AI can be a powerful tool for casual relief teachers who often don’t know what the day will bring. Platforms like ChatGPT can help you get across a concept before you have to teach it, turn one worksheet into three levels, build a starter for a class you have never met, draft the handover note and more. Here are 7 prompts that are ready to use to get you started.Â
These prompts and learnings have been pulled from Matt Esterman’s course, AI for Teachers, now available free on ClassCover Learn.Â
Why a cover day is a different problem
Almost every article about AI in teaching assumes you wrote the unit, know the class, and have the week to think about it. A cover day gives you none of that. You get the room, the roll, and whatever was left on the desk, and you have the time between the car park and the bell to work out how the next six hours run.
That is the gap worth filling. Casual work is a structural part of the profession, not an edge case: in 2024 there were 567,103 registered teachers in Australia and 16% of the workforce was employed on a casual basis, according to AITSL’s June 2026 workforce report.
One thing to settle before the prompts. You stay the teacher in the room. The Australian Framework for Generative Artificial Intelligence (AI) in Schools, approved by Australian education ministers in November 2023 and re-endorsed after review in June 2025, puts it plainly in its guiding statement on teacher expertise: generative AI tools should be used “in ways that support teacher expertise, and teachers are recognised and respected as the subject matter experts within the classroom” (Framework Resource Poster). AI drafts, you decide.
Seven ways to use AI on a cover day
Each of these takes a couple of minutes on your phone or a school computer. Copy the prompt, replace the bracketed parts, and read what comes back before you use it.
1. Get across a concept you are about to teach
The hardest moment in relief teaching is being asked to explain something outside your subject area, in front of thirty people, with no notice. You have two options: find a definition from a reliable source, and get a second explanation you can put in your own words.
This prompt comes straight from Matt Esterman’s session, and it works because it makes you do the thinking:
You are an expert learning coach who can assist learners through a Socratic teaching method. I need to learn about [topic]. Please teach me about this by asking me questions and giving feedback on my answers. Ask questions one at a time.
Pick something genuinely difficult. Answering four questions about a concept lands it better than reading four paragraphs about it, and you finish the exchange able to explain it out loud.
2. Turn one worksheet into three levels
You have been handed a single worksheet for a class with a range you cannot see yet.
Here is a worksheet I have been asked to run with a Year [X] [subject] class. Give me three versions: one for students who will find this difficult, one at the level written, and one for students who will finish early. Keep the same learning focus and the same format. Use Australian spelling.
Paste the questions in, or photograph the page and upload it. Print what you need, or write the harder version on the board as an option.
3. Build a five-minute starter for a class you have never met
The first five minutes decide how the rest of the lesson goes.
Give me five starter activities for a Year [X] [subject] class. Each one runs in five minutes, needs no printing and no devices, and works with a whiteboard and nothing else. It has to be runnable by a teacher who has not met the class. Number them and tell me exactly what to say first for each one.
Keep the good ones. Over a term you build a bank of starters that work across year levels, which is worth more than any single day’s plan.
4. Have extension work ready before you need it
Early finishers create most of the behaviour problems on a cover day.
Students in a Year [X] [subject] class are finishing [task] early. Give me three extension tasks that go deeper into the same content rather than adding more of the same. Each one needs to be explainable in two sentences and completable in 15 minutes with pen and paper.
5. Draft the handover note
The note you leave behind is how a school decides whether to book you again. It takes two minutes and most people skip it.
Write a short handover note from a relief teacher to the regular class teacher. Cover what the class got through, what we did not get to, and anything the teacher should know for tomorrow. Five lines, plain and professional. Here are my notes: [your rough notes]
6. Turn your handwritten notes into something you can send
You wrote notes through the day on the back of the roll. Photograph the page and ask the tool to type it out.
Current tools handle this. OpenAI’s documentation confirms that where “there is text in an image, the model can also understand the text” (OpenAI, Images and vision), and Google’s Gemini apps accept photos alongside documents and other files to “get answers, summaries, and insights about your content” (Google, Upload and analyse files in Gemini Apps).
Ask for it like this:
Type out the text in this image exactly as it is written. Then list anything you could not read clearly so I can check it.
That second sentence matters. OpenAI’s own documentation notes the model “may generate incorrect descriptions or captions in certain scenarios”, and flags reduced accuracy on small text and on rotated or upside-down text. Asking the tool to name what it could not read gives you a list to check rather than a clean-looking page with two invented words in it.
7. Write the reflection or feedback you owe the school
Some schools ask for a short reflection, a note on a student, or feedback on how a block went. Draft it, then rewrite it in your own words.
Draft a short professional reflection on a day of relief teaching for [Year X, subject]. Cover what worked, what I would change, and one thing I want to follow up. Keep it to 150 words and plain language. Here is what happened: [your notes]
How to make any prompt work
Four parts get you most of the way: a role for the tool, the task, the context it needs, and the format you want back. “Give me ten engaging lesson starters for Year 8 Science, as a numbered list” beats “lesson starters” every time.
Two habits matter more than the wording. Talk to it rather than firing one long request, because a short back-and-forth surfaces options you would not have thought of. And treat it as a capable intern rather than an expert: useful, fast, and in need of checking.
Here is a general-purpose version from the same session, for anything you have been asked to run and have never run before:
You are an expert in explaining work related tasks in clear language. Please explain the steps I might need to take to [insert task] and give me advice as to how to begin.
Three rules that keep you professional
You are in someone else’s school for the day. These three rules cover almost everything.
1. Ask what the school allows. Ask whether they have a licensed AI platform you should be using, and ask what their policy says about AI in student work. Schools without a specific AI policy will have an assessment policy or an acceptable use of IT policy that covers it. The national framework expects the expectation to be set in the task itself, asking that work designed for students “clearly outlines how generative AI tools should or should not be used”. If nobody has told you the class can use AI, assume the answer is no until you check.
2. Keep student information out of consumer tools. The framework’s guiding statement on protection of student inputs asks that “students, teachers and staff take appropriate care when entering information into generative AI tools which may compromise any individual’s data privacy”. Strip names and identifying detail before you paste anything in, and use the platform the school is licensed for where one exists.
3. Check the output before it reaches a student or a colleague. This is the rule Matt closes his session on, and the framework makes it a principle: teachers and school leaders “retain control of decision making and remain accountable for decisions that are supported by the use of generative AI tools”. Anything you send to a school, a parent or a class is yours. Read it first.
AI for Teachers – from Matt Esterman
Seven prompts get you through a day. Matt Esterman’s full session goes further: how prompting works, where AI fits into assessment, what students are already doing with these tools, and how to hold the professional line while you use them.
AI for Teachers runs 1.5 hours on ClassCover Learn, presented by Matthew Esterman, Director of Innovation and Partnerships, AI consultant and author. It maps to Standard 2.6 and includes a Q&A session. A ClassCover Learn subscription is $99 and gives you access to the full course library.
FAQs
Yes, with two checks. Ask the school whether they have a licensed AI platform you should use, and ask what their assessment or acceptable use policy says about AI in student work. Schools that have no specific AI policy will usually have an assessment policy or an IT acceptable use policy that covers it.
Australian education ministers approved the Australian Framework for Generative Artificial Intelligence (AI) in Schools in November 2023, and endorsed a review of it in June 2025. It sets six principles rather than a ban, and individual schools, sectors and state departments set their own rules on top of it. Check with the school you are working in.
The tasks with a short fuse: getting across a concept before you teach it, turning one worksheet into three difficulty levels, building a five-minute starter for a class you have not met, generating extension work for early finishers, and drafting the handover note at the end of the day.
Give it four things: a role, the task, the context, and the format you want back. Then keep talking to it. A short conversation gets you further than one long request.
Most current tools accept a photo and can read text inside an image. Accuracy drops with small, rotated or messy text, and OpenAI’s documentation notes the model “may generate incorrect descriptions or captions in certain scenarios”, so read the output against the original before you use it.
Take care. The national framework asks students, teachers and staff to “take appropriate care when entering information into generative AI tools which may compromise any individual’s data privacy”. Strip names and identifying detail, or use the platform the school is licensed for.
Structured PD does. AI for Teachers on ClassCover Learn runs 1.5 hours and maps to Standard 2.6, which covers using ICT to expand curriculum learning opportunities and integrating it into teaching programs (AITSL, Australian Professional Standards for Teachers). The course page notes that it does not carry NESA accreditation, because it was published after the August 2024 NESA changes, and that since those changes all PD activities count towards required PD hours.
