A 10-Minute Weekly AI Lesson Planning Workflow

Lesson planning is not one task. It is four tasks wearing the same coat: deciding what to teach, sequencing it into a period, writing the materials, and predicting where it will break. AI is useful for two of those and useless for the other two.

The workflow below takes about ten minutes per lesson once you have run it twice. It is built around that split rather than around a tool.

The four jobs inside "planning a lesson"

Separate them before you open anything.

1. What to teach. Decided by your curriculum map and last week's assessment. Not a job for a model.
2. How to sequence it. Objective, warm-up, modelling, practice, exit ticket, timed. This is the part worth drafting.
3. The materials. Worksheet wording, example sets, question stems. Worth drafting.
4. Where it breaks. Which step will confuse whom, and what you will do about it. Yours alone.

Write down job one before you start. A single line — "equivalent fractions, because six students still missed it on Friday" — is enough to keep the draft anchored to your class instead of the internet's idea of grade four.

Step 1: build one brief for the whole week

Instead of briefing each lesson separately, write one block that applies to the week:

• Grade and subject
• The standard or standards being assessed
• Period length
• Class profile: size, reading level, accommodations
• What happened last week that matters
• The format you want back

Then, for each lesson, add one line: the topic and the specific skill.

This is the single biggest time saving in the whole workflow. The class profile is the part that never changes, and it is the part that most improves output quality. Keep it in a notes file and paste it in each week.

Step 2: generate one plan, not five

Generate the first lesson. Read it. Fix it. Only then generate the rest.

Generating a week of plans at once feels efficient and is not: you get five versions of the same structural mistake, and you have to fix it five times. Generate one, learn what the tool gets wrong for your subject, then batch.

Step 3: the three edits that matter

Every generated plan needs the same three corrections.

Cut one activity. Plans routinely include four activities in a 45-minute period. Four activities with transitions, instructions, and one interruption is sixty-five minutes. Remove the weakest one before you teach it.

Rewrite the objective as something observable. "Students will understand equivalent fractions" cannot be assessed. "Students will identify two equivalent fractions using a visual model and explain why they are equal" can.

Fix the exit ticket. Generated exit tickets often test recall rather than the objective. One question, matched to the objective, printed on the plan.

Three edits, about four minutes. They are also the three things an experienced teacher would flag in a colleague's plan, which is a useful sanity check.

Step 4: keep a change log

After the lesson, note two things: what actually took longer than planned, and which students the plan lost.

The next week's brief gets one line from that note. Over a term, the brief becomes a precise description of your classes, and the drafts start arriving closer to usable. This is the compounding part of the workflow — not a better tool, a better brief.

What to generate and what to write yourself

Draft these:

• The timed sequence
• Example sets and practice items
• Differentiation scaffolds for the two or three students who need a different entry point
• A parent-friendly summary of what the unit covers

Write these yourself:

• The opening two minutes. The hook, the question, the odd fact. It sets the tone and it is the part students remember.
• Any example that depends on your class's shared history.
• The decision about what to cut when the period runs short.

A realistic time budget

For a single 45-minute lesson, the numbers usually look like this:

Step	Without a draft	With a draft
Sequencing	15 min	2 min
Materials	20 min	3 min
Differentiation	10 min	2 min
Judgement and edits	—	4 min
Total	~45 min	~11 min

The saving is real but it is not the whole period. What you get back is roughly thirty minutes per lesson, which is two and a half hours a week for a full timetable — enough to matter, not enough to replace planning.

Frequently asked questions

Does this work for subjects other than maths?

The workflow transfers; the edits change. In English, the recurring problem is generic comprehension questions rather than pacing. In science, it is safety instructions and equipment assumptions. In history, it is sources that do not exist. Generate one plan, read it like a sceptic, and write down the failure mode for your subject.

How do I handle mixed-ability classes?

Put the spread in the brief rather than asking for "differentiation". Naming the reading level and the specific scaffolds your class already uses produces usable options; asking for differentiation produces the word "scaffolding" four times.

Can I reuse plans from last year?

Yes, and it is faster than generating fresh ones. Paste last year's plan in with the instruction to update it for the new standard and the current class profile. You keep the parts that worked and the tool handles the rewording.

Bottom line

A weekly AI lesson planning workflow is not a prompt you paste. It is one reusable brief, one plan at a time, and the same three edits every week. Do that for a month and the drafts get closer, the edits get faster, and the planning block shrinks to something you can do with a coffee instead of an evening.