Real-Time Trials Need More Than AI: Why Operational Trial Infrastructure Matters

The FDA’s recent announcement to explore real-time clinical trial proof-of-concepts and to pilot artificial intelligence (AI)-enabled optimization in early-phase research marks an important signal for the industry. The agency’s direction is clear: we’re moving to earlier visibility of safety signals and endpoints, better-informed early decisions, and a more efficient path through clinical development. But the announcement is not simply a path forward for AI. It’s also a critical directional signal on how oversight should and must be executed.

Real-time review changes the tempo of clinical development. Historically, trial data has moved from site to sponsor to regulator in stages, often with significant lag. Now, the FDA wants to explore models in which predefined signals can be reviewed far earlier, potentially reducing the dead time between study milestones and improving how quickly promising therapies advance. In parallel, the agency’s proposed pilot asks how AI-enabled technologies might improve efficiency, safety monitoring, dose selection and early go/no-go decisions, while still meeting rigorous scientific and regulatory standards and aligning with trustworthy AI principles.

That matters because AI can only support better decisions if the trial itself is operationally capable of responding. A recommendation engine, a dose model or an earlier safety signal has limited value if the downstream study machinery cannot keep pace. In practice, real-time research depends on controlled randomization, maintained blinding, reliable supply continuity, change control, auditability, and coordinated updates across connected systems and stakeholders. These are not fringe concerns; they are the mechanics that determine whether a real-time insight can become a real-time action.

This is where operational trial infrastructure becomes strategically important. In our experience at Almac, leveraging an IRT platform purposely designed to support patient randomization and trial supply management across study phases, including simple-to-complex adaptive designs, real-time allocation methods, blinding controls, reporting and integrations—has been crucial. Taken further, we’re also utilizing a broader workflow-focused eClinical environment that brings together IRT, eCOA, eConsent, visit management and sponsor oversight features completes this operational infrastructure. The real differentiator in this emerging model is not just access to more data, but the operational readiness to keep protocol, participant and supply activities aligned and act on new information without disrupting trial execution.

The key point is not that IRT can “do AI.” It is that the operational systems of the trial must be fit for a more adaptive model of development. If a cohort opens, pauses or closes, randomization rules may need to change. Supply assumptions may need to change. Visit workflows, tasking, consent handling and participant communications may also need to change. In more complex studies, the challenge is often less about any one algorithm than about the ability to coordinate the ripple effects of change across the study ecosystem without compromising data integrity or site execution. That is why the most credible industry conversation is about orchestration, not hype.

For sponsors, this has an immediate implication. The real readiness question is not just “Do we have AI?” It is “Can our trial operating model absorb faster information and respond in a controlled way?” Organizations that still rely on fragmented workflows, manual reconciliations and slow amendment propagation may find that real-time review exposes operational bottlenecks rather than removes them. By contrast, sponsors that invest in operationally disciplined infrastructure covering randomization, supply, workflow coordination, and governed integrations will be better placed to explore the next generation of adaptive and data-responsive trial execution.

The regulatory direction is exciting, but it should be interpreted carefully. The FDA has opened the door to new models of visibility and decision-making. The winners will not be those who talk most loudly about AI. They will be those who combine trustworthy analytics with operational systems capable of turning insight into action, quickly, compliantly and at global scale.

Jeremy Jakubowski, Director - Technical Strategy and Delivery; and Richard Wzorek, Director – New Products & Services; both with Almac Clinical Technologies