Grounded chat & teacher prompting
This is the heart of the day: moving from vague requests to structured teaching instructions — and treating prompting as a conversation of instruction, review and refinement, not a search for one magical sentence.
The CLASS structure — build every serious prompt with it
Who are the students?
What should students understand or do?
What should NotebookLM produce?
Length, difficulty, examples, format, exclusions.
What must be grounded in sources, verified or avoided?
👀 Watch the demo — weak to strong in four steps
Which class? What purpose? How long? Which examples? How will we know students understood? None of it is stated.
The class is set — but the objective, examples, length and checks are still missing.
Grounded in sources, familiar examples named, difficulty controlled, and it ends with a check for understanding.
You can ask NotebookLM to critique its own output — then verify the result yourself.
"Prompting is not about finding one magical sentence. It is a conversation of instruction, review and refinement."
🖐 Now you do it — the prompt ladder
Using your own notebook and topic, write four versions:
- A vague prompt — the kind we all write on day one.
- Add class and subject.
- Add learning objective and artifact.
- Add constraints and a verification requirement — then generate this version and identify one weakness in the output.
Record for your evidence: initial prompt, improved prompt, what changed, why it improved, and one further revision made after seeing the output.
Templates to keep
Eight ready-made teaching templates — concept explanation, lesson plan, worksheet, misconceptions, revision, differentiation, story-based learning and assessment — are in the Prompt Library with one-tap copy. Replace the [highlighted] parts with your class, topic and constraints.
✋ Checkpoint — show your trainer
Checkpoint complete — you can drive the conversation!