The SWAR framework is a methodologically sound response to a genuine gap in the field and treating AI tool evaluation as a platform trial rather than a one-time benchmark exercise is exactly the right conceptual move.

One aspect that could be further developed in the full protocol concerns the operational rules governing interim analyses. Specifically, it would be valuable to clarify whether tool removal will follow pre-specified stopping criteria (analogous to O'Brien-Fleming bounds in adaptive clinical trials) or whether decisions will rest primarily with the Adjudication and Data Monitoring Committee. In platform trials, this distinction is foundational: pre-specified rules protect against decision bias and enhance reproducibility, while discretionary adjudication offers flexibility but may introduce subjectivity that is difficult to audit retrospectively.

Making this explicit — perhaps through a decision framework or stopping rule table within the protocol — would not only strengthen methodological transparency but also make the CESAR blueprint easier to replicate by other review teams seeking to adopt this approach.

Dear authors, this is a minor but meaningful addition that could significantly increase the protocol's utility as a field-wide standard.

Congratulations to the entire CESAR team for a genuinely innovative contribution to evidence synthesis methodology. The field needed this. 👏

Vanessa Bertolucci