One class, thirty learners: how AI can help you plan for them all

September brings new classes, new timetables and, for many teachers, a familiar challenge. You may know what you want to teach. You may even have taught the lesson several times before. But the pupils sitting in front of you this year are different.

One needs considerably more scaffolding. Another has English as an additional language. Several have gaps in prior knowledge. One pupil reads well below their chronological age. A handful understood the topic before they entered the room. And somewhere in the middle are twenty other children who also deserve to be challenged.

Differentiation isn't new. What is new is the possibility of having an extraordinarily capable assistant sitting alongside you while you plan. That assistant is AI. And one of the most useful things teachers can do with AI this September is not ask it to plan their lessons for them but ask it to help adapt good teaching to the learners actually in front of them.

Start with your lesson, not with AI

The worst starting point is probably to ask AI to Create me a Year 8 lesson on photosynthesis. You will certainly get something. It might even look impressive. But AI knows almost nothing about what you have previously taught, what your pupils already know, the misconceptions you have encountered, the curriculum you follow or what you want pupils to remember in six months' time.

It is essential to give your AI of choice that context. For example:

I am teaching Year 8 science in an English secondary school. Pupils have previously studied plant and animal cells and understand that plants require water and light. The key learning for this lesson is that pupils should be able to explain photosynthesis as a chemical reaction and describe the factors affecting its rate. Here is my existing lesson plan. Suggest improvements but retain the learning objectives and overall sequence.

That is a fundamentally different relationship with AI. You have designed the learning. AI is helping you improve it.

Then tell AI about your class

This is where things become particularly useful. Without providing names or identifiable personal information, describe the range of learners you are planning for. You might say:

The class contains 30 pupils. Six have weaker literacy and find extended written explanations difficult. Three are high prior attainers who need greater challenge. Two pupils have EAL and benefit from explicit vocabulary instruction. Several pupils have difficulty retaining multi-step instructions. How could I adapt this lesson while keeping the same ambitious learning objective for everyone?

You now have something approaching a planning conversation with another teacher. It might suggest vocabulary support, sentence stems, visual representations, chunked instructions, retrieval activities and deeper extension questions. You decide which are appropriate.

Don't create three lessons. Create different routes into the same lesson

AI makes it tempting to generate a ‘low’, ‘middle’ and ‘high’ version of everything. Resist that temptation. Instead, use AI to help remove barriers while maintaining ambition. Take a paragraph pupils need to understand and ask:

Identify the five words in this passage most likely to prevent a weaker reader understanding it. Give me pupil-friendly definitions and suggest how I could pre-teach them.

Or

Create three increasingly supportive scaffolds that would help a pupil answer this question successfully. Do not simplify the knowledge being assessed.

That last sentence is important. Support access to the thinking rather than removing the thinking.

Use AI to anticipate misconceptions

This may be one of the most underrated uses of AI for lesson preparation. Before teaching something, ask:

What are the five most common misconceptions pupils aged 11–12 might have about this concept? For each, suggest a question I could ask that would reveal whether the pupil holds that misconception.

You can go further

Create five plausible incorrect answers to this question. Explain the misconception that might have led to each answer.

Suddenly AI isn't merely generating resources. It is helping you think about what might happen when you teach them.

Create better questions

Give AI the learning objective and ask it to generate questions at different levels of cognitive demand. For example:

Create eight questions about this topic. Start with two retrieval questions, then move through application and explanation to two questions requiring pupils to make connections with previous learning. Explain what each question is testing.

You could then ask

My most confident pupils will complete these quickly. Give me three questions that require deeper thinking rather than simply more work.

Turn one resource into several

This is where the workload benefits can become significant. Take a high-quality text, worksheet or explanation that you already trust and use AI to create:

• A vocabulary list
• Retrieval questions
• A knowledge organiser
• Multiple-choice questions with plausible distractors
• An exit ticket
• A model answer
• A partially completed model
• Extension questions or
• A revision activity

Ask AI to critique rather than create

This is perhaps my favourite tip. Don't always ask AI: What should I do? Sometimes ask: What have I missed? Upload or paste your lesson and try:

Act as an experienced subject leader. Review this lesson critically. What might prevent pupils achieving the intended learning? Where is cognitive load unnecessarily high? Where might misconceptions arise? Which pupils might struggle to access the lesson? Do not rewrite it - give me five things to consider.

You don't have to accept any of them, but you have just acquired a critical friend.

Which AI should teachers use?

There isn't one "best" AI for every teacher, and schools should always use tools approved within their own data protection and safeguarding arrangements. But for September 2026, these are the ones I would explore first.

ChatGPT is probably the strongest all-rounder. It is particularly good for iterative planning: upload existing material, discuss it, challenge the suggestions and progressively refine the result. OpenAI itself emphasises that AI-generated lesson materials are a starting point and that the teacher remains responsible for reviewing them.

Microsoft Copilot is an obvious choice for schools already using Microsoft 365. Microsoft's new ‘Teach’ environment is particularly relevant: it can create lesson plans, quizzes, rubrics and flashcards and adapt materials by reading level, difficulty and other criteria. For eligible Microsoft 365 Education staff, Teach does not require the paid Microsoft 365 Copilot licence.

Gemini for Education deserves similar consideration in Google schools. Gemini can generate learning objectives and lesson plans, while Google's education offer also includes tools for personalising instruction.

Ailafrom Oak National Academy has been designed specifically as a lesson assistant and draws upon Oak's curriculum and sequenced resources. It can adapt lessons for particular classes and contexts, including adjusting literacy demands.

Claude is also an excellent general-purpose model, particularly when you want to work through substantial documents or have an extended planning conversation. Anthropic launched a dedicated Claude for Teachers offer in July 2026, although that specific programme is currently aimed at verified US K–12 educators.

A simple September routine

If you're returning to school and want to experiment, don't attempt to transform everything.

Take one lesson you are already planning and try this five-stage process:

1. Give AI the context - subject, year group, prior learning, curriculum position and learning objective.
2. Give it your existing material - start with something you trust rather than     generating everything from scratch.
3. Describe the  learners - anonymously and without personal data.
4. Ask it to adapt, challenge and critique - rather than simply ‘make a lesson’.
5. Check everything - accuracy, curriculum alignment, reading level, bias, appropriateness and whether it actually improves the learning.

Remember that AI can produce information that sounds authoritative and is completely wrong. It doesn't know your pupils. It doesn't understand the relationships in your classroom. And it cannot exercise the professional judgement developed through years of teaching. The teacher remains the expert.

One final prompt to try in September

When you have finished planning your first lesson, paste the plan into your school's approved AI tool and try this:

You are my critical planning partner. I am an experienced teacher, and I remain responsible for all decisions. Review this lesson from the perspective of the different learners in my classroom. Identify where pupils may struggle, where greater challenge might be needed, likely misconceptions and opportunities to improve accessibility. Suggest no more than five changes that would make the biggest difference to learning. Explain why you are recommending each one. Do not change the learning objective or reduce expectations.

‍