AI prompts for parent communication
Written for teachers. The hard email — grades, behavior, a missed assignment — drafted calm and factual on the first try, with nothing identifiable pasted in.
On sale now
Parent Communication for Teachers
For the message you have been putting off since Tuesday.
Or the whole shelf: Every Teacher Kit — all six for $28.
01 — What is in Kit 03
- 01 Ten sections, from the good-news note in September to the reply sitting in your inbox at nine p.m.
- 02 30 fill-in prompts — the hard email, the reply to an angry parent, the conference sheet, the contact log entry, the handoff note
- 03 30 follow-up steps that take a first draft to something you would actually send
- 04 30 failure modes, each with the prompt that repairs it
- 05 An 84-page manual, a printable quick-reference page, and a plain-text file every prompt copies cleanly from
02 — The email you rewrite at nine p.m.
The prompt at the center of this kit does not ask for “a parent email.” It asks for the four things that decide whether one works. You give it your grade and subject; what happened, as counts and dates rather than adjectives; what you have already tried; and the one thing you want to happen next. What comes back is short, leads with events instead of a verdict, carries a single ask a parent can act on that evening, and closes on partnership without shrinking the facts. There is no student name anywhere in it — the drafts say “your child” throughout, so what comes back is sendable exactly as it stands, and the name goes on later, in your email program.
That first draft is usually most of the way there. The follow-up chain is the rest, sent in the same chat under the draft. One step warms a draft that reads colder than you meant it, using a genuine positive you supply — and it is built to ask you for that rather than invent one. One cuts the email under ninety words, counted, because parents read on phones, and it cuts background before it cuts the next step. One rewrites it in plain words for a parent reading fast or reading in a second language. All three hold the facts and the ask fixed while they move the tone.
Ten sections work the same way: the positive note home and the September introduction letter, the hard email, the reply to an angry email, conference prep, the whole-class update, the plain-language rewrite and the courtesy translation, the contact log entry and the “per our conversation” follow-up, the second touch and the handoff note to a counselor, the apology when the mistake is yours, and the template file you keep for next year. Each section gives you three ways in: a prompt you fill in, a prompt that matches a message you already wrote, and a prompt that interviews you five questions at a time when you are thumb-typing in the parking lot.
03 — When it comes back wrong
Models fail at writing to families in a small number of predictable ways, and the kit names all thirty of them before you meet one. Three of them show up no matter which message you are writing.
The first is the verdict. Ask for a hard email and the draft comes back calling your student “disruptive,” “unmotivated,” or says they “don’t seem to care.” Those are conclusions, and a parent who reads a character judgment stops reading facts and starts defending their child — the email has failed before the second paragraph. The repair replaces every conclusion with what was observed, dates and counts, and leaves the ask alone. Its cousin on an angry email is the blanket apology: the draft is sorry for something you did not do, which is an admission in a thread that gets forwarded to a principal in March.
The second is invention. A model handed a thin brief completes the picture: a departure time on a field-trip email nobody proofreads twice, a benchmark score on a conference sheet that gets read aloud, a call length in a log entry you never wrote down. The kit’s answer is structural — log prompts write “not recorded” rather than close a gap — and each has a repair when one gets through anyway. The third is the promise: drafts that commit you to a retake, a weekly progress email, or a grade coming up. Every promise in a parent email is one you will be held to in November, and the repair strips out everything except the one step you actually named.
04 — What is free, what is paid, and what never goes in the chat box
The free starter card carries the four rules and ten cross-cutting prompts at principle level — enough to get a usable first draft out of a model. It does not carry Kit 03’s follow-up chains, its thirty failure modes and repairs, or the quick-reference page. That line is deliberate and it does not move.
The privacy design is structural, not advisory. Every drafting prompt tells the model to write “your child” and never a name; log and handoff prompts say “the student.” Situation facts enter as counts and dates, never as the gradebook row. Another child in an incident is described by role and never named — a second family’s privacy is not yours to paste. And where a prompt asks you to paste something of your own — an email whose voice you want matched, last week’s newsletter, your school’s conference form — the instruction is to replace every student name with XX first, and the draft is told to come back with no XX left in it.
Some messages are not drafted here at all, and the manual says so on the rules page and again at the point of temptation. Custody and family-court matters, suspected abuse or neglect, medical information, formal discipline proceedings, and special-education legal process follow your district’s process, inside your school’s systems, with the people responsible for them. Nothing in the kit generates IEP or 504 content, grades anything, or tells you what you are required to record, retain, or report — one repair exists purely to strip that out of a draft that offered it and hand you the questions for your administrator instead. On models: the runs behind this kit were against Claude, in the exact wording that ships, and a dated model-notes page inside says that plainly. Nothing outside the Claude family has been tested, and the kit does not claim it.
How we handle this
Student privacy is absolute. Every prompt uses placeholders ([STUDENT], [GRADE], [SUBJECT]) — never instruct a reader to paste identifiable student information. No prompt may generate legally-binding IEP content or autonomous grading decisions; AI drafts, the teacher decides.
Teacher Kit 03 · $7 · 84-page manual · on sale now
05 — Questions
- Do I have to put a student’s name into it?
- No, and the kit is built so you never need to. Every drafting prompt tells the model to write “your child” and never a name, so the draft is sendable exactly as it stands — you add the name in your email program if you want it there. Log and handoff prompts say “the student.” Where a prompt asks you to paste one of your own messages, the instruction is to replace every student name first, including any other child the message mentions.
- Is this just a list of prompts?
- No. A prompt list gets you a first draft. Kit 03 ships 30 fill-in prompts, 30 follow-up steps that take that draft to something you would send, and 30 documented failure modes each with the prompt that repairs it — 90 blocks across ten sections. It comes as four files: an 84-page manual, a plain-text companion every prompt copies cleanly from, a printable quick-reference page, and a one-page START HERE.
- Can I use it for a custody question, a discipline case, or an IEP message?
- No. Custody and family-court matters, suspected abuse or neglect, medical information, formal discipline proceedings, and special-education legal process are not drafted anywhere in this kit. Those carry legal weight, follow your district’s process, and get written inside your school’s systems by the people responsible for them. The kit drafts communication and points you to your administration for obligations — it does not tell you what you are required to record, retain, or report.
- Which AI does this work with?
- ChatGPT, Claude or Gemini — including their free tiers. That portability is a design property, not a hope: every prompt is a plain instruction with no tool-specific syntax, no custom-GPT dependency and nothing that relies on one tool’s features, so the same wording carries. Our own live-testing runs were made on Claude, in the exact shipping wording, and the dated model-notes page inside the manual says so plainly rather than implying every tool was run.
No charge · No spam · Unsubscribe anytime
Start with the free card
Ten prompts and the four rules behind them — the free starter card, plus one useful message now and then. Nothing else.