Clinical trials are increasingly mired in operational bottlenecks and administrative inefficiencies, delaying approvals of new and potentially lifesaving therapies. Between 2020 and 2024 even as the time between individual clinical trials decreased by an average of seven months, the average clinical trial cycle times (defined as the total duration from the approval of a clinical trial protocol to the database lock) rose by 14 months.
Additionally, more than 80 percent of AI investments in pharmaceutical R&D since 2015 have targeted the early discovery phases. While new AI models for drug discovery deserve serious attention – AI has been used for identifying novel targets in areas like cardiomyopathy, generating novel antibodies, and even designing newer modalities like optimized mRNA vaccines for influenza –AI for clinical development has not progressed at the same pace. Large pharmaceutical companies now spend more than $6 billion on R&D per approved drug, compared to just $40 million (in today’s dollars) in the 1950s, with approximately 85 percent of that spending coming after discovery. The inequality between drug discovery and clinical trials is creating a bottleneck, stalling promising drugs from reaching patients.
Trial inefficiency not only delays patient access to new therapies, it also places in increased burden on clinical research associates (CRAs) who are on the front lines of trial execution. Many CRAs are overwhelmed by the ever-growing burden of manual tasks, number of systems to monitor, and volume of data generated by a trial. They become ensnared in clinical trial ‘white space’ –extended periods of inactivity that can stretch on for months and hinder trial progress.
Yet agentic artificial intelligence (AI) and automation, even in their earliest stages of maturity, have the potential to mitigate white space and accelerate new therapy approvals, as well as redefine the trial process and the role CRAs will play in it. As AI matures, agents will become digital teammates, enabling CRAs to focus on higher-value work. .
The growing burden of inefficiency in clinical trials
Clinical trials are inherently complex, multiphased processes, that must be navigated with incredible care to ensure alignment with the protocol and strict regulatory standards. Efficiency for efficiency’s sake has no business in clinical research, but many bottlenecks are not safeguards. Rather, they have emerged from rapidly deteriorating administrative systems.
For perspective, IQVIA reports that "white space" currently consumes close to half of a new drug’s development time. However, due to the study’s limited definition of the term, that already startling statistic does not even account for the myriad productivity gaps that occur within every clinical trial conducted across an individual drug’s development lifecycle.
Tufts Center for the Study of Drug Development (CSDD) attributes increasing trial complexity to an upward trend impacting ALL protocol design variables, especially related to phase 2 and phase 3 trials. From site selection and activation to endpoints and patient eligibility criteria, trials are more complicated today. And the sheer volume of data being produced and disseminated across multiple systems throughout each trial is significantly greater than ever before – a 2025 Tufts study found that phase 3 trials each average nearly six million data points, and the number of data points per trial has roughly tripled over the last decade!
CRAs are responsible for monitoring such mega-complex trials, supporting sites in the operational conduct of the protocol while ensuring data integrity and patient safety at every step. They must accurately monitor data from dozens of separate systems to ensure they can build an accurate picture of study status and identify areas of need plus the next best action to take. Instead of focusing on supporting sites and adding strategic value, CRAs are chasing missing data and are reactively tactical.
More than a chatbot: closing productivity gaps with agentic AI
Highly capable and experienced CRAs are in a constant battle against time, which causes ever-increasing volumes of white space that slow clinical trials. Moreover, it highlights the futility of increasing manual administrative processes in clinical research. Human CRAs will never be able to keep pace with expanding complexity and quantity of manual tasks without an automated solution. Worse still, CRAs are unable to proactively address site issues due to the murky picture of site status spread across multiple clinical systems.
This is where agentic AI has the potential to be transformative.
Unlike purely predictive generative large language models (LLMs) and chatbots, AI agents can become virtual teammates, working semi-autonomously and systematically toward a set of specific goals rather than merely responding to isolated inquiries. Additionally, the AI agents being developed for CRA support today are not trained on troves of generic, disconnected data but on deep, vertical-specific expertise and grounded in regulatory guidelines, SOPs, and study plans, ensuring their actions are reliable, ethical, and informed by the nuanced regulatory and scientific realities of clinical research.
At the most basic level, AI agents can be used to automate everything from routine administrative tasks to complete clinical trial workflows, which on their own is enough to generate significant time savings and free up CRAs to focus on more important, human-centric responsibilities. But the real benefit is agents’ ability to act as a strategic coordinator with an unlimited capacity to learn and improve their performance based on past experiences.
The traditional, linear approach to site activation is a notoriously long, step-by-step process that tends to yield considerable delays and white space. It can often take months to complete tasks and the collect and process data from all systems. Teams gain little constructive knowledge about the process – except that too many independent variables exist to conceive of a more effective strategy.
By contrast, an AI agent can tackle most of the daily administrative burden. Armed with standard knowledge of what is needed for each country, site, and Institutional Review Board (IRB)/Ethics Committee (EC), AI agents automate the collection, review, and compilation of document packages. CRAs are freed up to focus on strategic activities such as helping sites operationalize the protocol so that less time is wasted when stakeholders are working at different paces. The agent can then analyze all relevant operational data to identify patterns and potential areas for improvement, ultimately helping to tailor future strategies based on the most effective approach for different site types, regions, therapies, and levels of protocol complexity.
Perhaps the most attractive benefit of agentic AI in clinical research – at least from the perspective of an overworked and overwhelmed CRA – is the ability to automate and streamline site monitoring and data management processes. By consolidating complex trial data across dozens of systems, monitoring hundreds of trial variables simultaneously, and suggesting smart real-time next-best actions, agents give CRAs the tactical support they desperately need while enabling proactive intervention before small issues become major problems, allowing CRAs to return to their core role of supporting sites.
For a team of 1,000 CRAs, Medable forecasts savings of eight hours per CRA per week, totaling 384,000 hours saved per year (based on company-sourced primary data). This translates into an estimated $40 million in annual productivity based on an average $200,000 salary (48-week year) per CRA. (See Figure 1.)
Balanced with appropriate checks and humans-in-the-loop, AI agents optimize CRAs’ time so they can focus on the most important activities like monitoring patient safety and working with sites.
A crucial partner, not a replacement
While the development and ongoing refinement of agentic AI has massive implications for the efficiency and quality of clinical trials, the primary aim of this technology is to support and not to replace human CRA oversight and expertise. While the level of autonomy given to AI agents is very likely to increase over time, they will act as a complimentary digital co-pilot, relieving highly skilled CRAs of more tedious work so they can focus their time, where the human touch is crucial to advancing the trial, supporting sites, and caring for patients.
This clinical trial transformation has never been more urgent or more vital, especially considering the broader use of AI further upstream in drug development. As AI-driven drug discovery identifies more potential drug candidates faster with ever-increasing chances of success – thanks to continuously improving computational biology – the creaking and overburdened clinical trial funnel will become the single greatest obstacle to treatments reaching patients. The boundaries of science will melt away, and only process mechanics will block human healing. As Novartis’ Vice President/Head of Digital Innovation and AI in Drug Development Max Lawson put it at the "Novartis AI and Digital Day": “Accelerating new medicines is arguably the most important way AI will benefit humanity.”
With AI agents as partners across the clinical trial lifecycle, the industry can match the speed of AI-powered drug discovery in drug development. Fully leveraging AI across the entire development paradigm will result in the most monumental wave of new human medicine ever seen.
About the Author
Andrew Mackinnon oversees the Medable Customer Value Team. He has more than 20 years of experience in managing clinical trials at large pharmaceutical, biotech, and CRO companies, most recently as a senior executive at Covance. He can be reached at andrew.mackinnon@medable.com.
Newsletters
Receive the latest pharmaceutical news, personalities, education, and career development – weekly to your inbox.
