For decades, the TMF has been the memory of a clinical trial. Now AI is turning it into the system that does the work.
It’s a well-known secret in the clinical development world: even though all sponsors call their Trial Master File critical, no one expects it to actually do anything. And for good reason. Just consider how a single document moves through a trial today. It gets created, uploaded, reviewed, and then filed along with all other TMFs. Eventually someone notices a piece is missing, sends an email, and follows up until it arrives. No value was actually added in this cumbersome and lengthy process.
It’s a process that originated when software existed to capture information. But in the age of AI Teammates, it no longer makes any sense to keep asking people to do the chasing while the systems are fully capable of doing so.
And that’s the reality of clinical trials. Almost every system in clinical research was built to remember work rather than to do it. CTMS records study progress, TMF holds essential documents, EDC stores patient data, and so on. Each one is a store of a certain kind of information, but none of them executes the work that produces it because it’s being done by people as the execution layer.
Sponsors and CROs hire an army of specialists for various activities: coordinators chase documents, reviewers open files, associates send reminders, and managers escalate issues that slip through. The software to date has not been designed to make any of it happen, just to record it.
Another open secret is that established eTMF vendors have every reason to keep it that way. That’s because in the SaaS world of technology, their business depends on seats and services. More complex workflows actually generate more revenue for them. And within that model, an intelligent system that removes the manual work altogether threatens the revenue it depends on.
This is why incumbents often bolt AI on as a feature rather than rethinking the system from scratch. They may have gone from repository to intelligent repository, but at the end of the day, it’s still just a place where documents sit. Making the repository actually do the work threatens their cash flow.
But with AI Teammates, a third step is introduced. Instead of quietly confirming that a document was filed, the TMF begins asking why one has not arrived, which site is most likely to miss a deadline, whether a version is inconsistent with the copy on file, and whether an updated one should be requested now. It’s more than just record keeping. It is operational execution work TMF teams perform by hand right now.
In practice, the AI Teammate becomes responsible for ensuring that a document is checked the moment it arrives rather than during a periodic review cycle. When something is wrong, whether a missing signature, a date that does not match, or a conflicting version, the system groups the related issues into a single unit of work, identifies the right contact at the site, drafts the message, and follows up until the problem is resolved. When it cannot resolve something, it hands the task back to a person and flags it for attention.
The impact on quality is significant. Using this approach moves QC finding rates from the 40-60% typical of manual review down to under 3%, because the same standard is applied to every document the moment it lands rather than in a review cycle weeks later.
This comes partly from the fact that AI Teammates, unlike humans, run constantly, catching a problem before anyone has thought to even look for it. The people involved transition their roles from doing the repetitive execution to supervising it. They intervene when human judgment is required.
The truth is that AI Teammates go beyond simply improving document management to changing what the TMF actually is: an operational engine supported by skilled humans to supervise it rather than work it. In this new model, the work runs itself.
This shift, from systems that remember work to systems that actually run the work, is a fundamentally different operating model for clinical trials. It was also one of the central themes in a recent SCOPE X presentation by Perceive Biotherapeutics VP of Clinical Operations Paulius Ojeras. In it, he shares how AI Teammates are becoming trusted execution partners capable of planning, coordinating, and carrying out operational work. It’s a session well worth watching:
Register Now
