A clinical study setup has to match more than the protocol. It has to match every IRB-approved document behind it, from the consent form to the wording of each survey question. It also has to match the decisions made during configuration, like when a survey opens, when it expires, and which reminder a participant receives.
Our study configuration agent carries those requirements into the platform. Testing validates that they were configured correctly. A mismatch caught before launch means a fix. One caught after launch may end up as a protocol deviation, with participants getting a different experience than the one the IRB approved.
The standard approach to testing a clinical study configuration
Study teams test by hand. For each study, the project team works through dozens of test cases, comparing the setup against the approved documents side by side and walking through each participant journey to confirm it matches the intended workflow. When something changes, parts of that work must start over.
What does an AI testing agent do in clinical research?
At PicnicResearch, a testing agent checks every study configuration element before it goes to the sponsor. It reads the protocol and the source materials, drafts the test cases, and runs them on the draft study with test accounts, moving through the study the way a participant would. The rare test it can’t run on its own goes to the accountable person on our team. It saves screenshots as evidence for each test and, when testing is complete, produces a summary report.
What the testing agent checks
Our agent confirms two things: that the setup works as defined and that it matches what the IRB approved. Together, the tests cover the full spectrum of participant journeys: onboarding and consent, eligibility, surveys and symptom diaries, wearables, compensation, reminders, and withdrawal.
How failed tests get fixed and retested
The agent records the finding and logs a defect for our team with what it expected, what actually happened, and the evidence. Our team corrects configuration issues. For code defects, a background agent can take a first pass at the fix, and engineering reviews and approves each one before it goes in. The testing agent then reruns the test to confirm it passes.
How sponsors approve a study through user acceptance testing (UAT)
Once our team reviews the results and approves the setup, the sponsor runs their own user acceptance testing (UAT). Depending on the study, that means recorded walkthroughs, a live demo of key scenarios, or test accounts they use to go through the study themselves. Each route ends the same way: the sponsor gives explicit approval before the study can launch.
Example: catching a missing reminder before launch
Say a study’s IRB-approved materials tell participants they’ll get a reminder 2 days before each survey closes, but the configuration never schedules it. The testing agent moves through the study as a participant would, reaches the survey, and the reminder never arrives. It logs the defect with what it expected, what actually happened, and the evidence. Our team adds the missing reminder to the configuration, and the agent reruns the test to confirm it goes out on time. The setup doesn’t move forward until it passes.
What sponsors get: a validated study, with an audit trail
Sponsors get a study validated against its approved documents before they review it, and a UAT completion report built from the test results. The report documents what was tested, the results in each area, how findings were resolved, and the sponsor’s acceptance. It’s written for the Trial Master File as the record of system validation.
With the agent running the test cases, our study team spends less time clicking through test paths and more time on what needs judgment: reviewing findings, resolving edge cases, and working with the sponsor through UAT.
Testing completes the path that configuration starts: requirements traced to their sources, built into the configuration, and confirmed before launch.
The study the IRB approved is the study your participants get.
Talk with our team about your next study.