Protocol Feasibility

Stress-test a plan before it hardens

Protocol feasibility is a queue-backed AI workflow used as guided decision support inside trial planning. It helps teams challenge assumptions and generate a fresh viewpoint without silently rewriting the underlying trial record.

Tie it to the trial

flowchart TD
    Trial[Trial and module] --> Run[Feasibility run]
    Run --> Output[Generated output]
    Output --> Review[Human review]

Each run is tied to a specific trial and workflow module. Generation, run history, and output retrieval are separate parts of the same process, which makes it possible to revisit older runs as the planning context changes.

Let the template choose the useful visual

The active feasibility template owns the question set, scoring guidance, output structure, and any visual explanation requested for an individual question. Most answers remain concise prose and tables. Selected questions can request one evidence-backed Mermaid diagram when a flow, sequence, or state transition makes the operational risk easier to understand.

Diagrams are not generic decoration. TrialStack follows the question-specific diagram type and node limit, includes only relationships supported by the available trial evidence, and adds a short text reading of the same relationship. If the evidence cannot support a trustworthy diagram, the answer omits it and names the missing evidence instead of showing a placeholder chart. This lets scientific, safety, operational, financial, and site-capacity reviews use different visuals without changing the review UI.

Use it when planning needs a second lens

Run protocol feasibility when:

• the trial context has changed materially
• a specific planning module needs a fresh challenge or comparison
• the team wants structured decision support before committing to a direction

Treat output as advice

Feasibility output is useful when it sharpens the conversation, not when it replaces it. Queue availability, permissions, and run timing are part of the experience, and completed output still needs human interpretation before it affects live planning.