There is a reason generic AI lesson plans do not survive contact with a classroom. A chatbot that has never seen your syllabus, your reading level, or the vocabulary your department agreed on will write something plausible and slightly wrong, and fixing it takes longer than starting over.
ClassSense inverts that. You give it the material you already teach, a PDF, a slide deck, or a Word document, and it reads that document before it writes anything. The lessons that come back use your sequence, your terms, and your examples.
How it works
Create a course
Name it, set the grade level and subject. Takes about twenty seconds.
Upload a unit
PDF, DOCX or PPTX. No reformatting, no template to fill in first.
It finds the skills
The document is broken into topics, and each topic becomes a skill pack.
You approve
Everything lands in a review queue. Nothing reaches a student until you say so.
What a generated lesson actually contains
A skill pack is not a worksheet. Each one carries:
- Explanations at three depths, so a student who is lost and a student who is bored are not handed the same paragraph.
- Worked examples that show the reasoning, not just the answer.
- Common misconceptions, named explicitly, which is the part most generated content skips.
- Strategy variants, alternate routes through the same idea for students the first route did not reach.
- Practice items that feed the mastery tracker as students answer them.
You are still the editor
Generated content is a draft, and ClassSense treats it that way. Every pack sits in a review queue with your name on the approval. You can edit, reject, or regenerate. A teacher who cannot veto the machine is not using a tool, they are being used by one.
An honest limit. Quality depends on what you upload. A dense, well-structured unit produces good packs. A photocopied worksheet with three bullet points does not contain enough to build from, and no amount of AI fixes that. Start with your best material.
Where the time actually goes
The saving is not really in the writing. It is in the second, third and fourth artifact: the practice set, the review game, the assessment, and the re-teach for the four students who did not get it. Those are what get skipped at eleven at night, and those are what get built here automatically.
Common questions
Does it work with any subject?
It works from whatever document you upload, so it is subject agnostic in principle. It has been used most on science, math and literature. If you teach something unusual, upload one unit and judge the result before committing a term to it.
Will it replace my curriculum?
No. It builds from your curriculum. If your district mandates a scope and sequence, ClassSense follows the document you give it rather than substituting its own.
What if the generated content is wrong?
Reject or edit it in the review queue. Nothing publishes to students automatically. If a pack is consistently wrong for your material, that is worth telling us, because it usually points at something fixable in the extraction step.
Which AI model does it use?
Anthropic's Claude models, through their commercial API. Under those terms, data sent through the API is not used to train the models.