Starting with a small brief
The user provides a course name, intended duration and level. The system searches and proposes course-level learning objectives. A human reviews and approves these objectives before generation proceeds.
Building the content hierarchy
Each approved course-level objective becomes a lesson. The system recommends four to six learning objectives per lesson; once approved, these are expanded into topics. Full written content is generated from the approved hierarchy.
From text to deliverables
Generated material is divided between video, practical lab and retained text components, then assembled into a single course file. The same staged approach has been used for article writing. The owner describes local-model operation for privacy; this does not establish that any external research/search step is offline.
Engineering boundaries
The exact model stack, search provider, video generation method, lab runtime, export formats and supported LMS/CMS integrations still need source evidence. An associated course-factory repository describes validation, provenance and release gates but has not been proven to be the same implementation. This project is AI software, not a training-service offering.