Resources / Prompt library

AI Prompt Library for MedComms Agencies

Ready-to-use starting points for the scientific, editorial, strategic, client, and operational work agency teams do every day.

42
Practical prompts
5
Agency teams

Universal instruction

High-risk medical prompts

Accuracy rules: Do not invent facts, citations, statistics, study findings, quotations, page numbers, or client requirements. Clearly label anything that cannot be verified from the materials I provide. Distinguish between information explicitly supported by the source materials and your own inference. If evidence is missing, say so.

Browse the library

Find the right starting point.

Showing 42 prompts

01

Medical Writer Team

Scientific narratives, clinical data, advisory boards, claims, review comments, and content flow.

01Turn source materials into a first-draft scientific narrative

Accuracy rules are included when copied

I'm creating a [deliverable: advisory board deck / scientific platform / speaker deck / manuscript / MSL training / congress summary] for [audience] about [topic].

I'll provide source documents, references, client direction, and any existing materials.

First, identify:

  1. The core scientific story the materials support
  2. The 5-8 most important messages
  3. The logical sequence in which those messages should appear
  4. Evidence supporting each message
  5. Important evidence gaps or contradictions

Then create a proposed content outline.

For every section include:

  • Communication objective
  • Key message
  • Evidence to use
  • Suggested slide/page title
  • 2-4 supporting bullets
  • Reference(s)
  • Any claims that need additional substantiation

Do not start writing polished copy until you have organized the evidence into the narrative.

02Convert dense clinical data into slide-ready copy

Accuracy rules are included when copied

I'm going to give you clinical study information that is too dense for a presentation.

Rewrite it for a [HCP / specialist / payer / MSL / internal pharma] audience.

Create:

  • A takeaway-style slide title
  • Maximum 3-4 main bullets
  • Any necessary sub-bullets
  • A one-sentence "so what?" interpretation
  • Suggested visual treatment for the data

Preserve clinically meaningful qualifiers, denominators, endpoints, timepoints, statistical information, and limitations.

Do not make the study sound stronger than the evidence supports.

Then show me:
Original meaning -> Simplified wording

so I can confirm that simplification has not changed the scientific meaning.

03Find the actual insight in advisory board transcripts

Accuracy rules are included when copied

Analyze this advisory board transcript as a senior medical strategist.

Do not simply summarize what participants said.

Identify:

  • Repeated themes
  • Areas of strong agreement
  • Areas of disagreement
  • Unexpected comments
  • Unmet educational needs
  • Clinical decision drivers
  • Barriers to treatment/use
  • Evidence physicians appear to want
  • Misconceptions or confusion
  • Differences between participant types, if identifiable
  • Statements that could materially affect medical strategy

For every major insight provide:

Insight:
Evidence from discussion:
Why it matters:
Potential implication for the client:
Confidence: High / Moderate / Low

Separate genuine insights from comments that were merely mentioned once.

04Create an executive summary from an advisory board

Accuracy rules are included when copied

Using the transcript, discussion guide, objectives, and any notes I provide, draft an executive summary for a pharmaceutical client.

Structure it as:

1. What we heard
5-7 most important findings.

2. What it means
Strategic interpretation of those findings.

3. What the client should consider
Potential medical affairs, evidence-generation, education, communication, or field implications.

4. Areas requiring further investigation

Keep the distinction between participant feedback and strategic interpretation explicit.

Prioritize information based on importance to the advisory board objectives, not simply frequency of mention.

05Build a reference-supported claims table

Accuracy rules are included when copied

Review the materials I provide and create an evidence table for the claims in this deliverable.

Use these columns:

| Claim # | Exact claim | Source | Exact supporting evidence | Page/table/figure | Fully supported? | Qualification needed? | Risk/comments |

Evaluate whether each source supports:

  • The exact population
  • Endpoint
  • Comparator
  • Timepoint
  • Direction of effect
  • Magnitude
  • Statistical interpretation

Flag claims that:

  • Overstate the source
  • Combine findings from different studies
  • Drop important qualifiers
  • Imply causality where none was demonstrated
  • Use a broader population than the evidence
  • Would require additional substantiation
06Rewrite MLR comments into actual revisions

Accuracy rules are included when copied

I'm going to provide a piece of content plus MLR comments.

For every comment:

  1. Explain what the reviewer is actually asking for in plain English.
  2. Identify whether the issue is primarily medical, legal, regulatory, editorial, or unclear.
  3. Recommend the minimum revision needed to resolve it.
  4. Draft the revised copy.
  5. Identify any downstream slides/pages that may need corresponding changes.

Return everything in a table:

| Comment | What it means | Recommended action | Revised copy | Downstream impact |

Do not automatically accept a reviewer interpretation if it appears inconsistent with the provided source material. Flag those cases separately.

07Check whether a deck actually tells a coherent story

Accuracy rules are included when copied

Review this deck as if you were the Medical Director doing final internal review.

Ignore minor grammar initially. Focus on the scientific story.

For each slide tell me:

  • What job the slide is performing
  • Whether it advances the narrative
  • Whether it repeats another slide
  • Whether anything important is missing before or after it

Then assess the entire deck for:

  • Narrative flow
  • Evidence hierarchy
  • Redundancy
  • Unsupported leaps in logic
  • Disconnects between objectives and content
  • Excessive background
  • Missing clinical context
  • Weak conclusions

Finish by proposing a revised slide order and identifying slides that should be removed, combined, split, or created.

08Make a discussion guide generate better advisory board insights

Accuracy rules are included when copied

Review this advisory board discussion guide.

The objective is not to generate polite conversation. It is to uncover information the client can actually use.

For every question assess:

  • What insight it is intended to generate
  • Whether it is likely to generate that insight
  • Whether it is leading
  • Whether it is too broad
  • Whether physicians can answer it meaningfully
  • Whether a stronger probe is needed

Rewrite weak questions.

Add probes designed to uncover:

  • Why
  • Decision criteria
  • Tradeoffs
  • Evidence thresholds
  • Patient types
  • Barriers
  • Exceptions
  • Differences in practice

Identify any client objectives that the guide currently fails to address.

03

Medical Editor Team

Copyediting, consistency, abbreviations, references, final QC, quantitative checks, and style.

01Perform a structured medical copyedit

Accuracy rules are included when copied

Copyedit this content as a senior medical editor.

Review separately for:

Grammar and syntax

Clarity

Scientific terminology

Consistency

Numbers, units, symbols, and statistics

Abbreviations

Drug names

Study names

References

Tables/figures

Internal cross-references

Do not rewrite stylistically correct copy just because you prefer another phrasing.

Return only changes that have a defensible reason.

For substantive changes, explain why the change is needed.

02Find inconsistencies across an entire deck

Accuracy rules are included when copied

Review this document for internal inconsistencies.

Specifically compare:

  • Numbers appearing in multiple locations
  • Percentages
  • Sample sizes
  • Study names
  • Author names
  • Dates
  • Drug capitalization
  • Dosing
  • Units
  • Endpoint terminology
  • Population descriptions
  • Abbreviations
  • Figure/table labels
  • References

Produce an inconsistency log:

| Item | Location 1 | Location 2 | Conflict | Recommended resolution |

Do not assume which version is correct unless the source materials establish it.

03Audit abbreviations automatically

Accuracy rules are included when copied

Create an abbreviation audit for this document.

Identify:

  • Every abbreviation used
  • First occurrence
  • Whether it is defined at first use
  • Whether it is redefined unnecessarily
  • Abbreviations appearing only once
  • Inconsistent forms
  • Undefined abbreviations in tables/figures

Return:

| Abbreviation | Definition | First location | Issue | Recommended correction |

Also generate a clean master abbreviation list I can compare against the document.

04Verify references against copy

Accuracy rules are included when copied

Cross-check the reference list, in-text citations, and cited claims.

Identify:

  • References cited but missing from the bibliography
  • References in the bibliography never cited
  • Duplicate references
  • Incorrect author/year combinations
  • Citation numbering problems
  • Claims that appear to lack citations
  • Citations attached to claims they may not support

Do not fabricate bibliographic information.

Flag anything requiring manual verification.

05Run a final QC before client delivery

Accuracy rules are included when copied

Perform a final QC review of this deliverable as though it is about to leave the agency.

Look specifically for high-embarrassment errors:

  • Placeholder text
  • Comments or tracked changes
  • Wrong client/product names
  • Old dates
  • Wrong version numbers
  • Duplicate slides/pages
  • Missing slides/pages
  • Broken numbering
  • Inconsistent headers/footers
  • Incorrect references
  • Undefined abbreviations
  • Typographical errors in headings
  • Inconsistent capitalization
  • Figure/table labeling errors
  • Internal notes accidentally left in copy

Rank findings:
Critical / Important / Minor

06Check numbers without rereading everything manually

Accuracy rules are included when copied

Extract every quantitative statement from this document.

Include:

  • Percentages
  • Patient numbers
  • Doses
  • Durations
  • Dates
  • Confidence intervals
  • P values
  • Hazard ratios
  • Means/medians
  • Ranges

Group repeated values together so I can quickly identify inconsistencies.

Show:

| Metric | Value | Location(s) | Consistent? | Source verification needed? |
07Compare two versions and tell me what actually changed

Accuracy rules are included when copied

Compare Version A with Version B.

Ignore purely cosmetic formatting unless it changes meaning.

Categorize changes as:

  • Scientific/content
  • Client direction
  • Regulatory/compliance
  • Editorial
  • Structural
  • Data/statistical

Highlight any change that:

  • Alters scientific meaning
  • Removes a qualifier
  • Changes a number
  • Changes the population
  • Changes the strength of a claim
  • Introduces a new claim

Finish with a concise summary of changes that require medical review.

08Build a project-specific style sheet

Accuracy rules are included when copied

Analyze these approved client materials and create a project style sheet.

Identify established conventions for:

  • Product naming
  • Generic/brand naming
  • Capitalization
  • Abbreviations
  • Numbers
  • Units
  • Percentages
  • Statistical notation
  • Headings
  • Punctuation
  • Terminology
  • References
  • Client-specific exceptions

Separate:
Observed client convention

from

General editorial recommendation

so we do not accidentally convert client preferences into universal rules.

04

Brand Strategy Team

Research insights, positioning, competitive white space, workshops, archetypes, and messaging.

01Turn research into actual strategic insights

Accuracy rules are included when copied

Analyze these research materials, advisory boards, market research findings, competitive materials, and/or stakeholder interviews.

Do not summarize them chronologically.

Identify:

Observation - What the evidence says
Insight - What it reveals about HCP behavior, attitudes, or decision-making
Implication - Why that matters for the brand
Opportunity - What the brand could potentially do about it

Prioritize insights that could change:

  • Positioning
  • Messaging
  • Evidence communication
  • Education
  • Customer segmentation
  • Engagement strategy

Reject observations that sound interesting but have no meaningful strategic implication.

02Pressure-test a brand positioning statement

Accuracy rules are included when copied

Evaluate this proposed brand positioning:

[POSITIONING]

Analyze it against:

  • Target customer
  • Unmet need
  • Competitive alternatives
  • Differentiation
  • Clinical evidence
  • Credibility
  • Relevance
  • Defensibility
  • Memorability

Identify what a skeptical HCP or competitor could challenge.

Then provide:

  1. Strongest aspect
  2. Weakest aspect
  3. Evidence the positioning depends on
  4. Strategic vulnerabilities
  5. Three stronger positioning directions
03Find white space in a competitive landscape

Accuracy rules are included when copied

I'll provide competitor messaging, claims, campaigns, positioning, clinical evidence, and/or promotional materials.

Map the competitive landscape according to:

  • Core promise
  • Clinical benefit emphasized
  • Emotional/rational framing
  • Patient type
  • Decision driver
  • Evidence used
  • Perceived strength
  • Potential vulnerability

Then identify areas of credible strategic white space.

Do not call something white space simply because competitors have not said it.

It must also be:

  • Relevant to customers
  • Supported or supportable by the brand
  • Meaningfully differentiating
04Generate workshop hypotheses instead of generic discussion questions

Accuracy rules are included when copied

We are holding a strategy workshop about [challenge].

Based on these background materials, generate 10-15 hypotheses worth debating.

Each should take a position rather than ask a generic question.

Format:

Hypothesis:
Evidence supporting it:
Evidence against it:
Why resolving this matters:
Workshop exercise/question:

Focus on disagreements or uncertainties that would materially change strategy.

05Build HCP decision archetypes from evidence

Accuracy rules are included when copied

Using the research I provide, identify distinct HCP decision-making archetypes.

Do not invent demographic personas.

Segment based on observable differences such as:

  • Treatment priorities
  • Evidence requirements
  • Risk tolerance
  • Patient selection
  • Switching behavior
  • Barriers
  • Familiarity
  • Practice constraints

For each archetype provide:

  • Defining mindset
  • Decision drivers
  • Barriers
  • Evidence needed
  • Messages likely to resonate
  • Messages likely to fail
  • Potential engagement approach

Cite the evidence supporting the segmentation.

06Turn a strategy into messaging architecture

Accuracy rules are included when copied

Using this brand strategy and supporting evidence, develop a messaging architecture.

Create:

Strategic communication objective

Single-minded message

3-5 message pillars

For each pillar:

  • HCP need
  • Core message
  • Supporting evidence
  • Proof points
  • Appropriate qualifiers
  • Potential objection
  • Response to objection

Then identify which messages are:
Differentiating / Supporting / Hygiene

07Simulate a skeptical HCP reaction

Accuracy rules are included when copied

Act as a skeptical [specialty] physician reviewing this strategy/message.

Your job is to challenge it, not agree with it.

Tell me:

  • What sounds like marketing language
  • What lacks clinical relevance
  • What claim you would question
  • What evidence you would want
  • What competitor you would immediately compare it with
  • What patient type makes you hesitate
  • What practical barrier the strategy ignores

Then switch back into strategist mode and recommend how we should address each challenge.

08Create the "so what?" slide after research

Accuracy rules are included when copied

I have a large amount of research but need to explain what it actually means for the brand.

Reduce the findings into:

What we know

What we believe this means

What remains uncertain

What the brand should do next

Keep evidence separate from interpretation.

Prioritize no more than five strategic implications.

For each implication provide the evidence that led to it so senior stakeholders can challenge the logic.

05

Business Integration Team

Automation opportunities, operations, SOPs, bottlenecks, resources, governance, ROI, and knowledge.

01Find the best AI automation opportunities in an agency workflow

I'm going to describe an agency workflow from beginning to end.

Analyze every step and classify it as:

  • Requires senior judgment
  • Requires specialist judgment
  • Repetitive but necessary
  • Administrative
  • Data retrieval
  • Data transformation
  • Quality control
  • Handoff/waiting

Then identify where AI or automation could:

  • Remove work
  • Accelerate work
  • Reduce errors
  • Reduce handoffs
  • Improve knowledge retrieval

Score each opportunity from 1-5 for:
Time saved / Ease of implementation / Risk / Business value

Recommend the three opportunities worth piloting first.

Do not recommend automating tasks simply because they can technically be automated.

02Diagnose why a project keeps going over budget

Analyze this project using its scope, original estimate, timeline, timesheets, change requests, revisions, and project communications.

Identify the primary drivers of over-servicing.

Categorize lost time into:

  • Scope creep
  • Excessive client revisions
  • Internal rework
  • Poor briefing
  • Incorrect estimating
  • Senior staff doing junior work
  • Waiting/handoffs
  • MLR-related changes
  • Process inefficiency
  • Unplanned meetings

Quantify the impact where data allow.

Then recommend specific process changes that would prevent the same issue on the next project.

03Convert an informal workflow into an SOP

I'll describe how our team currently performs [workflow].

Convert it into an operational SOP containing:

Purpose

Trigger

Required inputs

Roles

Step-by-step process

Decision points

Quality checks

Handoffs

Escalation criteria

Final outputs

Documentation required

Common failure modes

Also identify where the current process depends too heavily on tribal knowledge or a specific individual.

04Find bottlenecks from project data

Analyze these project timelines, task logs, or project-management exports.

Identify:

  • Steps with the longest cycle time
  • Tasks with frequent delays
  • Repeated approval bottlenecks
  • Excessive handoffs
  • Tasks frequently reopened
  • Teams that regularly wait for inputs
  • Work that repeatedly requires senior rescue

Distinguish between:
Actual working time
and
Waiting time

Then identify the three process changes most likely to improve throughput.

05Build a better resource plan

Based on this project pipeline, deadlines, estimated hours, team roles, and current allocations, identify resource risks over the next [2/4/8] weeks.

Flag:

  • Individuals likely to exceed capacity
  • Underutilized staff
  • Projects competing for the same specialist
  • Concentrations of senior-level work
  • Deadline collisions
  • Work that could be shifted to another role

Provide:

Week -> Capacity problem -> Projects affected -> Recommended action

Prioritize changes that avoid both burnout and unnecessary freelance/contractor spend.

06Create a RACI that reflects how the project actually works

Using the project description, deliverables, departments, and approval process I provide, create a practical RACI.

Avoid assigning multiple people as Accountable unless absolutely necessary.

Include:

  • Briefing
  • Scientific development
  • Strategy
  • Client communication
  • Editorial review
  • Creative
  • Internal approval
  • Submission
  • Revision
  • Final QC
  • Delivery

Then identify:

  • Ambiguous ownership
  • Duplicate approvals
  • Missing decision makers
  • Handoffs likely to cause delays
07Decide whether a new AI tool is actually worth buying

Evaluate [TOOL] for use within our medcomms agency.

Our proposed use case is [USE CASE].

Assess:

  • Current workflow being replaced
  • Number of users
  • Frequency of use
  • Estimated hours saved
  • Quality improvement
  • Integration requirements
  • Training burden
  • Security/privacy considerations
  • Human oversight required
  • Failure modes
  • Vendor lock-in
  • Annual cost

Calculate a simple ROI if I provide salary/rate information.

Finish with:
Buy / Pilot / Build internally / Do nothing

and explain the decision.

08Find knowledge we keep recreating

Review these projects, folders, deliverables, and team workflows.

Identify knowledge the agency repeatedly recreates instead of reusing.

Examples might include:

  • Disease-state summaries
  • Competitor profiles
  • Client preferences
  • Approved messaging
  • MLR precedent
  • KOL information
  • Literature searches
  • Advisory board insights
  • Templates
  • Standard processes

For each candidate identify:

Knowledge asset
Who creates it
Who else could reuse it
How frequently it changes
Best storage format
Whether AI retrieval would add value

Then propose the first five knowledge bases the agency should create.

09Turn meetings into project-management actions automatically

I'll provide a meeting transcript.

Extract only information that should affect project execution.

Identify:

Decisions

Action items

  • Owner
  • Task
  • Deadline

Client requests

Changes to scope

Changes to timeline

Dependencies

Risks

Questions requiring follow-up

Items that should be added to the project tracker

Do not assign owners or deadlines that were not stated. Mark them TBD instead.

10Conduct a post-project operational autopsy

Analyze this completed project to determine what the agency should learn from it.

Review scope, timeline, financial performance, revisions, team feedback, client feedback, and project communications.

Answer:

What went well?

What caused unnecessary work?

Where did the original estimate fail?

Where did the process break down?

What work could have been reused?

What could have been automated?

What knowledge should be captured for future teams?

What should we do differently next time?

Finish with no more than five concrete process changes and assign each a suggested owner/function.