Auto Course Generator
Upload a PDF, DOCX or PPTX → a document extractor pulls the text → an LLM pipeline structures it into a course: modules, lessons and formatted content. The populate endpoint can turn a single document into ready-to-edit lessons.
Lesson Builder — 8 editors
The interactive Lesson Builder edits each lesson type:
- rich_text (Tiptap), slides (drag-and-drop with rich elements), quiz, video, pdf, assignment, live_class, external_link.
Interactive subtypes (drag-drop, fill-blanks, timeline, branching, interactive video, flashcard sets) are authored too — see Interactive content.
AI Course Editor agent
An on-demand agent lets admins create and manage course content through natural chat, using tools (a tool_use loop). It routes through the multi-provider LLM Router like every other LLM-backed agent.
CourseReviewAgent — data-driven suggestions
The CourseReviewAgent analyzes the course against real performance data — aggregated ConceptMastery, common StudentMisconceptions, RAG coverage and quiz scores — and returns up to 8 prioritized suggestions (add_lesson, edit_lesson, add_exercise, reorder, add_reinforcement), cached 24h. It never invents suggestions without a basis in data.
RAG-first & source labeling
Content generation is RAG-first: the platform only generates with the LLM when the institution's RAG doesn't cover the topic, and always labels content as curated vs AI-generated.