I teach seventh-grade history, and the running joke in my building is that summer exists so teachers can recover from the last year in time to dread the next one. The thing that actually drained me wasn't the kids or the grading. It was Sunday-night lesson planning. Not because I don't know my content — I can talk about the causes of World War I until someone physically removes me from the room — but because turning 'teach the causes of WWI' into a structured eighty-minute experience for thirty-two twelve-year-olds reliably took me two hours per lesson. Five lessons. You do the math; I refused to.
I kept rebuilding the same structure every single week
Every good lesson has the same skeleton: a hook, direct instruction, guided practice, independent practice, a closing assessment. I knew this in my bones. And yet there I was every Sunday, reinventing that skeleton from scratch for every topic like I'd never heard of it. The waste wasn't in the thinking, it was in the formatting and the scaffolding — the part a machine is genuinely good at.
The Build a Ready-to-Teach Lesson Plan from One Learning Objective prompt, which I run on GPT-4o because it's fast and cheap and I'm doing this five times in a row, forced me to do the one part that's actually mine to do: write a real objective. Not 'teach WWI' but 'Students will identify the four main causes of WWI and explain how each made the next one more likely.' The moment I wrote that precisely, I knew exactly what the lesson had to accomplish. Half my old planning time was just me being vague at myself.
The sections I'd have phoned in
GPT-4o filled the skeleton with timing, teacher moves, student moves, and materials. The HOOK section is where it earned its keep — it suggested opening with a real 1914 newspaper headline and asking the class to predict what happens next, which is exactly the kind of thing I'd have skipped at 9pm on a Sunday in favor of 'today we're learning about WWI, please open your books.' One of those gets thirty-two kids leaning forward. The other gets me a room of people quietly drawing on their arms.
The timing budget is the unsung hero. The prompt forces the sections to add up to the actual block length — 80 minutes, split across hook, instruction, guided practice, independent practice, and closure — with minutes attached to each. I used to chronically over-plan the instruction and run out of time for practice, which is backwards, because the practice is where the learning happens and the lecture is where I feel productive. Having a model hand me '12 minutes direct instruction, 24 minutes guided practice' in writing keeps me honest. I stop talking when the budget says stop talking, which my students have noticed and appreciate more than they'll admit.
- Writing a precise 'I can' objective clarified the whole lesson before any AI was involved
- The HOOK ideas pulled me out of my own ruts — primary sources, prediction prompts, what-would-you-do scenarios
- The single differentiation note covers both struggling and advanced students, so I stop forgetting one of them
- The forced time budget stops me over-lecturing and starving the practice time
- The PITFALL TO AVOID is often the best line in the output — distilled hard-won classroom wisdom
The pitfall line that saved a lesson
For the WWI lesson, the pitfall it flagged was that students conflate the assassination of Franz Ferdinand as 'the cause' when it was the trigger, and that if I don't separate trigger from underlying cause explicitly, half the class writes that one guy got shot and started the whole war. That is precisely the misconception I'd watched happen for three years and never thought to pre-empt in the plan itself. Now I build the distinction into guided practice. It's the difference between teaching the lesson and teaching the lesson I actually want them to learn.
Here's my honest workflow, caveats included: I plan my single best lesson of the week entirely by hand, because the lesson I care most about deserves my full brain. Then I run the other four through this prompt and edit hard — I cut anything that doesn't sound like me, and I never trust its materials list without checking I actually own the materials. Once it suggested a documentary clip that doesn't exist, which it asserted with total confidence, and that's the moment I learned to verify every external resource it names. The model is a planning partner, not a substitute teacher; it has never met my kids and it occasionally invents things, so the editing is not optional.
My planning went from ten hours a week to three. I spend the saved seven on the part no model can do: knowing my students well enough that the hook actually hooks them. The kid who lights up at a 1914 headline is not the same kid who needs a what-would-you-do scenario, and figuring out which is which is the actual job — the prompt just gives me the raw material to make those calls. When a topic needs a quick check-for-understanding I generate it with my Generate a 10-Question Quiz with a Self-Grading Answer Key prompt and drop it into the assessment slot, so the lesson and its check come out of the same Sunday hour. Grab the Build a Ready-to-Teach Lesson Plan from One Learning Objective prompt on Prompt Dock and get your Sundays back.