Your reading data came back and it's what you expected: some kids reading three grades above level, some kids reading two grades below, most of them somewhere in the middle, all in the same forty-minute block. Writing three or four completely separate lesson plans every night isn't sustainable, and it isn't actually necessary either.
The one-structure, many-texts model
The move that scales is keeping the *structure* of the workshop identical across groups and varying the *text and the scaffolding*, not the thinking you're asking kids to do. If every group is doing "identify the author's claim and one piece of evidence," the skill stays constant even though the reading level of the passage changes.
A single 40-minute block usually breaks down like this:
- Mini-lesson (8 min): whole class, one skill, modeled on a short shared text everyone can access
- Small group / independent reading (25 min): tiered texts on the same skill, rotating groups if you have the staffing, stations if you don't
- Share-out (7 min): each group reports on the same question, at whatever text level they read at
Where the actual planning bottleneck is
The mini-lesson is easy, it's one text for the whole class. The bottleneck is finding or adapting three or four versions of the same skill-focused text at different reading levels, every single week. That's the part that eats your prep time.
This is where AI genuinely helps, because text-leveling and scaffold-writing is exactly the kind of transformation task it's good at, as long as you give it a real anchor text and you check the output. Try this:
1. Paste your grade-level anchor text (or a paragraph of it) and the skill you're teaching.
2. Ask for two additional versions: one simplified in vocabulary and sentence length for your below-level readers, one with denser vocabulary or an additional layer of nuance for your above-level readers, both keeping the same content and the same skill focus.
3. Ask for a short scaffold (three guiding questions, or a sentence frame) for the below-level version specifically.
You're not asking the model to invent new content from nothing, you're asking it to re-render content you chose at different reading levels, which is a narrower and more reliable task.
Checking the output before it reaches kids
Always read the simplified and extended versions yourself before handing them out. Models sometimes over-simplify to the point of losing the actual meaning, or they add a detail that isn't in your original text. If a leveled version changes what the passage is actually about, it's not usable, it's a rewrite you need to fix or a version you should redo from your own edit of the original.
Our [differentiation plan generator](/differentiation-plan) builds this same tiered structure across a full lesson, not just one text, if you want the mini-lesson, small group task, and share-out all mapped out together instead of assembling the pieces separately each week. And if part of your spread includes ELL students specifically, our [ESL/ELL lesson and scaffolds generator](/esl-lesson) adds language-level scaffolding on top of the reading-level tiering, so you're not solving two differentiation problems with two unrelated tools.
What this buys you
The workshop still takes real planning, that doesn't go away. What changes is where your planning time goes: instead of writing three unrelated lesson plans, you're writing one strong structure and one strong anchor text, then spending the saved time checking that the leveled versions are actually right for the kids reading them.
