Regulatory Affairs Is the Connective Tissue Pharma’s AI Strategy Is Missing
Contributed Commentary by Hans van Leeuwen
October 29, 2026 | Every major pharma organization I’ve come across in recent years has its own version of the same AI-based ambition. It is looking to use the technology to compress the R&D timeline, reduce late-stage attrition, and make regulatory submissions faster and inspections less tense. And yet, almost without exception, progress will stall because the product data foundations are flawed.
Even today most pharma companies still lack a single authoritative version of their own product data. Clinical, manufacturing, quality, and pharmacovigilance teams operate from systems designed for their own respective purposes, with the result that any cross-functional anomalies are invisible until highlighted during an integration project, a regulatory inspection, or an AI initiative. I’ve seen regulatory and PV teams attempt a seemingly straightforward data-sharing exercise only to discover that their product databases have drifted so far apart that the same compound appears under different identifiers, with different attribute values reconciled through years of manual workarounds. Putting things right can mean consequential changes across a dozen downstream dependencies. That’s before anything has been submitted.
AI Creates Urgency Around Product Data Consistency
The data fragmentation problem has existed for a long time, and organizations have mostly lived with it. A big difference now comes from companies’ operational expectations around AI. AI tools fed conflicting versions of the same data point will either fail outright or, more dangerously, produce outputs that may look credible but can’t be trusted. No amount of investment in model sophistication compensates for that. The data architecture has to be fixed first, and the people who do that need to understand product data well enough to spot when something is wrong and what that could mean downstream.
This is where regulatory affairs (RA) comes in. No other function has as complete a picture of a medicine across the full lifecycle. RA sits at the intersection of clinical, manufacturing, quality, and safety data, and has historically been the function that catches inconsistencies before submissions reach a health authority. But managing submissions is too narrow a description of what the function could be doing, given the data at its disposal. I would argue that RA should serve as an organization’s custodian of trusted product data; not just for submission purposes, but also for every internal function that might need that data.
Regulators are beginning to expect exactly this kind of proactivity. The FDA’s Real-Time Oncology Review program is just one indicator of evolving regulatory expectations. By allowing review of critical efficacy and safety data to begin before a complete application is assembled, RTOR requires that data be submission-ready well ahead of the conventional timeline. This isn’t a submission management challenge, but rather a data governance challenge. Someone needs to own the pipeline of trusted product information that makes early review possible.
Who Currently Owns the Governance Gap?
The standard explanation given for why cross-functional data projects stall surrounds cultural resistance—that functions don’t want to share or are protective of their own systems. But that’s rarely the issue I see. A more common problem is that nobody has ever mapped the connections between silos well enough for people to understand that they’re working from the same underlying data—in different systems, under different labels, interpreted differently. Once that mapping exists, the case for alignment usually makes itself.
This might suggest a major governance initiative, but this is more about inserting regulatory data stewards in optimal places: people who are sufficiently data-literate to recognize when the same product element appears in two different forms and understand the downstream consequences of any discrepancies.
There also needs to be a master data strategy: a unifying approach to regulatory information management that holds product data to a common standard across systems, avoiding costly re-engineering every time a technology platform changes.
ISO IDMP standards provide the common vocabulary that makes this achievable, allowing the same data point to carry the same meaning whether it sits in a quality system, a clinical database, or a PV platform. European compliance deadlines are already concentrating minds on implementation, but the opportunity is larger than conformance for its own sake. Once product information is standardized and governed properly, the same data assembled for a submission can be distributed in real time to every function that needs it, feeding PV, informing supply chain, and supporting manufacturing quality. The submission then becomes a by-product of good data management rather than the reason for it.
Envisioning a Different Kind of Regulatory Function
The core work of regulatory submission management won’t go away, but it will shrink as a proportion of what RA does and this is a good thing. Structured data pipelines and AI will progressively absorb the routine compilation work that currently consumes significant resource. What should expand in its place is the data governance and stewardship remit, in other words maintaining the integrity of the product data record, arbitrating ownership questions across functions, overseeing AI outputs that draw on product data, and serving as the organization’s primary point of contact with agencies on evolving data standards.
This future role for RA professionals requires a different and evolving profile. Regulatory expertise combined with data literacy and a working understanding of enterprise data architecture isn’t a common combination today, but organizations that build it now will be positioned to compress development timelines, reduce inspection risk, and extract real value from AI as the technology matures. This won’t be because they’ve bought the right tools, but because the data those tools depend on can now be trusted.
The function that has always been closest to the product’s complete data story has the clearest claim to take on this remit. If regulatory affairs doesn’t, another function will, and probably less effectively.
Hans van Leeuwen is a regulatory affairs and life sciences technology expert with more than 25 years of experience spanning pharmaceutical development, regulatory operations and data management. He most recently served as Global Head of Regulatory Affairs Data and Systems at Astellas Pharma, where he led global regulatory operations and oversaw initiatives involving eCTD, IDMP, XEVMPD and other regulatory submission and data standards. Drawing on decades of experience at the intersection of regulatory affairs, technology and business transformation, Hans is currently a regulatory affairs consultant to ArisGlobal. He can be reached [email protected]. This article builds on themes discussed in a recent life sciences industry podcast, which can be accessed in full here.