AI prompts for lesson planning that actually save time

Most disappointing experiences with AI in lesson planning come down to one thing: a vague prompt. Type "write me a lesson plan on customer service" and you get something generic that needs rewriting almost from scratch, which defeats the point. A better prompt takes thirty extra seconds to write and saves twenty minutes of editing.

The four-part prompt structure

A reliable structure for education prompts has four parts, and it is worth learning it properly rather than winging it each time.

Role

Tell the tool what kind of expert to act as. This shapes tone and vocabulary. Example: "Act as an experienced FE tutor teaching Level 2 Hospitality and Catering."

Context

Give the specific detail that makes the output usable rather than generic: group size, prior knowledge, time available, setting, any access needs in the group.

Task

State exactly what you want produced, not just the topic. "Write a 15-minute starter activity" is far more useful than "help with a lesson".

Constraints and format

Specify length, structure, reading level, and what to avoid. This is the part people skip most often, and it is usually the difference between a usable draft and one you throw away.

Worked example 1: a starter activity

Prompt: "Act as an FE tutor teaching a Level 2 Hospitality group of 14 learners, mixed ability, some with reading difficulties. Write three starter activity options, each taking 10 minutes, on the topic of food allergen awareness. Keep language at approximately Entry Level 3 reading level. Format as a numbered list with a one-line instruction for each."

Why it works: the role sets tone, the context flags the reading level need before it becomes a problem, the task is time-boxed and specific, and the format request means you get something you can paste straight into a lesson plan.

Worked example 2: differentiated resources

Prompt: "Act as a functional skills English tutor. I am teaching a session on formal letter writing to a group of adult learners at Entry Level 2. Write a core worksheet with a model letter and five gapped sentences, then a simplified version with fewer gaps and a supported version with sentence starters. Use plain English throughout, no idioms."

Worked example 3: assessment-ready scenarios

Prompt: "Act as an assessor writing a realistic workplace scenario for a Level 3 Health and Social Care learner. Write a 150-word scenario involving a medication error in a care home, suitable for use in a written assessment question. Do not include real names of care providers or medications, use generic placeholders instead."

Note the constraint about real names and medications, added specifically to avoid factual errors that would need fact-checking against actual drug names and dosages.

Worked example 4: scheme of work summary

Prompt: "Summarise the following unit specification into a one-page scheme of work with six sessions, each with a learning objective, a key activity and an assessment checkpoint. Keep the language of the original specification's assessment criteria unchanged, do not paraphrase the criteria themselves." Then paste the specification text. This constraint matters because paraphrased assessment criteria can drift from what an awarding body actually requires.

What still needs a human check every time

• Any factual claim, statistic, date, piece of legislation or named standard
• Assessment criteria wording, always checked against the actual specification
• Reading level claims, which AI tools estimate inconsistently
• Anything going in front of learners with additional needs, checked against what you know of the actual group, not the generic description you gave the tool

Iterating rather than starting over

If the first output is not right, do not start a new prompt from scratch. Refine: "make this shorter", "remove the group discussion activity and replace with an independent task", "lower the reading level further". This iterative approach usually gets to a usable result faster than rewriting the whole prompt.

Time actually saved, realistically

Staff who use this structure consistently report saving perhaps 20 to 40 percent of planning time on a typical session, mostly by cutting the blank-page stage. It does not remove the need to know your group, your specification and your subject, which is exactly why this is a companion piece to How teachers are using AI in the classroom: 12 real examples rather than a replacement for good subject knowledge.

Building this into staff CPD

Prompt writing is a skill worth teaching explicitly rather than assuming staff will pick it up by trial and error. A short staff CPD session on the four-part structure, with a shared bank of good example prompts for common tasks, pays for itself quickly in staff time. This kind of structured skills development sits well alongside our qualifications and CPD courses, which increasingly build digital and AI literacy into teaching qualifications rather than treating it as a separate add-on.

A closing caution

Good prompting makes AI output more useful, but it does not make it more accurate. Constraints reduce the chance of an obviously wrong answer, they do not eliminate it. Treat every output, however well-prompted, as a draft written by an enthusiastic but occasionally unreliable colleague, and you will get the time savings without the embarrassing mistakes.