It's a Thursday again. Seven clients seen, a short stack of notes waiting, that familiar 7:15 heaviness settling in. But something's different tonight. You don't open a blank chat window and stare at it, trying to remember how you worded that thing last time. You open a doc you already made. You scroll to the note prompt. You fill in four blanks. Done.
That doc is what today is about.
Thirteen weeks ago, we started with note debt riding home with you. Since then we've built a whole quarter together: the medical-necessity paragraph that helps insurance say yes, the three versions of the parent email, treatment-plan language, psychoeducation your clients will actually read, intake and progress reviews, care coordination, session prep, and the school-side work on IEP goals and behavior-plan wording. Every one of them ran on the same rule, and none of them touched a client's data.
Today is the finale, and it's the one that makes all the others stick: your own prompt library. One home for every workflow you actually used, saved as reusable, de-identified templates, so you never start from scratch again.
The redaction reflex, made permanent
I've said the same rule every week, because it's the one that keeps all of this safe: Never put client names, records, or identifying details into an AI tool. Ever.
Here's the quiet gift of a library. You stop performing the redaction reflex from memory every single time and start baking it into the template itself. The [brackets] force generic placeholders. The guardrail line is already sitting at the bottom. A house-rules header sits at the top, telling the tool up front that everything it sees is de-identified.
This is the lowest-tax issue of the quarter — there's no client note to strip today. You're saving PHI-free language once, so the safe way becomes the default way instead of a thing you have to remember at 7:15 on a Thursday.
The workflow
Make one home. A single doc or note titled My Prompt Library. Not ten files scattered everywhere. One.
Give every entry the same four parts: a title, one line on when to use it, the prompt with [brackets], and your own quick de-ID reminder for that one.
Steal from the last thirteen weeks. Drop in the prompts you reached for most — start with just three.
Put your house rules at the top. Paste the standing header once, so every prompt inherits the same guardrails.
Tune as you go. When a draft comes back a little off, fix the prompt, not just the output. In a month it sounds like you.
Two short prompts get you started. First, the header you save at the very top of your library:
Standing instructions for every draft you help me with:
- I am a licensed therapist. You are helping with wording and structure only.
- Everything I paste is de-identified. Never ask me for names or identifying details.
- Use only the details I provide. Do not invent history, symptoms, quotes, dates, or diagnoses.
- Keep the tone [professional / warm / plain-language] and the length [short / medium].
- Give me a draft I will review and finalize myself.Second, a template-maker — for turning a rough prompt you keep retyping into a clean, reusable one:
Here is a prompt I keep rewriting from scratch: [paste the rough version you've been using].
Turn it into a clean, reusable template I can save:
- Replace every specific detail with a labeled [fill-in-the-blank] field.
- Keep it de-identified — no names, no identifying details anywhere.
- Add a one-line note at the top for when to use it.
- End it with: "Use only the details I provided. Do not invent history, symptoms, or quotes."
Return just the finished template.Finish it like the clinician you are
A library is a starting point that gets better, never a finished thing. The drafts it produces are still drafts — you add the identifying specifics back inside your own EHR or secure email, and you own every word for clinical, legal, and billing accuracy. The library just means you're never starting from zero.
The prompts were never the point. The point is the 7:15 that doesn't own you anymore — the blank screen that turned into a doc you built, on your terms. You start every hard task already halfway done.
Try it this week
Open one doc. Title it My Prompt Library. Add exactly three entries — the three prompts from this quarter you'd miss the most. That's the whole assignment. You can grow it from there.
This is the last issue of the launch quarter, so I especially want to hear this one: hit reply and tell me which single workflow earned a permanent spot in your library. That answer tells me what to build next.
And if building your own library sounds like one more thing on a list that's already too long — reply 'pack' and I'll send details on a done-for-you set of the most-used prompts from this quarter, de-identified and ready to paste.
Talk soon,
Shaun
P.S. That's a wrap on the first quarter — thirteen weeks, thirteen tools, zero client data. If one of them handed you an evening back, forward this to the therapist down the hall who's still writing notes at 9 p.m. It's free, and the whole thing starts over for them at week one.
A quick but important note: this newsletter is for general educational and informational purposes only. It shares time-saving ideas, not advice. The views here are the author's own and don't represent any employer, school district, or agency. Nothing here is legal, clinical, compliance, or supervisory advice, and AI can be wrong. Always follow your licensing board's rules and your employer's and state's policies (especially on AI use), along with your HIPAA and FERPA obligations, and check with the right person before acting on anything significant. You are the professional: review and finalize everything yourself. And never put client names or protected health information into any AI tool.