All projects

AI Systems

ScenarioLab

A clinical scenario practice platform for event health services officers — AI-generated scenarios, adversarially validated and human-approved before publication.

TypeScriptClaude APIAI PipelineGitHub Actions
ScenarioLab

Overview

ScenarioLab is a clinical scenario practice platform for event health services officers. Trainees work through realistic dispatch scenarios and reveal the correct clinical response; facilitators use printed cards to run live scenario sessions.

Safety-First AI Pipeline

All scenarios are AI-generated from official clinical practice guidelines, adversarially validated by a second AI pass, and require human clinical staff sign-off before publication. The published website contains no live AI — only pre-approved, static scenarios.

Engineering Highlights

  • Offline CLI pipeline: fetch clinical resources → generate scenarios → adversarial validation → human review
  • Polite resource fetching with ETag / If-Modified-Since caching and robots.txt crawl-delay compliance
  • Weekly GitHub Actions workflow that opens a pull request when source guidelines change
  • Per-organisation theming generated from a branding config
  • TypeScript monorepo with a static, zero-runtime-AI web frontend

Availability

The repository is private while the platform is piloted — a walkthrough or demo is available on request.