Skip to main content

Promp guidelines

Provide clear, specific prompts that help eVa AI generate the most accurate and relevant results.

Yuliia Skrybantovych avatar
Written by Yuliia Skrybantovych
Updated over 2 weeks ago

Effective prompting is your gateway to leveraging eVa AI capabilities in pharmaceutical contexts, where precision, compliance, and accuracy are critical. A well-crafted prompt serves as clear instructions that guide eVa AI to understand your intent and deliver exactly what you need.

Examples:

Basic: "Write about drug side effects"

Better: "Create a summary of common side effects for cardiovascular medications, organized by drug class, for inclusion in a patient information leaflet."

Best: "Act as a medical writer creating patient education materials. Develop a comprehensive side effects summary for cardiovascular medications targeting adult patients with heart conditions. Organize content by major drug classes (ACE inhibitors, beta-blockers, statins, diuretics). For each class, include: most common side effects (>5% incidence), serious but rare effects requiring immediate medical attention, and when to contact healthcare providers. Use clear, non-technical language appropriate for patients with high school education level. Format as a structured reference guide with consistent formatting, approximately 800 words, suitable for FDA-regulated patient information materials."

Poor prompting results in vague outputs, missed regulatory requirements, and potentially non-compliant content that could compromise submissions or patient safety. Strategic prompting enables eVa AI to generate precise, regulation-compliant responses that meet pharmaceutical industry standards.

Dos and Don'ts

An effective prompt is key to unlocking accurate and relevant AI responses, and mastering the Dos and Don'ts ensures you communicate your intent clearly while avoiding common pitfalls. This section outlines essential best practices to help you craft precise, context-rich prompts that guide eVa AI to deliver optimal results.

Be clear and specific

  • State exactly what you need: "Generate a summary of Phase III trial results for {drug name}"

  • Include the desired format: "Create a bulleted list of..." or "Write a paragraph explaining..."

Provide rich context

  • Include relevant background: "For a regulatory submission to FDA..."

  • Specify your audience: "For HCPs treating oncology patients..."

  • Mention constraints: "Within 500 words and compliant with promotional guidelines"

Structure complex requests

  • Break multi-part requests into clear components:

    Task: Create educational content about {condition}
    ​Audience: Healthcare professionals
    ​Format: Email with 3 key messages
    ​Tone: Professional but accessible
    ​Requirements: Include efficacy data and safety profile

Request specific outputs

  • "Provide 3 examples of..."

  • "List the top 5 considerations for..."

  • "Generate a comparison table showing..."

Use industry-standard language

  • Leverage pharmaceutical terminology correctly

  • Reference specific regulations: "Following ICH guidelines..."

  • Mention relevant stakeholders: "For medical affairs teams..."

Avoid vague or ambiguous language

  • Don't use undefined pronouns ("it," "this," "that")

  • Avoid colloquialisms in regulatory contexts

  • Don't assume eVa AI understands implied pharmaceutical contexts

  • Avoid ambiguous medical terminology

  • Vague questions can lead to unclear or irrelevant response. Instead of "Tell me about treatments", use "Explain first-line treatment options for {disease name} in {patient type}"

Don't overwhelm with information

  • Avoid including irrelevant background information

  • Don't combine multiple unrelated requests

  • Avoid excessive technical jargon without context

  • Don't include outdated regulatory references

Don't ignore regulatory requirements

  • Never skip compliance considerations

  • Don't assume eVa AI knows current regulatory status

  • Avoid requesting non-compliant content

  • Don't ignore safety labeling requirements

Don't neglect quality control

  • Don't accept first-draft outputs without review

  • Avoid skipping medical/legal review processes

  • Don't use outputs without proper validation

  • Avoid bypassing established approval workflows

Don't assume context carryover

  • Don't rely on previous conversation context for critical details

  • Avoid referencing undefined studies or compounds

  • Don't assume eVa AI remembers previous regulatory discussions

  • Avoid incomplete therapeutic area specifications

Prompt template structure

For effective communication with eVa AI, we suggest you follow this prompt template structure:

[ROLE] + [CONTEXT] + [TASK] + [CONSTRAINTS] + [OUTPUT FORMAT]

Here's an example.

Role: "Acting as a {specific pharmaceutical role}..."

Context: "For a {therapeutic area} compound in {development phase} targeting {indication}..."

Task: "Create/Retrieve/Generate/Analyze {specific deliverable}..."

Constraints: "Following {regulatory guidelines} and ensuring {safety/compliance requirements}..."

Output: "Deliver in {specific format} with {structure/length requirements}..."

Natural prompting tips

You can write prompts using natural language. eVa AI also understands and overlooks minor mistakes to provide relevant and coherent responses.

To get accurate and helpful responses in the eVa AI conversational hub, besides the general guidelines, adhere to the following tips:

  • Include the word "suggest" in your prompt to increase the likelihood of retrieving information from Veeva Vault DAM.

  • You can enter a prompt in any language supported by the eWizard platform; however, eVa AI provides suggestions only in languages of the content stored in Veeva Vault DAM and your knowledge base.

  • Don't request data that is unrelated, inappropriate, harmful, or in violation of legal, ethical, or platform-specific guidelines.

Reference prompt library

Our library of reference prompts is a starting point for your own interactions with eVa AI. Each prompt is crafted to demonstrate clear intent, concise language, and optimal structure, and to support you in the text and image generation journey. By referencing these examples, you can quickly learn how to phrase your requests to achieve accurate, relevant, and actionable responses.

Text generation

Explore the following reference prompts you could use in requests to eVa AI for text generation:

  • Act as a {specific pharmaceutical role} with expertise in {required area}, and suggest some effective {disease treatment}.

  • Suggest quality of life & patient-reported outcome claims for {Product}.

  • Retrieve {expected result context}.

  • Follow FDA guidance for clinical protocols in {disease treatment} and generate {expected result}.

Image generation

Discover the following reference prompts tailored to generate compelling and visually striking images by eVa AI:

  • Suggest an image representing {image context}.

  • Find an image symbolizing {image context}.

  • Generate pictures of {image context}.

  • Suggest an image highlighting key benefits or advantages of {Product}.

Email generation

We recommend using one of the following prompts to eVa AI when generating interactive emails:

  • Turn this chat brief into an interactive email based on the selected template.

  • Generate an interactive email from the shared brief in the current chat.

  • Expand my outline into a professional interactive email.

  • Create a complete interactive email with the provided brief or texts. For an upcoming email, apply the selected template.

  • Compose a marketing interactive email using the brief below.

  • Create an invitation interactive email for the upcoming event.

  • Generate a professional interactive email to announce our new feature.

Did this answer your question?