your perspective, where do you believe the right balance between AI and human judgment lies?