Prompting Basics
This lesson focuses on prompt discipline: making the task explicit, controlling the output shape, and using a revision loop that saves time instead of generating more rework.
Keep the wrapper, swap the srcdoc for the real provider embed URL, and the page structure remains unchanged.
Key takeaways
- Tell the model what role to take, what job to perform, and what output format to return.
- Use constraints to reduce cleanup work: length, tone, structure, and source boundaries all matter.
- Iterate with targeted corrections instead of rewriting the entire prompt every time.
Step-by-step workflow
- Define the output first: email draft, meeting summary, comparison table, checklist, or plan.
- State the audience and tone so the model has context for how formal or direct the response should be.
- Provide the source material and make clear whether the model can infer, summarise, or only restate.
- Ask for a structured answer that you can review quickly instead of a long freeform response.
- Refine with one correction at a time and capture the version that consistently works.
Companion notes
Reserve this space for a prompt worksheet, example prompts, or a review checklist once the assets are ready.
Recommended tools
- A prompt library or text-expander for common instruction blocks
- A shared team document for agreed output formats
- A review checklist to keep factual checks and tone checks consistent
Need access help?
If your login state looks wrong, verify the approved email first. The lesson route itself depends on Cloudflare Access, not a separate site account.
Support