30.4% of US pharmacy preceptors surveyed in 2025 had used AI chatbots (93% of them ChatGPT), and 51.5% planned to start or continue using themSource ↗
In a drug information study, ChatGPT answered only 10 of 39 real medication questions satisfactorily, and every question's response included at least one fabricated referenceSource ↗
75.3% of surveyed pharmacists said they were unlikely to make a patient care decision based on information from a chatbotSource ↗
OpenAI offers US pharmacists with an NPI a free verified ChatGPT for Clinicians tier with clinical search and cited answers — but no HIPAA BAA on the free tierSource ↗
writingClaudeChatGPT

Prior authorization and formulary-exception letters that answer the denial

Rejections land at the pharmacy counter, but the supporting letter that gets them overturned usually waits on whoever has 30 free minutes — and nobody does. Drafting prior authorization and formulary-exception letters is one of the best-documented generative AI wins in clinical practice, and OpenAI's clinician tier now includes prior-auth drafting as a headline feature for pharmacists.

Prompt
You are a clinical pharmacist who writes prior authorization support letters and formulary exception requests that address the plan's stated criteria point by point. Draft a letter for this request.

Medication requested: {{medication_requested}}
De-identified clinical context (diagnosis, prior therapies, relevant labs): {{clinical_context}}
Denial reason or plan criteria, if known: {{denial_reason}}

Requirements:
- Write from the pharmacy to the plan's pharmacy review department, ready for the pharmacist and prescriber to review and sign.
- Structure: one-line request, clinical background, why formulary alternatives failed or are inappropriate, a direct response to each element of the denial reason, closing request with a review timeframe.
- Use only the facts I provided. Do not invent labs, dates, doses, or trial durations. Where the plan will expect a fact I did not give you (dates of prior trials, lab values, documented intolerance), insert [NEED: description] so we can pull it from the record.
- Reference plan criteria language only if I pasted it — never guess what the plan's policy says.
- Plain clinical prose, under 350 words, no hedging filler.
- End with a checklist of supporting documents this request type usually needs.

Fill in your details and the prompt updates live — then copy.

communicationClaudeChatGPTGemini

Plain-language counseling handouts for new prescriptions

Counseling happens in the 90 seconds a patient will actually stand at the counter, and the leaflet the system prints is dense enough that most of it goes unread. Pharmacists now use AI to turn their own counseling points into one-page handouts at a 6th-grade reading level — the pharmacist supplies the clinical content, the model does the plain-language work, and the pharmacist verifies before anything is handed over.

Prompt
You write patient education materials for {{pharmacy_name}}. Create a one-page counseling handout for {{medication}} at a 6th-grade reading level.

Counseling points I want covered (from the labeling and my own review): {{counseling_points}}

Structure:
1. What this medicine is for — one sentence, everyday words.
2. How to take it — timing, food, and missed-dose guidance, only as I described in my points.
3. What you may notice at first — common effects with realistic time frames and what usually helps.
4. Call the pharmacy or your prescriber if — warning signs, each on its own line.
5. Three short Q&As answering: {{common_questions}}.

Rules:
- Use only the counseling points I provided. Do not add doses, interactions, or warnings I did not list — if you believe something standard is missing, put it at the end under "Pharmacist: consider adding" rather than inside the handout.
- No jargon without a plain-word translation, no scare language, no promises about results.
- Under 400 words so it prints on one page. Then give me a 2-sentence version I can say verbatim at the counter.

Fill in your details and the prompt updates live — then copy.

analysisClaudeChatGPT

Drug information answers with citations you then verify

Summarizing information is the single most common AI use pharmacists report — 53% of chatbot users in a 2025 survey — but the same literature shows why raw chatbot answers cannot be the final word: one study found ChatGPT answered only 10 of 39 real medication questions satisfactorily and fabricated references. The workable pattern is a structured prompt that forces the model to separate what it knows from what it is guessing, feeding a verification step the pharmacist was going to do anyway.

Prompt
You are a drug information pharmacist. Answer this question for a practicing pharmacist, not a patient.

Question: {{question}}
Setting and relevant de-identified context: {{context}}

Format:
1. Bottom line — 2-3 sentences, a direct answer with the strength of the evidence stated plainly.
2. What the evidence says — key findings, clearly distinguishing FDA labeling from practice guidelines from primary literature.
3. What I could not verify — anything you are uncertain about, stated explicitly.
4. Verify before acting — the exact items I should confirm in Lexicomp, Micromedex, or the package insert before this touches a patient.

Rules:
- If you are not certain a source exists, write "no citation available" — never fabricate a citation, author, or journal. I will check every reference you give.
- Tag every dose, interaction severity, and stability claim with [VERIFY] — those come from primary references, not from you.
- If the question cannot be answered without patient-specific data I did not provide, say what is missing instead of assuming.
- Under 400 words.

Fill in your details and the prompt updates live — then copy.

planningClaudeChatGPT

Drug-shortage response plans and conversion charts

Hospital pharmacy teams average around 20 hours a week managing drug shortages, and every substitution generates the same artifacts: a staff memo, a conversion table, counter talking points, and prescriber calls. AI drafts that whole package from the alternatives your wholesaler can actually supply — with every conversion flagged for an independent pharmacist check, because dosing math errors are exactly how shortages turn into medication errors.

Prompt
You are a pharmacy operations lead helping manage a drug shortage. Build a shortage response package for {{shortage_drug}} in a {{care_setting}}.

Alternatives we can actually obtain, per our wholesaler (with strengths and forms): {{available_alternatives}}

Produce:
1. A staff memo — what is short, expected duration and impact, and the substitution hierarchy in order of preference. Under 300 words.
2. A conversion table — each alternative beside the original with the strength and volume arithmetic shown step by step, and a [PHARMACIST VERIFY] tag on every row. No conversion goes live without an independent check.
3. Counter talking points — how to explain the switch to patients in plain language, and which questions get routed to the pharmacist.
4. A short prescriber call script for cases where a therapeutic (not generic) substitution needs authorization.

Rules:
- Use only the alternatives I listed. Do not propose other drugs, imports, or compounding options.
- Show your arithmetic on every conversion; if information needed for a safe conversion is missing, say so instead of estimating.
- Flag every point where state law or P&T policy likely requires prescriber authorization rather than pharmacist substitution, marked for local confirmation.

Fill in your details and the prompt updates live — then copy.

automationChatGPTClaude

Refill, adherence, and vaccine outreach messages

Refill reminders and vaccine campaigns work — large text programs get roughly 44% of reminded patients to refill — but the identical blast message ages fast, and writing fresh copy per campaign never makes it up the priority list. Pharmacies now use AI to draft segmented multi-touch sequences with merge fields, then load them into their pharmacy messaging platform, which handles consent and opt-outs.

Prompt
You write patient outreach messages for {{pharmacy_name}}, a community pharmacy. Create a 3-touch campaign for: {{campaign}}.

Touches:
1. Text message — under 160 characters, one clear ask, includes {{contact_method}}, ends with "Reply STOP to opt out."
2. Follow-up text 10-14 days later — a different angle (convenience, insurance coverage, protecting family), not a repeat of touch 1. Under 160 characters.
3. A 30-second phone script for staff or our IVR — greeting, reason for the call, one question, an easy close.

Rules:
- Use [FIRST_NAME] and [MED_OR_SERVICE] merge fields. Never write real patient names, and do not attach a specific drug name to a specific patient in this chat.
- Nothing in a text may reveal a health condition or a specific medication — anyone holding the phone can read a text. "Your prescription is ready to refill" is fine; the drug name is not.
- No pressure tactics, no fake deadlines, no exclamation-point pileups.
- Give two variants of touch 1 to A/B test, plus a one-line success metric for the campaign.

Fill in your details and the prompt updates live — then copy.

writingClaudeChatGPT

MTM notes, medication action plans, and prescriber letters

A comprehensive medication review produces three documents — a patient-facing action plan, a personal medication list, and often a prescriber letter — and the writing regularly outlasts the visit itself. Turning visit shorthand into those deliverables is one of the cleanest AI wins pharmacists report: the clinical findings stay yours, the model does the formatting and the plain-language translation, and vendors in this space report saving 30-45 minutes of documentation a day.

Prompt
You are a clinical pharmacist documenting a comprehensive medication review (CMR). From my visit shorthand, produce the standard deliverables plus a prescriber letter.

De-identified medication list: {{med_list}}
Drug therapy problems I identified and what was discussed: {{visit_shorthand}}

Produce:
1. Medication Action Plan — patient-facing, 6th-grade reading level, structured as "What we talked about / What I need to do / Questions for my next visit," one entry per issue.
2. Personal Medication List entries — each drug with "what I use it for" in the patient's own terms and "how I take it," using only the directions I provided.
3. Prescriber letter — one drug therapy problem per paragraph: the finding, why it matters clinically, and my specific recommendation, closing with a respectful request to advise.

Rules:
- Document only the problems and interventions in my shorthand. Do not add interactions, diagnoses, or recommendations I did not make. If you notice a possible issue I did not list, put it under "Pharmacist review suggested" at the end, clearly separated from the record.
- Where a standard element is missing (indication, follow-up date, prescriber placeholder), insert [NEED: element].
- Patient materials under 350 words total; letter under 250 words.

Fill in your details and the prompt updates live — then copy.

Common questions from pharmacists

Is it a HIPAA violation to use ChatGPT at my pharmacy?

Using it with patient-identifiable information is — OpenAI signs a Business Associate Agreement only for its ChatGPT for Healthcare product, not for free, Plus, Team, or Enterprise accounts, and removing the name alone is not de-identification. The compliant pattern is to fully de-identify before pasting, or use a BAA-covered tool for anything tied to a real patient.

Can I trust AI for drug interaction and dosing questions?

Not as a final source. In published testing, ChatGPT answered only 10 of 39 real medication questions satisfactorily, fabricated references, and missed a genuine Paxlovid-verapamil interaction. Use it to draft and summarize, then verify every clinical claim in Lexicomp, Micromedex, or the labeling — the same check you would do anyway.

Is there a free AI tool built for pharmacists?

OpenAI offers ChatGPT for Clinicians — a free, NPI-verified tier for US pharmacists with clinical search and cited answers, plus drafting tools for documents like prior-auth letters. Note it still carries no HIPAA BAA, so the de-identification rule applies exactly as it does on any consumer account.

Will AI replace pharmacists?

The evidence points the other way: ASHP's Statement on Artificial Intelligence in Pharmacy frames AI as a tool operating under pharmacist oversight, and 75% of surveyed pharmacists would not base a patient care decision on a chatbot. AI is absorbing the typing — letters, handouts, documentation — while verification and clinical judgment remain the licensed work.

Related professions