Why capitalizing on AI will require a new operational model

An Expert View from Rob Goodwin, Chief Operating Officer, Parexel.

It’s indisputable: AI has the potential to shorten cycle times in clinical development. Across the industry, organizations are already seeing a significant reduction in the typical time spent on labor-intensive workflows. But while gains have been realized within functions, there is still significant opportunity across the development spectrum.

The solution, however, isn’t to layer AI over legacy processes, as this will exacerbate the fragmentation that exists in current operational models. AI can only transform clinical development if we’re willing to first transform our organizations and challenge the status quo of how trials are conducted.

Through AI and advanced data capabilities, we have the chance to not just accelerate clinical trials but evolve the way we deliver them. Traditionally, our workflows have been structured around functions, with a focus on moving projects from one handoff to the next. This model made sense in the pre-AI era, when information constraints required deep specialization — clinical operations, data operations, biostatistics and pharmacovigilance each working in isolation.

Two fundamental shifts are reshaping this approach. First, AI enables teams to access integrated data that informs cross-functional insights in real time. Second, organizations are breaking down silos through collaboration, with teams learning across domains and sharing responsibility for outcomes. The result? A workforce that thinks and acts not as isolated specialists but as orchestrators, driving change across the trial continuum.

Beyond functional boundaries: Research driven by purpose

In traditional models, data is locked in function-specific systems, preventing downstream teams from accessing timely information. Teams working within these silos may not fully understand the consequences of their decisions on other functions. This creates a cascading problem: reviews, which require manual aggregation from multiple sources, occur late in the process at predetermined milestones in episodic batches. As a result, AI can only be applied within individual silos rather than across the full data landscape.

Consider how this plays out at clinical trial sites. Site interactions are generated by people in a range of roles: CRAs, data managers, vendors, medical reviewers and subject matter experts, among others. At the same time, data is aggregated in multiple steps from a myriad of sources, including EDC, eCOA and ePRO, lab results and wearable-technology vendors. This aggregation creates even more interactions.

At Parexel, we’re addressing this challenge through Integrated Data Delivery (IDD), which centralizes data capture and review within a unified technology platform. Rather than requiring multiple rounds of verification across single-function systems, we use ParexelAI to consolidate data from all sources. Then clinical data monitors (CDMs) work with sites to resolve all issues at one time. With this process, SDV and SDR become confirmatory steps rather than laborious efforts. The result is a better experience for everyone: sites spend less time on data queries and more time on patient care, sponsors get cleaner data faster and patients benefit from less administrative burden during their trial participation. The result is a more streamlined experience for sites, sponsors and patients alike. We see this philosophy as “Every Patient, Every Visit.”

The CDM is integral to the success of the IDD approach. CDMs serve as a primary point of contact for sites in all data-related matters, from data review and query management to site visit coordination. This streamlines communication and planning for sites and study teams alike. IDD also redefines the role of the clinical research associate (CRA). By significantly reducing the data validation work required of CRAs, we allow them to be site managers instead of data cleaners. This fundamentally improves the relationship between sponsors and sites by freeing CRAs to provide site support, education and patient safety oversight.

Equipping an AI-ready workforce

In a recent industry analysis, Parexel found that only one in five biopharma leaders believe that drug development professionals are well prepared to use AI. Adopting AI will therefore require a new approach to staff training and development. Instead of task execution, we must prioritize critical thinking and proactive, informed decision making.

With human-in-the-lead AI, workers are active participants in AI-generated insights. This includes providing input on how AI systems are developed and assessing AI outputs for accuracy and reliability. This is where developers and subject matter experts work shoulder-to-shoulder to develop solutions that solve the true business need.

With the advantage of near-real-time data, we can reduce reliance on siloed expertise, equipping staff to be data-literate integrators who collaborate in parallel throughout the trial rather than working in prescribed sequence. As a result, we will shorten review cycles and greatly reduce the possibility of overlooking critical details — both of which will lead to higher quality decision making, speed and improved safety monitoring. For example, we can now ingest a protocol and immediately provide competitive intelligence on inclusion and exclusion criteria, helping sponsors optimize a trial design before enrollment begins. We've also developed AI-powered prompts that automate the creation of clinical trial documents — from CSRs to informed consent — dramatically reducing the time from database lock to reporting. Shaping our processes around critical data also allows us to detect and mitigate risk earlier and to learn continuously across the trial lifecycle, adjusting with responsiveness and agility.

Of course, realizing these benefits also requires strong governance frameworks, clear accountability for AI-assisted decisions and continued investment in workforce readiness.

Designing a new approach

AI is part of our landscape and organizations that adopt new operational models and embrace enterprise-level change will be best positioned to capitalize on what advanced data solutions offer. Rather than incremental optimization, this is an opportunity to rethink the building blocks of clinical trial execution.

As you begin to reimagine your approach to clinical trial delivery, ask:

• Which decisions should be made earlier? Real-time data should enable faster, more confident decision making at critical trial junctures.
• Which handoffs can be eliminated through cross-functional processes? Cross-functional cooperation reduces delays and the errors that can come with a functional mindset.
• What work should disappear entirely? Some activities, like redundant data reviews, may no longer be necessary in a real-time data environment.
• How can we incentivize integration over optimization? Reward teams for trial-level outcomes, not function-specific metrics.
• How will we define success? Tracking timeline adherence is valuable but we recommend using additional benchmarks such as decision confidence, time from data generation and site satisfaction.

AI isn't simply a tool for efficiency, it's a catalyst for redesigning outdated clinical and operational models. With a modernized approach to clinical trial delivery, we can better integrate data, learn from it earlier, adapt to it with greater agility and bring critical therapies to market faster — while simultaneously reducing the administrative burden on patients who make clinical research possible. By designing trials around data and collaboration rather than fragmented processes, we help patients spend less time navigating trial logistics and increase our ability to provide care, leading to better retention and faster paths to market for life-changing treatments.