Claude for K-12: Teacher Planning and Classroom Boundaries

Claude is an AI assistant from Anthropic that K-12 teachers use to prepare lessons faster, build differentiated versions of the same activity, and draft rubrics and feedback stems they can edit and use. It does not grade student work automatically or connect to your gradebook, but it handles the text-heavy preparation work that takes time away from instruction.

If you plan lessons across multiple periods or units, a tool like fenbs can keep your approved teaching instructions, prompts and planning tasks in one place that any connected AI reads before a session, so you stop retyping the same context every time you open a new chat.

What K-12 teachers use Claude for

Most teacher use falls into a few repeating categories: drafting lesson plans from a standard, generating comprehension questions at different difficulty levels, creating differentiated activity versions, writing family communications, and producing rubrics or feedback language.

• Lesson plans: give Claude the standard, grade level, available time and constraints; it drafts a structure you revise.
• Differentiated activities: ask for the same content at three reading levels or with scaffolding for English learners.
• Assessment questions: generate recall, analysis and application questions for a unit.
• Rubrics and feedback stems: describe the assignment and criteria; Claude produces a draft rubric or feedback phrases you edit.
• Family communications: draft newsletters, permission forms or meeting summaries in plain language.

Claude works best when you are specific. "Write a 45-minute lesson plan for fourth-grade science on the water cycle, including three activity segments and an exit ticket, for a class that includes six English learners" gives a far more usable first draft than a vague request.

Checking your school's AI policy before you start

Before using Claude for school work, read your district's AI use policy. Many US districts have published policies covering which tools staff may use, how student data must be handled, and whether AI-generated content must be disclosed to students or families.

FERPA, the Family Educational Rights and Privacy Act, governs student records in the US. Do not paste student names, ID numbers, grades or any detail that could identify a student into a third-party AI tool unless your district has a signed data agreement with that vendor. Draft all materials with generic placeholders instead.

If your school belongs to a state consortium or uses a district-managed Claude account, check whether those accounts carry different terms than the public version at claude.ai. The terms on your account determine what data protections are in place.

Building differentiated materials efficiently

Differentiation is one of the clearest wins for AI-assisted lesson prep. Write the core activity first, then ask Claude to adapt it: a simpler version for students reading below grade level, a challenge extension for advanced learners, and sentence-frame scaffolding for students still developing academic language.

Be specific in your instructions. "Rewrite this reading passage at a fifth-grade Lexile level, keeping the science vocabulary but adding a short parenthetical definition after each key term" produces a more useful draft than "simplify this." The more detail you provide upfront, the less editing you do afterward.

Say a middle school language arts teacher has a close-reading exercise on a primary source. She asks Claude for a version with bolded vocabulary, a brief glossary, and three guided annotation prompts. She then edits for accuracy and decides which annotations suit her class. Claude drafts; she decides what goes to students.

Preparing lessons and assessments

A useful Claude prompt for a lesson plan includes: the subject and grade level, the learning standard or objective, the available time, any constraints (no devices, students need visual supports, limited printing budget), and the output format you want. The more context you give, the less revision the draft needs.

Review every draft before using it. Claude occasionally places a standard in the wrong grade band, suggests an activity that assumes materials you do not have, or proposes a timing allocation that does not fit your class period. You catch those issues; Claude saves you the blank page.

For assessments, give Claude the standard being measured, the grade level, and the format you want — multiple choice, short answer, constructed response, or discussion prompt. Ask for questions at multiple cognitive levels if you are planning a summative test. Review every item before using it.

Feedback and formative checks

For written feedback, describe the assignment and the criteria, then ask Claude to generate a bank of specific feedback stems: comments about organization, use of evidence, thesis clarity or argument structure. You pull from that bank and personalize each comment. This does not replace reading student work; it speeds up writing comments you have already decided to give.

Exit tickets and quick formative checks are another strong use. Give Claude a learning objective and ask for three to five low-stakes questions in a format students can answer in five minutes. Edit the questions to match your class vocabulary before using them.

For teachers across different roles in education, the blog topic on AI by role covers how educators at all levels approach AI-assisted workflows.

What Claude cannot replace in your classroom

Claude does not know your students. It does not know which child needs to sit near the front, which pairs of students work well together, or which topic a class found genuinely confusing last week. Those professional judgments belong to you, and no prompt changes that.

It also should not substitute for IEP meetings, 504 conferences, mandated reporting, or any communication that carries legal standing. Use Claude to draft language you review and edit, not to produce documents you sign without reading.

Talk with your grade-level team or department about how everyone is using AI. Shared norms around disclosure, student data and what counts as AI-assisted material make it easier to maintain consistent practice across classrooms.

Building a prompt library for consistent results

The prompts that produced good results in September will work again in spring if you saved them. Build a small library organized by task type: differentiation, exit ticket creation, rubric drafting, family communication. When you need a new version of a familiar task, start from the prompt that worked rather than writing a new one from scratch.

Adapt prompts for new contexts rather than rewriting them. A differentiation prompt that worked for a reading passage can work for a math word problem with small changes to the subject-specific vocabulary instructions. Keeping the core structure and changing the subject and grade-level details saves significant time across a full year of preparation.

When a prompt needed heavy editing to be useful, note what was missing: too vague, wrong grade assumption, assumed materials you do not have. That note becomes the instruction for the next attempt. Over time, your prompts get more specific and the drafts they produce require less revision.

Pay attention to the prompts that produced results you used with minimal changes. Those are your highest-value instructions. The wording that produced a well-structured rubric for a sixth-grade science argument essay is specific enough to be valuable again next term. Keep it exactly as it worked. Sharing these high-value prompts with colleagues multiplies their value: a lesson preparation prompt that works well for one teacher's class can often work equally well across an entire department.

How to do this with fenbs

1. Create a free account at fenbs.ai. Start on your personal board to try the workflow before inviting colleagues.
2. Add a planning project for the unit or subject you are currently preparing. Give it a short goal and add the current lesson tasks.
3. Write your best prompts and any team format requirements in Skills and prompts. Include the exact wording that produced the best results so far.
4. Connect Claude or another AI assistant from Integrations. At the start of each planning session, ask your AI to read the project context and the saved skills first.
5. Before closing a session, ask your AI to save progress: which lessons are drafted, what still needs review, and where to pick up next time.

A free personal account works for solo planning. If you share prep with a co-teacher or department, a team account lets each member connect their own AI while reading the same project record.

fenbs does not import your Claude or ChatGPT chat histories, gradebook data or student files. The shared record is what you write into it: goals, tasks, approved instructions and saved prompts.