Why I Built This
Writing a teaching syllabus in the Outcome-Based Education (OBE) format is repetitive: learning outcomes, mapping to programme outcomes, assessment methods, week-by-week plans. Lecturers at Dai Nam University spend hours on it every semester. OBE-Agent is an AI agent that does the structuring so the lecturer can focus on the content.
How It Works
The lecturer describes the course; the agent drafts the syllabus sections in the OBE structure, asks for missing details, and lets the lecturer edit the result. I built the application end to end — the backend that orchestrates the agent and the web interface lecturers use.
Key Decisions
- Agent, not a single prompt. A syllabus has dependencies between sections (outcomes drive assessments), so the agent works step by step and checks consistency instead of generating everything at once.
- Editable output. The agent produces a first draft, never the final word; every section can be changed by hand.
- Built to generalise. The same orchestration is being extended into an assistant for businesses.
What I Learned
Building around an LLM is mostly product work: deciding what the agent asks, when it stops, and how a user corrects it. The product is working and still under active development.