If you ask classroom teachers what AI tool actually saved them time this year, the answer is rarely a chatbot they use with students. It is a lesson-planning assistant that handled a piece of work the teacher used to do alone on a Sunday afternoon. The shift from “AI in front of students” to “AI behind the teacher’s planning” is one of the clearer practical wins from the past two years, and it is worth a careful look at what these tools actually do, which ones are working, and what they cannot replace.

What lesson-planning AI is good at

The use cases that consistently work share a pattern: they take a task that is mechanical and time-consuming for a human, and they produce a draft that the teacher edits. The draft is not the final product. The teacher’s judgment is. The AI just gets the blank-page work done faster.

Differentiation. Generating five versions of a reading at different reading levels used to take an English teacher most of a prep period; it now takes minutes. Same for math word problems written for different student interests, vocabulary lists adjusted for ELL learners, or a lab procedure simplified for students who need scaffolding.

IEP and 504 alignment. Adapting a lesson to a student’s accommodations is high-leverage and time-consuming work. Lesson-planning tools can produce a first draft of an accommodation plan from the teacher’s existing lesson plus the student’s IEP categories, and the teacher refines from there. Special education teachers we hear from describe this as the single most valuable thing AI has done for their workload.

Rubric and exit-ticket generation. A four-criterion rubric for a specific assignment, an exit ticket aligned to a particular standard, a discussion-question set for a reading. These are tasks that the teacher knows how to do; the AI just removes the friction of starting from scratch.

Differentiated parent communication. A three-paragraph progress note to a parent in plain language, a translation of a teacher’s draft into a family’s home language, a tone adjustment when an email is going to be hard to read. AI handles these credibly enough that the teacher’s revision time is short.

The specific tools

The market has consolidated meaningfully since 2023. A few names that have established themselves.

Magic School AI is the most-named platform among teachers we hear from, with the broadest tool set: differentiation, IEP adaptation, rubrics, exit tickets, behavior intervention drafts, parent communication, and a long tail of smaller tools. The free tier is generous, and the paid tier through schools is reasonable.

Brisk Teaching runs as a Chrome extension over Google Docs, which makes it the lowest-friction option for teachers already living in the Google Workspace ecosystem. Strong differentiation tools and a feedback feature for student work that some teachers love and some find too automated.

Diffit specializes in reading-level adaptation. If the differentiation use case is the only one a teacher needs, Diffit is the most refined tool for that specific job. Less useful for general planning.

Curipod and Eduaide.ai cover similar territory with somewhat different interfaces; teachers who have tried multiple tools tend to stick with one of these or with Magic School.

For more general use, ChatGPT and Claude work directly with careful prompting. The dedicated tools save time on prompt engineering and provide guardrails that pure chat does not, but the underlying capability is similar.

What lesson-planning AI is not good at

Generating a complete lesson from a standard. The output looks polished and is shallow. The lessons that AI produces from scratch tend to be generic, fail to anticipate the misconceptions a real classroom will surface, and ignore the particular kids in the room. Teachers who try to use AI to replace lesson planning end up with worse lessons than teachers who use AI to accelerate the planning they were already doing.

Replacing pedagogical judgment. AI does not know which question to ask first, which student needs which scaffold, which lesson concept will trip up your fifth-period class specifically. These are the parts of teaching that take expertise to develop and the parts AI cannot manufacture. Lesson-planning AI is a tool for teachers who already plan well; it does not turn a weak planner into a strong one.

Producing fact-accurate content without verification. AI hallucinates dates, citations, and specific facts. Anything that goes in front of students should be checked. This sounds obvious; it is not always done. Teachers who skip the verification step quietly accumulate errors in their materials.

Privacy considerations

The privacy contracts on AI lesson-planning tools matter and are not all equivalent. Some tools store teacher prompts indefinitely; some allow opt-out; some are explicitly compliant with FERPA and state student-data laws and some are not. Districts adopting these tools at scale should read the data-use clauses with the same scrutiny they apply to LMS contracts. The same logic applies to LMS data clauses, and the AI add-ons inside LMSes inherit some of those privacy decisions.

Teachers using these tools individually rather than through a district should be careful about pasting student names, IEP details, or other PII into general-purpose AI tools that have not been vetted. The dedicated education tools handle this better than ChatGPT or Claude do by default.

A teacher’s evaluation framework

If you are deciding whether a tool is worth integrating, three questions worth asking before adopting.

Does it speed up a task you were already doing well? If yes, it probably saves real time. If no, it is replacing your judgment with the tool’s, which is the wrong substitution.

What is the failure mode? If the AI produces a flawed output, will you catch it before it reaches students? If the answer requires the tool to be perfect, the tool is the wrong tool.

What does it cost in attention? Some teachers find the tools genuinely freeing; others find that interacting with the prompts is itself another form of cognitive load. Try a tool for a week; if it is not making your week lighter, it is the wrong fit.

The broader picture on AI in classrooms is still moving. Our earlier write-up on three years of AI in schools covers the larger question of where the technology helps and where the panic was misplaced; lesson-planning tools are the corner of that landscape that has settled into something useful.