AI Prompts for Teachers: 10 Templates for Lesson Planning & Feedback (2026)

Generic prompts get generic worksheets. The fix is to load four specifics: the exact grade or level, the learning objective or standard, the time you actually have (a 45-minute period is not a unit), and the classroom reality (class size, range of readiness, available materials). Add an output shape — a table, a numbered plan, a rubric grid — and the model produces something you can drop into your planner.

Two guardrails matter for accuracy. First, ask the model to mark anything it's unsure of rather than inventing a fact, date, or statistic — it will fabricate plausible-sounding 'facts' for content if you let it, so verify any subject-matter claim before teaching it. Second, keep student data out: describe students by role or need ('a student reading two grades below level'), never by name or record.

These prompts run well on any current model. For high-volume routine work (quiz questions, vocabulary lists, email drafts), an efficiency-tier model like Gemini 3.1 Flash-Lite or gpt-5.4-mini is plenty; for nuanced differentiation and feedback, a frontier model reasons better. Prices as of June 2026 (OpenAI, Gemini).

When to use: you have an objective and a time slot and need a structured plan you can teach from.

``` Write a lesson plan for [GRADE/SUBJECT]. Objective / standard: [PASTE the exact standard or learning objective] Time available: [e.g. one 50-minute period] Class context: [size, readiness range, any constraints/materials] Structure: objective in student-friendly language, hook (5 min), I do / we do / you do progression with timing, a quick formative check, and an exit ticket. List materials needed. Keep activities concrete and doable in the time given. If the objective is too big for the time, say so and suggest where to split it. ```

Why it works: giving the model the real standard and the real clock keeps it from designing a 90-minute lesson for a 50-minute period. The 'too big for the time, say so' instruction is what makes it honest about pacing instead of cramming.

When to use: you have one lesson and need to reach students across a range of readiness.

``` For the lesson/objective below, give me three versions of the core task: - Below grade level (scaffolded, with supports listed) - On grade level - Above grade level (extension that deepens, not just "more work") Also suggest one accommodation each for: a student who needs language support (multilingual learner) and a student who needs the task chunked into smaller steps. Keep the learning objective the same across all versions — only change the access and challenge. Describe students by need, not by name. Objective / task: [PASTE] ```

Why it works: 'same objective, change access and challenge' is the core principle of good differentiation, and stating it stops the model from just shortening the assignment. 'Deepen, not more work' fixes the most common bad extension.

When to use: you need a clear, fair rubric students can actually understand.

``` Write a rubric for this assignment: [DESCRIBE / PASTE prompt]. Grade level: [LEVEL]. Use [3 or 4] performance levels. For each criterion: a clear name, and a descriptor at each level written in specific, observable language (what the work shows, not vague words like "good" or "excellent"). Limit to the [4-6] criteria that matter most. Add a short student-facing version: "To earn top marks, make sure you..." in plain language. ```

Why it works: 'specific, observable language, not vague words' is what separates a rubric students can self-assess against from one that just relabels A/B/C. The student-facing version turns the rubric into a checklist they can actually use while working.

When to use: communicating with families — a check-in, a concern, or good news — clearly and warmly.

``` Draft a short email to a parent/guardian about [SITUATION, described generally — no student name or record]. Tone: warm, specific, collaborative (partner, not adversary). Structure: open with something genuine, state the situation factually, what we're doing about it at school, what would help at home, and an invitation to connect. Under 180 words. No jargon or acronyms. For a concern, frame it as 'here's what I'm seeing and how we solve it together,' never as blame. ```

Why it works: the 'partner, not adversary' framing and the 'no blame' rule are what keep a concern email from triggering defensiveness. Banning jargon and acronyms makes it accessible to every family. Adapt it with the Business Email Generator.

When to use: you need questions at the right cognitive level, fast, with an answer key you can trust.

``` Write [N] questions on [TOPIC] for [GRADE]. Mix the levels: [X] recall, [Y] application/analysis. Format: [multiple choice / short answer / mix]. For multiple choice, make distractors plausible (based on common misconceptions), not obviously wrong. Provide an answer key with a one-line explanation of why each answer is correct. Mark any question where the correct answer depends on a fact I should verify, rather than presenting it as certain. ```

Why it works: 'distractors based on common misconceptions' is what makes multiple-choice questions diagnostic instead of guessable. The 'verify' flag is essential — models confidently produce wrong answer keys on factual content, so cross-check before printing.

When to use: writing feedback that helps a student improve — without pasting any identifying information.

``` Help me write feedback on this piece of student work (no name, no identifying details). Grade level: [LEVEL]. Assignment goal: [GOAL]. Structure: 2 specific strengths (quote what the student did, not generic praise), the single most important thing to improve next with a concrete next step, and one guiding question to push their thinking. Tone: encouraging and honest. Address the work, not the student. Keep it actionable — one main thing to work on, not ten. Student work: [PASTE — REDACT name and any identifiers] ```

Why it works: 'one main thing to improve, not ten' respects how much a student can actually act on, and quoting specific strengths makes the praise credible. 'Address the work, not the student' keeps feedback growth-oriented.

When to use: planning a multi-week unit and you want a coherent arc, not a pile of lessons.

``` Outline a [N]-week unit on [TOPIC] for [GRADE]. Standards/objectives: [PASTE] Class time: [periods per week, length] Produce: the essential question, a week-by-week sequence (what's taught, how it builds on the prior week), where formative checks go, and the summative assessment. Identify the 2-3 likely sticking points and where to reteach. Keep it realistic for the time given. Flag anything that won't fit so I can cut deliberately. ```

Why it works: anchoring on an essential question gives the unit a spine so it builds instead of sprawling. Naming likely sticking points up front lets you plan reteaching before you're behind, and the 'won't fit, flag it' rule keeps the plan grounded in real minutes.

When to use: a text is too hard for some readers and you want a version that keeps the content but lowers the barrier.

``` Rewrite the passage below for a [TARGET reading level / grade]. Keep the key ideas and vocabulary that matter for the content; simplify sentence structure and replace only the words that block comprehension (and gloss the important terms instead of dropping them). Then list the 5 academic vocabulary words worth pre-teaching, each with a student-friendly definition. Do not change facts or oversimplify to the point of being wrong. Passage: [PASTE] ```

Why it works: 'gloss important terms instead of dropping them' preserves the academic vocabulary students need rather than dumbing the content down. The 'don't oversimplify to the point of being wrong' guard keeps accuracy intact while lowering the reading load.

When to use: a concept is hard to make concrete and you want multiple ways in.

``` Give me ways to teach [CONCEPT] to [GRADE]: - 3 hooks/opening questions that create curiosity - 2 analogies relatable to this age group (and the limit of each analogy so I don't create a misconception) - 2 worked examples, from simplest to harder, with the thinking shown - the most common misconception students have, and how to confront it Keep examples age-appropriate and culturally neutral. ```

Why it works: asking for 'the limit of each analogy' is the move that prevents a helpful analogy from hardening into a misconception. Pairing worked examples with the common misconception lets you teach toward the error students actually make.

When to use: you're out and need a plan a substitute can run with no prep and no special knowledge.

``` Write a substitute lesson plan for [GRADE/SUBJECT], [period length], that a sub with no background in the subject can run independently. The activity must: need only [materials available], require no grading or specialized knowledge from the sub, keep students productively occupied, and connect loosely to [current topic] if possible. Include: minute-by-minute timing, exactly what the sub says/does, what students turn in, and a simple backup if they finish early. ```

Why it works: writing for 'a sub with no background' forces the plan to be fully self-contained — the failure mode of sub plans is assuming knowledge the sub doesn't have. The early-finish backup prevents the chaos of a class with ten minutes and nothing to do.

For routine, high-volume work — quiz banks, vocabulary lists, email drafts, sub plans — an efficiency-tier model is cost-effective and fast: gpt-5.4-mini ($0.75 in / $4.50 out per 1M) or Gemini 3.1 Flash-Lite ($0.25 / $1.50). For nuanced differentiation, feedback, and unit planning, a frontier model — Claude Opus 4.8 ($5 / $25), gpt-5.5 ($5 / $30), or Gemini 3.1 Pro (~$2.00 / $12.00) — reasons more carefully about pedagogy. Prices as of June 2026; check the live rate cards (OpenAI, Anthropic, Gemini).

Always verify subject-matter facts, dates, and answer keys before teaching them — models produce confident, wrong content. And keep student data out of general AI tools: describe by role and need, never by name or record, and follow your district's privacy policy.

Sources and further reading: DAIR.ai Prompt Engineering Guide, Learn Prompting, Gemini prompting strategies, Claude prompt engineering overview. Pricing current as of June 2026.