Lesson plans, differentiation, resources and feedback — with the marking left to you. Below are 9 copy-ready prompts. Fill in the [BRACKETS], copy, and paste into ChatGPT, Claude, Gemini or any capable assistant.
Teachers were early adopters for an obvious reason: the preparation load is enormous and much of it is structured work that AI genuinely does well.
The 9 prompts
Plan a lesson that produces understanding
Design a session around what students will be able to do.
Help me plan a lesson. TOPIC: [WHAT YOU ARE TEACHING] STUDENTS: [AGE OR LEVEL, CLASS SIZE, WHAT THEY ALREADY KNOW] LESSON LENGTH: [MINUTES] WHAT THEY SHOULD BE ABLE TO DO AFTERWARDS: [THE OUTCOME] RESOURCES AVAILABLE: [WHAT YOU HAVE] CURRICULUM REQUIREMENTS: [IF ANY] WHAT USUALLY GOES WRONG with this topic: [IF YOU KNOW] Produce: 1. THE OUTCOME AS A CAPABILITY - rewritten so it is observable. 'Understand photosynthesis' cannot be checked; 'can explain why a plant in a dark cupboard dies, referring to the inputs and outputs of photosynthesis' can. 2. THE PREREQUISITE CHECK - what students must already know, and a quick way to check it in the first few minutes. Teaching over a gap is the most common cause of a lesson not landing, and the students who have the gap rarely say so. 3. THE MISCONCEPTIONS - what students commonly believe about this topic that is wrong. These are stable, confident and must be addressed directly; they do not disappear when the correct version is presented alongside them. Say how to surface and confront each. 4. THE LESSON STRUCTURE with timings: - The hook: a question, a demonstration, or a problem they cannot yet solve - Prior knowledge activation and check - The new content, in chunks with a check after each - Worked example: the thing done fully, with the reasoning spoken - Guided practice: they try it with support - Independent practice: they try it alone - The check: how you know it worked 5. THE CHUNKING - working memory is limited. Break the content into segments with processing time between them. Say how many new ideas is too many for this length and level. 6. THE CHECKS FOR UNDERSTANDING - throughout, not at the end. Specific techniques that reveal what every student thinks, not just the confident ones: mini-whiteboards, hinge questions with diagnostic wrong answers, and cold calling. Asking 'does everyone understand?' reveals nothing. 7. THE DIFFERENTIATION - what to do for students who finish early and for those who are stuck. Both need planning in advance. 8. THE RESOURCES - what you need, and what can be improvised from what I said is available. 9. THE CONTINGENCY - if it takes longer than planned, what to cut. If it goes faster, what to extend into. 10. THE TAKE-AWAY - what they leave with, and the homework if any, tied to the outcome.
What you get: A capability-stated outcome, prerequisite and misconception handling, a chunked structure with checks throughout and contingency plans.
Tip: Point 3 is what separates a lesson that works from one that seems to. Misconceptions survive being taught the right answer unless they are confronted directly.
Explain a hard concept to a specific audience
Find the explanation that works for these particular learners.
Help me explain this concept. THE CONCEPT: [WHAT IT IS] WHO I AM EXPLAINING IT TO: [AGE OR LEVEL, AND WHAT THEY ALREADY KNOW] WHY THEY NEED IT: [WHAT THEY WILL DO WITH IT] HOW I HAVE EXPLAINED IT BEFORE: [YOUR CURRENT EXPLANATION] WHERE THEY GET CONFUSED: [IF YOU KNOW] Produce: 1. WHAT MAKES THIS CONCEPT HARD - the specific difficulty. Concepts are hard for identifiable reasons: it is counterintuitive, it requires holding several things in mind at once, it uses a familiar word in an unfamiliar way, it conflicts with an existing belief, or it is abstract with no everyday referent. Name which applies, because the explanation follows from it. 2. THE PREREQUISITE CHAIN - what must be understood first. Explaining a concept to someone missing a prerequisite produces memorisation, not understanding. 3. THE CORE IDEA IN ONE SENTENCE - stripped of terminology, at the level of my stated audience. 4. THE CONCRETE FIRST - the specific example before the general principle. Learners build abstractions from instances, not the reverse. Give the example to start with. 5. THE ANALOGY, WITH ITS LIMITS - one analogy, and explicitly where it breaks. An analogy taught without its limits creates a misconception that is harder to correct than the original confusion, because it feels like understanding. 6. THE PROGRESSION - the order to build it: what to establish first, what to add second, and what to leave until the basics are secure. Most bad explanations present everything at once. 7. WHAT TO LEAVE OUT INITIALLY - the exceptions, the edge cases and the precise definition. These matter, and introducing them too early prevents the main idea forming. Say when to add them. 8. MY CURRENT EXPLANATION, ASSESSED - where it assumes too much, moves too fast, or uses a term the audience will not have. 9. THE CONFUSION POINT - I said where they get stuck. Address it specifically, and say whether it is a prerequisite gap, a terminology problem, or a genuine conceptual difficulty. 10. THE CHECK - the question that reveals whether they have understood or are repeating words. It should require applying the idea to a case not covered in the explanation.
What you get: The specific reason the concept is hard, a prerequisite chain, concrete-before-abstract progression, an analogy with limits and a diagnostic check.
Tip: Point 5's insistence on stating an analogy's limits matters. An unlimited analogy becomes the student's mental model, and the errors it produces surface months later.
Write questions that reveal what students think
Design assessment that diagnoses rather than just scores.
Help me write diagnostic questions. TOPIC: [WHAT YOU ARE ASSESSING] STUDENTS: [LEVEL] PURPOSE: [check understanding mid-lesson / end of topic / identify gaps before teaching] WHAT I NEED TO KNOW: [THE SPECIFIC UNDERSTANDING YOU ARE CHECKING] FORMAT: [multiple choice / short answer / written / verbal] Produce: 1. THE MISCONCEPTIONS to test for - the specific wrong models students hold about this topic. Diagnostic questions are built around these, not around the correct answer. 2. THE HINGE QUESTIONS - if multiple choice, the most valuable format for diagnosis. Each with: - The question - The correct answer - Distractors, each of which is the answer a student holding a specific misconception would choose - What choosing each distractor tells you This is the whole technique: a wrong answer should identify which misunderstanding the student has, not merely that they are wrong. Random plausible distractors waste the question. 3. THE ANSWER-REVEALING QUESTIONS - questions a student can only answer correctly if they genuinely understand, and cannot answer by pattern-matching or recall. Usually these involve: an unfamiliar context, a case where the obvious rule does not apply, or asking for the reason rather than the result. 4. WHAT TO AVOID - questions answerable by elimination, questions where the correct answer is the longest option, questions testing reading comprehension rather than the subject, and questions where a student who understands might reasonably choose a different answer. 5. THE 'EXPLAIN WHY' FOLLOW-UP - for each question, the follow-up that distinguishes a lucky guess from understanding. 6. THE PRE-TEACHING VERSION - if the purpose is identifying gaps before teaching, questions that reveal starting points without being discouraging. 7. HOW TO USE THE RESULTS - for each pattern of responses, what to do next. A question is only diagnostic if the answer changes your teaching. 8. THE WHOLE-CLASS TECHNIQUE - how to get an answer from every student rather than the confident few: mini-whiteboards, voting, or written responses collected. Hands up tells you about four students. 9. THE MISCONCEPTION THAT SURVIVES - the one students often keep even after correct teaching, and what it takes to shift it.
What you get: Hinge questions whose distractors each diagnose a specific misconception, answer-revealing questions and a response-to-action map.
Tip: Point 2 is the technique worth learning. A multiple-choice question whose wrong answers each identify a different misunderstanding is worth ten questions that just score.
Give feedback that improves the next piece of work
Mark work in a way that produces improvement.
Help me give feedback on this student work. THE WORK: """ [PASTE OR DESCRIBE] """ THE TASK THEY WERE SET: [WHAT WAS ASKED] STUDENT LEVEL: [STAGE] ASSESSMENT CRITERIA: [PASTE, if any] WHAT THEY STRUGGLE WITH GENERALLY: [IF YOU KNOW] HOW MUCH TIME I HAVE PER STUDENT: [REALISTICALLY] Produce: 1. THE ONE THING - if they change one thing in the next piece, what. Feedback with twenty comments produces no change; feedback with one clear priority produces some. Identify it. 2. WHAT IS WORKING - specific and quoted, naming the technique rather than offering encouragement. 'Your second paragraph gives evidence before the claim, which makes it convincing' teaches something; 'good work' does not. 3. THE FEEDBACK, WRITTEN - phrased so it can be acted on: - Specific to a place in the work, not general - About the work, never about the student - Naming the improvement, not just the problem. 'This paragraph does not support the argument' is a diagnosis; 'add a sentence linking this evidence back to your claim' is actionable. - Forward-looking: what to do next time, not only what went wrong this time 4. WHAT NOT TO COMMENT ON - everything that is not the priority. Marking every error buries the important feedback and discourages. Say what to leave alone this time. 5. THE UNDERLYING CAUSE - the errors in this work may share a cause. Several unrelated-looking mistakes often trace to one misunderstanding. Identify it if there is one, because addressing the cause is worth more than correcting the symptoms. 6. THE RESPONSE TASK - feedback that is read and not acted on changes nothing. Give the specific short task that makes them use the feedback: redraft this paragraph, redo this question, or apply this to a new example. This is the step that produces improvement. 7. THE TIME-EFFICIENT VERSION - given how long I have per student, what to do. Whole-class feedback on the common issue plus one personal line per student is often more effective than extensive individual comments, and takes a fraction of the time. 8. THE GRADE QUESTION - if a mark is given alongside comments, students read the mark and stop. Say whether to separate them, and how. 9. THE TONE - direct and specific, without softening that obscures the point, and without discouragement. Show the phrasing.
What you get: One priority improvement, specific praise naming techniques, actionable feedback, a response task and a time-efficient whole-class alternative.
Tip: Point 6 is what makes feedback work. Comments students read and file away change nothing; a five-minute task that forces them to use the feedback does.
Design practice that builds a skill
Create exercises that actually develop capability.
Design practice exercises. THE SKILL: [WHAT STUDENTS SHOULD BE ABLE TO DO] STUDENTS: [LEVEL AND WHAT THEY CAN DO NOW] HOW MUCH PRACTICE TIME: [AVAILABLE TIME] FORMAT: [in class / homework / self-paced] HOW THEY GET FEEDBACK: [teacher marks it / answers provided / peer / none] Produce: 1. THE SKILL BROKEN DOWN - the component sub-skills. Complex skills are built from parts, and practising the whole thing repeatedly is inefficient when one component is the bottleneck. Identify the components and which are likely weak. 2. THE PROGRESSION - exercises ordered so each builds on the last: - Isolated component practice, where one element is varied and everything else held constant - Combined practice of two components - Full-skill practice in a familiar context - Full-skill practice in an unfamiliar context, which is where transfer is tested 3. THE EXERCISES - written out, at each stage. 4. THE DIFFICULTY CALIBRATION - practice should be difficult enough to require effort and easy enough to succeed most of the time. Too easy produces no learning; too hard produces guessing and discouragement. Say what proportion of correct answers indicates the right level. 5. THE VARIATION - vary the surface features while keeping the underlying principle constant. Students who practise ten near-identical problems learn to recognise the format, not the principle, and fail when the wording changes. 6. THE INTERLEAVING - mix this skill with previously learned ones rather than practising it in a block. Blocked practice produces better performance during the session and worse retention afterwards; interleaved practice does the opposite. Say how to mix these exercises with earlier topics. 7. THE FEEDBACK MECHANISM - given what I said is available. Practice without feedback embeds errors. If no teacher marking is possible, what self-checking can be built in: worked answers, a checklist, or a way to verify the result independently. 8. THE COMMON ERRORS - what students will get wrong, and an exercise designed to surface each. 9. THE SPACING - when to revisit this skill after the initial practice. 10. THE TRANSFER CHECK - the final exercise, in a context they have not seen, which is the only real test of whether the skill was learned rather than the exercise format memorised.
What you get: The skill decomposed, a staged progression with varied surface features, difficulty calibration, interleaving guidance and a transfer check.
Tip: Point 6 is counterintuitive and well established. Blocked practice looks better in the lesson and interleaved practice produces more learning that lasts.
Handle a topic students find boring or hard to engage with
Teach something that does not sell itself.
Help me teach a topic students find dull. THE TOPIC: [WHAT IT IS] WHY IT MATTERS: [THE GENUINE REASON, not the curriculum requirement] STUDENTS: [LEVEL AND CONTEXT] WHAT I HAVE TRIED: [YOUR CURRENT APPROACH] WHY THEY DISENGAGE: [YOUR READ ON IT] CONSTRAINTS: [TIME, RESOURCES, CURRICULUM] Produce: 1. THE DIAGNOSIS - disengagement has causes, and they need different responses: - IT SEEMS POINTLESS: they cannot see why it matters - IT IS TOO HARD: disengagement is often avoidance of failure. This is the most common and the most misread. - IT IS TOO EASY: no challenge - IT IS ABSTRACT: no concrete anchor - IT IS PRESENTED AS CONTENT TO ABSORB rather than a problem to solve From what I described, which applies. 2. THE TOO-HARD CHECK - specifically. Students who appear bored are frequently lost, and looking bored is more socially acceptable than admitting confusion. Test for this before trying to make the topic more interesting; making a topic they cannot follow more fun does not help. 3. THE GENUINE HOOK - based on why this actually matters. Not a contrived link to popular culture, which students see through immediately. The real reasons a topic is interesting are usually: it answers a question someone actually had, it explains something they have noticed, it was once controversial, or it has a consequence they care about. Find the real one. 4. THE PROBLEM FRAMING - present it as a problem to solve rather than material to learn. Start with a question they cannot yet answer and want to. 5. THE CONCRETE ANCHOR - a specific case, object, or situation to build from. 6. THE ACTIVITY - something they do rather than receive. For this topic and my constraints, what is feasible. 7. WHAT NOT TO DO - the approaches that reliably fail: forced relevance to things students like, competition that engages the confident and disengages everyone else, gamification that becomes the point, and a video that fills time without producing thinking. 8. THE HONEST ACKNOWLEDGEMENT - some material is genuinely dry and necessary. Saying so directly, explaining why it is needed, and getting through it efficiently is more respectful and more effective than pretending it is exciting. Say whether that applies here. 9. THE MINIMUM VERSION - if my constraints do not allow an elaborate approach, the smallest change that would help.
What you get: A cause diagnosis including the too-hard check, a genuine hook rather than a contrived one, problem framing and an honest option for dry material.
Tip: Point 2 is the check to run first. Apparent boredom is very often concealed confusion, and the fix is teaching, not entertainment.
Build a scheme of work for a unit or course
Sequence a whole topic over several weeks.
Help me plan a unit of work. THE SUBJECT AND UNIT: [WHAT IT COVERS] STUDENTS: [LEVEL AND PRIOR KNOWLEDGE] DURATION: [NUMBER OF LESSONS AND THEIR LENGTH] WHAT THEY MUST BE ABLE TO DO BY THE END: [OUTCOMES] ASSESSMENT: [HOW AND WHEN THEY ARE ASSESSED] CURRICULUM REQUIREMENTS: [WHAT MUST BE COVERED] RESOURCES: [WHAT YOU HAVE] Produce: 1. THE END POINT - the capabilities students should have, stated observably. Plan backwards from these. 2. THE CONCEPT MAP - what depends on what. Some concepts cannot be taught before others; some are independent and can be sequenced freely. Identify the dependencies, because the sequence follows from them rather than from the textbook order. 3. THE SEQUENCE - lesson by lesson: the focus, the outcome, and what it builds on. Include lessons for consolidation and practice, not only for new content. A unit that introduces something new every lesson leaves no time for anything to stick. 4. THE HARD PARTS - the two or three concepts in this unit that students find most difficult. Allocate more time to these and less to the straightforward material, rather than dividing time evenly. 5. THE MISCONCEPTION MAP - the common misconceptions in this unit and which lesson addresses each. 6. THE RETRIEVAL SCHEDULE - when earlier material is revisited. Building spaced retrieval of week one's content into weeks three and five is what makes the unit stick, and it takes five minutes per lesson. 7. THE ASSESSMENT ALIGNMENT - against my stated assessment. Does the sequence prepare students for what they are actually assessed on? Flag any mismatch, and any assessed content the plan does not cover. 8. THE FORMATIVE CHECKPOINTS - where to check understanding before moving on, and what to do if a checkpoint shows the class is not ready. 9. THE SLACK - lessons deliberately unallocated. Units always overrun, and a plan with no slack means either rushing the end or cutting the consolidation. 10. THE RESOURCE LIST - what needs preparing, and when, so it is not being made the night before. 11. WHAT TO CUT - if the time is shorter than the content requires, what goes, decided now.
What you get: Backwards-planned outcomes, a dependency-based sequence, time weighted to the hard parts, a retrieval schedule and deliberate slack.
Tip: Point 6 costs almost nothing and changes retention substantially. Five minutes of retrieval on earlier content at the start of each lesson beats an hour of revision at the end of the unit.
Differentiate one lesson for mixed ability levels
Adapt a single lesson so every student is challenged at the right level.
Differentiate this lesson for a mixed-ability class. SUBJECT AND AGE / GRADE: [DETAILS] LESSON OBJECTIVE: [WHAT STUDENTS SHOULD KNOW OR DO BY THE END] CURRENT LESSON PLAN OR ACTIVITY: [PASTE] CLASS PROFILE: [e.g. 5 students need support with reading, 4 finish early, 2 English language learners] TIME AND RESOURCES: [MINUTES, MATERIALS] Produce: 1. THREE PATHWAYS for the main activity: support, core and stretch - same objective, different routes 2. SCAFFOLDS for the support pathway (sentence starters, worked example, vocabulary list) 3. STRETCH TASKS that go deeper, not just more work 4. LANGUAGE SUPPORT for English learners without lowering the thinking 5. A QUICK CHECK at the end that works for all pathways 6. GROUPING advice Keep it practical for one teacher in one room. Do not label students in ways they would see as fixed ("low group").
What you get: Support, core and stretch pathways with scaffolds, stretch tasks, language support, an exit check and grouping advice.
Tip: Stretch should mean deeper, not more. Extra worksheets for fast finishers teach them that finishing early is a punishment.
Write a parent email about a concern
Raise an academic or behaviour concern with a parent clearly and constructively.
Write an email to a parent about a concern. STUDENT (first name only): [NAME] THE CONCERN: [WHAT HAS HAPPENED - specific, factual] EXAMPLES AND DATES: [EVIDENCE] WHAT I HAVE ALREADY TRIED: [ACTIONS] WHAT I WOULD LIKE FROM THE PARENT: [e.g. a meeting, support at home, information] SCHOOL POLICY POINTS TO MENTION (if any): [POLICY] Write the email so that it: - Opens with something genuinely positive about the student (ask me for one if I did not give it: [POSITIVE]) - States the concern factually, without labels or judgements - Gives 1-2 specific examples - Explains what I am doing to help - Makes one clear request and offers times to talk - Is under 200 words Also give me: a shorter version for a school messaging app, and 3 things to prepare before a follow-up meeting. Do not include other students' names or details.
What you get: A factual, constructive parent email under 200 words, a shorter app version and meeting preparation points.
Tip: Facts and examples, not labels. 'Handed in 2 of the last 5 homework tasks' gets a better response than 'is not trying'.
Where AI actually helps here
- Lesson plans with objectives, activities, timings and the questions to ask
- Differentiating one activity three ways for different levels
- Generating worked examples, practice sets and their answer keys
Where it falls down
- Matching your curriculum. It will produce something for a syllabus that is not yours
- Knowing your class. Pace and pitch are your judgement
- Grading real student work — accuracy aside, many institutions restrict it outright
The mistake almost everyone makes: Not giving it the specification
Paste the actual learning objectives from your specification, with the exam board and level, and say what the class did last lesson. Without that you get a generically competent plan for a course that does not exist — and adapting it costs more than writing from scratch.
Free tool: Prompt Builder
Runs in your browser. No sign-up, nothing uploaded.
Questions people ask
Can AI grade student work?
It applies a rubric consistently, which is not the same as grading well, and many institutions have policies against it. Use it to draft feedback comments from your own marks rather than to assign the marks.
How do I make AI lesson plans fit my curriculum?
Paste the specification points verbatim, name the exam board and year group, and say what was covered last lesson. It is the difference between a usable plan and a plausible one.