3 prompts every teacher should try this week
Matt Dunn ยท April 27, 2026
Most teachers have tried AI by now. You've typed something into ChatGPT or Claude, scanned the output, and thought: that's not usable. Too generic. Wrong year level. Missing the point entirely.
The problem isn't the tool. It's the prompt. A vague request gets a vague answer, and vague answers create more work, not less.
Here are three prompts you can copy, paste, and use today. Each one solves a different everyday teaching problem, and each one works regardless of what you teach or which AI tool you prefer. They come from our course, Practical AI for Busy Teachers, where we walk teachers through dozens of workflows like these.
The most common mistake teachers make with AI planning is asking for too much at once. "Plan me a week of maths lessons" sounds efficient, but the output is surface-level and takes longer to fix than it would to write from scratch.
The fix is simple: one learning intention, one lesson.
Prompt structureThe output will be specific enough to teach from with minimal editing. Compare that to what you'd get from "make me a lesson plan about the water cycle." The details you provide are what make the difference.
You know the feeling. It's Wednesday evening, you need an engaging activity for tomorrow, and you're stuck. AI can help here, but the trick is asking for options rather than a single answer.
When you ask for three activities, the AI has to generate genuinely different approaches. You then pick the best one, combine elements from two, or refine one further. It turns AI into a brainstorming partner, not a single-answer machine.
Prompt structureYou'll get three distinct options you can scan in a couple of minutes. If one is close but not quite right, tell the AI what to change. "Make option 2 work as a paired activity instead of students working independently" is all it takes.
Marking is where most teachers lose their evenings. AI can help, but there's an important boundary: never paste student names or student work into a public AI tool. That data ends up on external servers, and you can't control what happens to it.
The safer approach is to describe the assessment criteria and let AI generate a bank of comments you can select from and adapt. No student data leaves your computer.
Prompt structureYou'll get a bank of 27 comments you can reuse across a whole class. Select the one that fits each student, tweak the wording, and move on. It turns a two-hour marking session into a much shorter one, without compromising quality or privacy.
Why these prompts work
You might have noticed that all three prompts follow the same structure. That's not an accident. Each one uses a framework called RTFC: Role, Task, Format, Context.
It's a simple structure, and once you see it, you can apply it to almost anything: report comments, parent emails, unit plans, quiz questions, differentiated readings. This is one of the first things teachers learn in Practical AI for Busy Teachers, and it underpins every workflow in the course.
In a future article, we'll share a technique that takes your prompts even further, one that gets the AI to do the hard thinking for you so you get better results with less effort on your end. Keep an eye out for that one.
Try one this week
Pick whichever prompt matches your most pressing need right now. Copy it, swap in your own details, and paste it into ChatGPT, Claude, Copilot, Gemini, or whichever tool you have access to. Notice the difference between what you get back and what you've been getting from vague, one-line prompts.