46% of nurses say they use generative AI in the workplace, and 62% say AI-enhanced onboarding and training accelerates staff productivity and confidenceSource ↗
Nurses spend an average of 31% of a 12-hour shift documenting in flowsheets alone, per a 2025 JMIR Nursing study — before narrative notes, handoffs, or post-shift chartingSource ↗
ChatGPT produced complete NANDA/NOC/NIC care plans in 35 seconds versus an average of 30 minutes for nurses — but its plans did not always reflect individualized decision-makingSource ↗
A targeted AI education workshop raised the share of nurses with satisfactory AI knowledge from 16.3% to 82.8% immediately afterwardSource ↗
writingClaudeChatGPT

Turning shift shorthand into a complete narrative note

Nurses spend roughly a third of a 12-hour shift on flowsheet documentation alone, and narrative notes routinely push charting 30-60 minutes past the end of shift. Most nurses already keep shorthand on a brain sheet during the shift — the slow part is reconstructing it into complete, defensible prose at 1930 when the next shift has already taken report.

Prompt
You are an experienced {{unit_type}} nurse who writes clear, defensible narrative notes. Convert my shift shorthand into a complete narrative note ready to paste into the EHR.

My shorthand (already de-identified — no names, room numbers, or dates of birth): {{shift_shorthand}}

Format:
- Chronological entries using only the times I gave you
- Objective, factual, past tense — chart what was observed and done, not opinions
- Standard nursing abbreviations are fine; expand anything ambiguous
- Include assessments, interventions, patient response, and provider notifications exactly as I noted them

Rules:
- Use only what is in my shorthand. Never invent vital signs, times, medication doses, or assessment findings. Where a standard element is missing (pain reassessment after a PRN med, order read-back, safety checks), insert [VERIFY: element] so I can confirm before signing.
- Quote the patient's own words for subjective complaints where I noted them.
- No blame language about the patient, family, or other staff.
- After the note, list anything a charge nurse or risk manager would expect for this kind of shift that is still missing.

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

communicationClaudeChatGPTGemini

Rewriting discharge instructions patients actually follow

Nurses do the real discharge teaching — about 22 minutes per patient, versus 3-4 minutes for physicians — yet 40-80% of patients forget or misunderstand instructions shortly after leaving, and most printed materials read far above the recommended sixth-grade level. Rewriting a dense discharge packet into plain language by hand takes time no bedside nurse has at 1400 with two other discharges pending.

Prompt
You are a patient education specialist working with a nurse at discharge. Rewrite these discharge instructions in plain language at a {{reading_level}} reading level for a patient going home from {{care_setting}}.

Instructions to rewrite (de-identified): {{discharge_instructions}}

Output format:
1. "Your condition in plain words" — one or two sentences
2. "Your medicines" — a simple table: name, what it is for, how to take it, one key warning
3. "What to do at home" — short bulleted steps
4. "Call your doctor if..." — warning signs
5. "Go to the emergency room if..." — emergency signs
6. Three teach-back questions I can ask to check understanding

Rules:
- Do not add, remove, or change any medication name, dose, frequency, or medical instruction. Simplify only the words around them. If an instruction is ambiguous, keep the original wording and mark it [ASK PROVIDER].
- Do not add any medical advice that is not in the original instructions.
- Short sentences. Everyday words. Second person ("you"). Avoid words over three syllables where a simpler word exists.

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

planningChatGPTClaude

Drafting care plans you critique instead of write from scratch

Writing a care plan from a blank page averages about 30 minutes per patient. A 2025 comparative study found ChatGPT drafted complete NANDA/NOC/NIC care plans in 35 seconds and scored higher on structure and terminology than practitioners — but its output "did not always reflect individualized decision making," especially on complex cases. The realistic workflow is the reverse of copying: generate a draft fast, then spend your time individualizing it.

Prompt
You are a clinical nurse educator who drafts care plans for nurses to critique — not to copy. Draft a nursing care plan for this de-identified scenario in {{care_setting}}: {{patient_scenario}}

For each of the top 3 nursing diagnoses, prioritized by risk:
- The diagnosis in NANDA-I style (label, related factors, defining characteristics), flagged [CONFIRM AGAINST CURRENT NANDA-I] because taxonomy wording changes between editions
- Two measurable, time-bound expected outcomes
- Four to six interventions, each with a one-line rationale
- Evaluation criteria for end of shift

Rules:
- Use only the assessment findings I gave you. Never invent labs, vital signs, or history. If a finding you would normally need is missing, list it under "Assess first" instead of assuming it.
- Order interventions around my shift priority: {{shift_priorities}}
- End with a section called "What this plan cannot know" — the individual factors (patient goals, psychosocial context, home situation, cultural needs) I must add before this plan is usable, because a template is not an individualized plan.

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

analysisClaudeChatGPT

Breaking down new research and policy changes for the unit

Practice councils, journal clubs, and policy rollouts all land on the same people: nurses who have to read a 20-page study or a rewritten protocol and explain to the unit what actually changes at the bedside. Most nurses report low confidence with research appraisal — one workshop study found only 16.3% had satisfactory AI knowledge at baseline — and the reading happens on unpaid time if it happens at all.

Prompt
You are an evidence-based practice mentor for bedside nurses. Summarize the following article or policy for a 10-minute unit huddle with {{audience}}.

Pasted text: {{document_text}}

Output:
1. One-sentence bottom line
2. "What changes at the bedside" — at most 5 bullets of concrete practice changes for {{unit_context}}
3. "Strength of the evidence" — study design, sample size, setting, and limitations in plain words
4. "What it does NOT say" — the most likely overreadings to warn the unit against
5. Three questions the unit should ask before adopting this

Rules:
- Use only claims that appear in the pasted text. Quote key numbers exactly and name the section they came from. If the text does not address something, write "not addressed" rather than filling the gap.
- Do not add citations, statistics, or studies from memory — I will pull supporting literature myself.
- Plain language throughout; no jargon the newest nurse on the unit would not know.

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

creativeClaudeChatGPTGemini

Building preceptor scenarios and quizzes for new nurses

Precepting comes with no extra hours: preceptors build teaching scenarios, competency checklists, and quizzes on their own time, and quality varies wildly between units. In Wolters Kluwer's 2025 survey, 62% of nurses said AI-supported onboarding and training accelerates new-staff productivity and confidence — and scenario generation is the piece a general-purpose AI does well, because the patients are fictional by design.

Prompt
You are a nurse educator who designs simulation scenarios. Create a fully fictional, realistic training scenario for a {{orientee_level}} orientee on a {{unit_type}} unit, focused on {{skill_focus}}.

Structure:
1. The report the orientee receives at the start, in SBAR format
2. Three staged updates that unfold in sequence, each requiring a decision — include realistic vitals and assessment changes
3. For each stage: the expected nursing actions, and the common mistakes new nurses make at that point
4. Five debrief questions that probe clinical reasoning, not recall
5. A 5-question knowledge quiz with answers and a one-line rationale for each

Rules:
- The patient must be entirely fictional with placeholder demographics. Do not model the scenario on any real case, even if my description resembles one.
- Where the correct action depends on facility policy (escalation thresholds, rapid response criteria, restraint use), write [PER FACILITY POLICY] instead of inventing a rule, so I can insert ours.
- Keep drug doses and vital sign ranges clinically plausible, and flag any value I should double-check against a current drug reference before teaching it.

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

automationClaudeChatGPTCopilot

Drafting fair schedules and the admin work around them

Charge nurses and managers lose hours to schedule-building, swap requests, and the follow-up messages nobody enjoys writing. In focus groups, nurses said an AI scheduler could act as "a neutral entity" that reduces the emotional burden on whoever builds the schedule and distributes weekends more fairly — but every group insisted a human keeps final authority, because "a computer can't feel how someone's really doing."

Prompt
You are a scheduling assistant for a charge nurse. Draft a schedule proposal for {{schedule_period}} and flag fairness problems. The final call is always mine.

Staffing rules: {{staffing_rules}}

Requests and constraints — staff are listed by code only, never by name: {{shift_requests}}

Output:
1. A schedule table: staff code by day, shift type in each cell (D/N/O)
2. "Rule violations I could not avoid" — every cell that breaks a rule, and which rule it breaks
3. "Fairness check" — weekend, night, and holiday counts per person compared to the unit average
4. "Requests I could not honor," each with the specific reason
5. A short, neutral message template I can adapt for staff whose requests were not honored

Rules:
- Never invent availability, requests, or qualifications that are not in my list. If information is missing for someone, leave their row blank and flag it.
- Apply my rules exactly as written. Where two rules conflict, show me the conflict — do not silently pick a winner.
- No judgments or speculation about why anyone requested time off.

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

Common questions from nurses

Is it a HIPAA violation to use ChatGPT or Claude at work?

It depends entirely on what goes in. Consumer accounts have no Business Associate Agreement, so entering protected health information — names, MRNs, dates, or detail combinations that could identify a patient — is a violation. De-identified text is generally permissible, but check your facility's AI policy first; many now have one.

Can AI write my nursing notes for me?

It can draft them from your own observations, but you are legally accountable for every word you sign — boards of nursing and courts treat the signed note as yours, not the AI's. Read every line, delete anything you did not observe, and follow your facility's documentation policy.

Will AI replace nurses?

The profession's own bodies say no, and the evidence backs them. The American Nurses Association and American Academy of Nursing both hold that AI must support nursing judgment, not supplant it — and studies show AI performs worst exactly where nursing matters most, on complex cases needing individualized, contextual judgment.

What do nursing organizations actually say about using AI?

The ANA has an ethics position statement on AI in nursing practice, and in 2026 called for nurse-led guardrails, citing accountability gaps, algorithmic bias, and erosion of professional judgment as the top risks. The practical takeaway is that nurses should use AI with human oversight, protect patient data, and stay accountable for outcomes.

Related professions