Skip to content
Courses/AI Router and content system/Project: content system for Aulafy
Lesson

Project: content system for Aulafy

The final project combines routing, scoring, and human review to generate or improve educational content without losing editorial judgment.

Architecture

TerminalCode
sources -> brief -> local draft -> frontier review -> score -> human edit -> publish
              \-> shadow candidates -> comparison dataset

Minimum entities

  • source: URL, date, type, license, and summary.
  • brief: goal, audience, tone, level, and constraints.
  • generation: model, route, prompt, cost, latency, and output.
  • review: score, human changes, decision, and reason.

Editorial judgment

The router can speed up research and drafts, but it does not replace human decision-making. In Aulafy, publishing means checking sources, testing commands when possible, and declaring limits.

Final output

You finish with a small tool that receives a topic, proposes a model route, drafts content, requests review, and saves traces to improve the system with evidence.

Shadow mode and comparisonsBack to course