Between the queue, the phone, and the rejections, you write more than most people realize: counseling points, prior authorization justifications, MTM documentation, drug information responses, P&T monographs, shortage memos, technician training. A chat assistant drafts any of those in a minute, in the structure you specify, and it is patient with the fourth revision.
What it cannot do is replace Lexicomp, Micromedex, the package insert, or your own check. A general model will state a renal dose adjustment, an interaction severity, or a beyond-use date with total confidence and no guarantee of being right, and its knowledge stops at a training cutoff while labeling changes every month. Nothing clinical leaves the draft stage until you have verified it in a validated reference.
The data rule is simple. Patient names, dates of birth, prescription numbers, and the other HIPAA identifiers never go into a consumer chatbot; they belong only in systems your employer has covered under a Business Associate Agreement. Rewrite the situation as a fictional or fully de-identified case and the general assistant becomes safe for the writing.
Quick wins this week
- Draft a one-page counseling handout for a medication class you dispense every day at a sixth-grade reading level, then check every statement against the package insert and your counseling references.
- Paste a payer's published prior authorization criteria and a de-identified clinical summary, and ask for a justification letter that walks through each criterion in order.
- Turn a new FDA safety communication or labeling change into a three-line huddle note for the team, with the link to the source.
- Ask for a counter script for the hard conversations: a refill-too-soon rejection, an early fill request on a controlled substance, a cash price that shocked someone.
- Build a continuing education plan for the year from your state board's renewal requirements, including the mandated topics your state requires.
What AI can do for pharmacists, task by task
Patient counseling materials and medication education
Describe the drug or class in general terms and ask for a handout at a specific reading level with sections for purpose, how to take it, common side effects, serious warnings, and when to call. Keep doses out of it; those come from the label on the vial. Check every statement against the package insert, the Medication Guide where one is required, and your organization's approved education content, because models restate old labeling with total confidence.
Prior authorization and formulary exception letters
Give the model the payer's published criteria, the de-identified clinical picture, and the prior therapies with outcomes, and ask for a letter that quotes each criterion and answers it. Verify every guideline or study it cites by opening the source; invented references are common. Anything that needs real patient detail gets finished inside an approved system, not a consumer tool.
MTM and comprehensive medication review documentation
The findings and recommendations are yours; the assistant turns them into plain-language entries for the Medication Action Plan and the cover letter in the standardized CMR format Medicare requires. Work with de-identified findings in a general tool, or inside an approved platform with the real record. Check that nothing you did not recommend has been added and that the reading level suits the patient.
Drug information responses and P&T monographs
Gather the primary sources yourself, upload them to NotebookLM, and ask questions that must be answered from those documents with the supporting passage shown. Use Perplexity to find candidates, then read the paper. The model organizes and drafts; every efficacy number, dosing statement, and citation is verified against the source before the monograph goes to committee.
Shortage memos and therapeutic alternative communications
Paste the ASHP or FDA shortage entry and your P&T-approved alternatives, and ask for a memo to prescribers and nursing that explains the situation, the approved substitutions, and the ordering change. Do not let the model supply conversion doses or alternatives that are not on your approved list; that is where a shortage memo becomes a medication error.
Prescriber interventions and clarification requests
When you have identified a therapy problem from your references, ask the model to draft the fax or secure message in a tight SBAR structure with your recommendation stated once and the supporting facts in two lines. Use a de-identified version in a general tool or the real one in an approved system. Check that it has not softened your recommendation or added an alternative you did not verify.
Technician training, SOPs, and compliance documentation
Paste the relevant section of USP <795>, <797>, or <800>, your DSCSA procedure, or your controlled-substance handling policy, and ask for an SOP draft, a competency checklist, or a ten-question quiz that cites the section for every item. Everything is checked against the chapter text and your state board's rules. Models like to add requirements from generic guidance, so include the instruction 'do not add anything not in the text'.
Prompts for pharmacists
Replace the bracketed placeholders, paste into any chat assistant, and iterate on the result.
Counseling handout for a new medication
Act as a clinical pharmacist who writes patient education. Write a one-page handout for a patient starting [MEDICATION OR DRUG CLASS] in a [SETTING, E.G. COMMUNITY PHARMACY OR HOSPITAL DISCHARGE] setting at a [READING LEVEL] reading level. Sections: what this medicine is for, how to take it (general instructions only, no doses), common side effects and what to do about them, serious warnings and when to call the pharmacy or doctor, what to do about a missed dose in general terms, and storage. Short sentences, second person, no patient details. End with a numbered list of every clinical statement in the handout that I should verify against the package insert before using it.
Tip: Read the verification list against the current label, then the handout against your counseling references. The draft saves typing, not verification.
Prior authorization justification from the payer's criteria
You are a pharmacist writing a prior authorization justification for [MEDICATION]. Payer criteria, pasted from the published policy: [PASTE THE PA CRITERIA]. De-identified clinical summary (no names, dates, or identifiers): [DE-IDENTIFIED CLINICAL SUMMARY]. Prior therapies, durations, and outcomes: [PRIOR THERAPIES AND OUTCOMES]. Write a letter that addresses each criterion in order: quote the criterion, then state the supporting facts from the summary. Cite nothing beyond the criteria and the summary. If a criterion is not clearly met, tell me in a separate note instead of implying it is. Formal, under 400 words.
Tip: Finish the letter with real patient detail only inside a system your organization has approved for PHI.
Drug information response from provided sources only
Act as a drug information pharmacist. Question, de-identified: [CLINICAL QUESTION]. Source excerpts I have gathered: [PASTE THE SOURCE EXCERPTS WITH TITLES AND DATES]. Write a formal drug information response with these sections: question, background, evidence summary, answer, and limitations. Every statement in the evidence summary and the answer must cite one of the provided sources by title. If you need to add anything from your general knowledge to make the answer coherent, put it in a separate section headed 'Unverified, from general knowledge' so I can check it. Do not state any dose or interaction that is not in the sources.
Tip: The 'Unverified' section is the point. It separates what you can defend from what the model guessed.
Shortage memo for prescribers and nursing
You are a pharmacy operations manager. Write a shortage memo about [DRUG IN SHORTAGE]. Shortage details, pasted from the ASHP or FDA entry: [PASTE THE SHORTAGE ENTRY]. Our P&T-approved alternatives and ordering changes: [PASTE THE APPROVED ALTERNATIVES LIST]. Format: situation in two sentences, what changes in ordering, approved alternatives exactly as listed (do not add, remove, or convert anything), expected duration if the source states one, and who to contact. Under 250 words, plain language, suitable for a huddle and an email.
Tip: If the model adds a conversion or an alternative you did not paste, delete it and remind it. Never let a memo carry a dose you have not verified.
Technician quiz from a USP chapter or SOP
Act as a pharmacy educator. Using only the text below, write a 10-question quiz for a [ROLE, E.G. STERILE COMPOUNDING TECHNICIAN]: six multiple choice and four short answer. Each question must cite the section it comes from. Provide an answer key with a one-sentence rationale that quotes the text. Do not add any requirement that is not in the text. Text: [PASTE THE SOP OR CHAPTER SECTION]
Tip: Pair it with a competency checklist request using the same text, and have your pharmacist in charge sign off before it counts as training.
Counter script for a difficult conversation
You are an experienced community pharmacist coaching a colleague. Situation: [SITUATION, E.G. REFILL TOO SOON ON A CONTROLLED SUBSTANCE]. Our policy: [PHARMACY POLICY IN PLAIN TERMS]. What we can offer the patient: [WHAT YOU CAN OFFER]. Write a short script for the pharmacist at the counter: an opening that acknowledges the patient, a plain explanation of the policy without blame, the options, and a closing line. Then give two responses to likely pushback. Keep it under 200 words, no clinical claims beyond the policy, and no promises we cannot keep.
Tip: Adjust the tone to your store. The structure is what you are buying; the words should sound like you.
Want a prompt for something else? Use the Prompt Builder.
Skills to build
Knowing what counts as PHI at the counter
Why: A prescription number, a date of birth, a fill date, or a phone number is enough to make a prompt a disclosure under HIPAA, and pharmacy staff paste them without thinking because they are on every screen.
How: Keep the 18-identifier Safe Harbor list at the workstation and rewrite every real situation as a fictional case before it goes anywhere near a general assistant.
Separating drafting from verification
Why: The model is a strong writer and an unreliable pharmacist. Its interaction and dosing statements are plausible text, not checked references, and the errors are the confident kind.
How: Keep a fixed hierarchy: package insert, Lexicomp or Micromedex, primary literature, then the model for wording only. If a wrong answer could reach a patient, the answer comes from the reference.
Reading generated text for the missing warning
Why: Handouts and memos fail by omission: the dropped 'do not crush', the missing boxed warning, the interaction that was in your source and not in the draft.
How: Compare the draft to its source line by line instead of judging it on its own. Read for what is absent before you read for what is wrong.
Grounding every prompt in the actual document
Why: A model working from memory blends old labeling with new. A model working from the pasted policy, chapter, or insert quotes the right line and stays inside it.
How: Build a folder of the documents you reuse: payer criteria, USP sections, your SOPs, your counseling standards. Start every prompt by pasting the one that applies.
Knowing your board's rules and your corresponding responsibility
Why: Federal law places a corresponding responsibility on the pharmacist for every controlled-substance prescription, state boards regulate what may be delegated, and no software carries either duty for you.
How: Read your state board's current guidance on AI and technology, ask your compliance officer which tools are approved under a BAA, and keep the PDMP check, the red-flag review, and the final verification as human steps in writing.
Tools worth knowing
ChatGPT
The general-purpose AI assistant most of your coworkers already use.
Claude
A careful writing and analysis assistant that shines on long documents.
Microsoft Copilot
AI inside Word, Excel, Outlook, and Teams, with your company's data protections.
NotebookLM
A research notebook that only answers from the sources you give it, with citations.
Perplexity
An answer engine that cites its sources, built for research rather than chat.
Cautions for pharmacists
Consumer chatbots carry no Business Associate Agreement, and a prompt with a name, date of birth, prescription number, or any other HIPAA identifier is a disclosure. Never paste confidential patient data into a consumer AI tool unless your organization has approved that tool and signed the agreements. Use fictional or fully de-identified scenarios for drafting and finish anything with real detail inside an approved system.
Models make arithmetic errors, cite superseded labeling, and rate interactions inconsistently, all in the same confident tone. Renal and hepatic adjustments, pediatric weight-based doses, interaction severity, beyond-use dates, and IV compatibility come from your validated references and your own verification. If AI explains a concept to you, confirm it before it changes what you dispense or recommend.
Your corresponding responsibility under federal law, the PDMP check, the red-flag assessment, and the decision to fill or refuse cannot be delegated to a drafting tool. AI can help you word the conversation or the documentation of a refusal; it cannot decide it. DEA and state board rules on records, inventory, and electronic prescribing are unchanged by any of this.
A model's knowledge has a cutoff date, and labeling, REMS requirements, shortage status, and boxed warnings change after it. Models also fabricate studies with real journal names. Paste current source text into the prompt rather than asking for recall, and open every citation before it appears in a monograph or a letter.
State boards discipline pharmacists for misleading patient information regardless of who drafted it, so generated handouts must match approved sources. The subtler risk is trusting a tool because it is usually right; the same alert fatigue that makes DUR warnings easy to click past applies to AI drafts. Keep the review step real, especially on the tenth draft of the day.
Your 30-day plan
- Week 1: Ask your compliance officer or pharmacist in charge which AI tools are approved, which carry a BAA, and what may be pasted into each. Read the HIPAA identifier list. Use a general assistant for two non-patient tasks: a technician training quiz and a huddle note.
- Week 2: Draft two counseling handouts for high-volume medications and check them line by line against the package insert. Build one prior authorization letter from a payer's published criteria using a de-identified summary.
- Week 3: Load a USP chapter or your compounding SOPs into NotebookLM and use it for a week of lookups. Draft a shortage memo from a current ASHP entry and your approved alternatives, and have a colleague check it.
- Week 4: If your pharmacy system or employer has an approved AI feature, learn its workflow and its review step. If not, write a one-paragraph proposal for your manager on the one task where a licensed tool would help most.
- End of month: Write down the three tasks where AI saved real time and the one place it was confidently wrong, and share both with your team.
Frequently asked questions
Will AI replace pharmacists?
Can I use ChatGPT for drug interaction checks?
Is it a HIPAA violation to type a patient's medication list into ChatGPT?
Can AI help with prior authorizations in the pharmacy?
What about AI features built into pharmacy software?
Terms used on this page
Related roles
- AI for PhysiciansThe documentation and inbox load that follows you home is where AI pays off first. It drafts; you decide, verify, and sign, and nothing with PHI leaves the systems your organization has covered under a Business Associate Agreement.
- AI for NursesAI can take the writing, summarizing, and studying load off your shift without ever touching a patient record, as long as you know where the HIPAA line is and stay on the right side of it.
- AI for Healthcare AdministratorsAdministrators drown in documents: policies, payer correspondence, board decks, survey prep, budget narratives. AI drafts and summarizes all of it, if you keep PHI inside covered systems and set the rules for your organization before someone else sets them for you.
- AI for Medical Coders and BillersAI can read a note faster than you can, and it will also cheerfully assign a code that was deleted two years ago. Used well, it speeds up the reading, the denial letters, and the research; the code selection and the compliance stay yours.