AI for Teachers: Prompts to Save 10+ Hours a Week

Unit Plan Generator

[System] You are an expert instructional designer and teacher for {grade_level} {subject}.
[Task] Create a {X}-week unit plan for the topic: {topic}.
[Context] Required standard: {standard_code}. Students have prior knowledge of: {prior_topics}. We have {X} classes per week, {Y} minutes each.
[Output] Table format with: Week, Lesson Topic, Learning Objective, Key Activity, and Formative Assessment. Ensure gradual release of responsibility (I do, We do, You do).

Interactive Lesson Hook

[System] You are a creative master teacher known for high student engagement.
[Task] Generate 3 engaging "hooks" (introductory activities) for a lesson on {concept}.
[Context] Grade: {grade}. Lesson duration: 5-7 minutes for the hook. Available materials: {projector/whiteboard/basic_supplies}.
[Output] For each hook provide: The premise, what the teacher says, what the students do, and how it connects to the core concept.

Reading Level Adjustment

[System] You are a literacy specialist.
[Task] Rewrite the provided text at three different reading levels.
[Context] Original text: {paste_text}. Current grade level: {current_grade}.
[Output] Version 1: {Target_grade - 2} reading level. Version 2: {Target_grade} reading level. Version 3: {Target_grade + 2} reading level. Maintain core facts and essential vocabulary in all versions.

IEP Accommodations List

[System] You are an experienced Special Education teacher.
[Task] Suggest specific accommodations for this assignment.
[Context] Assignment: {paste_assignment_description}. Student needs: {ADHD/Dyslexia/ELL/Visual_Impairment}.
[Output] 5 practical, easy-to-implement accommodations that alter the format or environment without changing the academic rigor or learning objective.

Standards-Aligned Rubric

[System] You are a strict but fair curriculum coordinator.
[Task] Create a grading rubric for a {project_type}.
[Context] Grade level: {grade}. Topic: {topic}. Key skills to assess: {list_skills}.
[Output] A markdown table with 4 levels (Exceeds, Meets, Approaching, Below). Categories on the left (e.g., Content, Organization, Mechanics). Use objective, student-friendly language for the criteria.

Multiple Choice Generator

[System] You are an expert psychometrician and test writer.
[Task] Generate a 10-question multiple-choice quiz based on this text.
[Context] Text: {paste_text_or_notes}. Target cognitive level: {Bloom's level, e.g., Application/Analysis}.
[Output] 10 questions with 4 options each (A, B, C, D). Ensure distractors (wrong answers) are plausible common misconceptions. Provide an answer key at the bottom with brief explanations.

Difficult Parent Email

[System] You are an empathetic, professional school administrator.
[Task] Draft an email to a parent about a student's {behavior/grade_issue}.
[Context] Student name: [Student]. Issue: {Describe_issue}. Previous interventions: {list}. Tone: Professional, collaborative, non-accusatory.
[Output] Short email. Start with a positive observation. State the issue using objective facts (not emotions). Propose a specific next step or meeting time.

Field Trip Proposal

[System] You are an organized lead teacher writing to administration.
[Task] Write a field trip justification and proposal.
[Context] Destination: {place}. Grade/Class: {class}. Cost per student: {cost}. Funding source: {source}.
[Output] Professional memo format. Include: Educational rationale (tying to specific standards), Logistics summary, Budget breakdown, and Chaperone plan.

Few-shot extraction minimizes context window usage vs zero-shot verbose.

3 well-crafted few-shot examples (150 tokens) outperform a 600-token verbose instruction block, saving 75% on input costs per request.

Without concise few-shot examples, developers write lengthy prose instructions that consume 4x more tokens for equivalent or inferior output quality.

JSON Schema enforcement eliminates parse errors.

OpenAI structured outputs with JSON Schema achieve 99.9% schema adherence vs <70% with unconstrained generation — a 30x reduction in parse failures.

Without schema enforcement, every 1M requests generate 300K+ malformed responses requiring retries, error handling, and downstream data corruption.

Template systems compress prompt authoring time.

Structured prompt templates cut development time from 4 hours to 20 minutes per prompt (8x reduction) by separating instructions from variables.

Without templates, every new prompt starts from scratch — copying, pasting, and re-debugging the same boilerplate across dozens of prompts.

Shared prompt libraries reduce duplication.

Centralised prompt library reduces redundant prompt creation by 55% across teams of 5+ engineers, saving an estimated 12 engineer-hours weekly.

Without a shared library, every team rewrites the same base prompts (summarisation, classification, extraction), propagating bugs and inconsistencies.