Prompt Engineering

Prompt engineering is how you communicate with AI models to get useful, accurate results. These techniques work across all models on campusGenAI.

Core principles

Be specific

Vague prompts produce vague results. Adding context, constraints, and a clear goal dramatically improves output quality.

Instead of	Try
”Write me an email"	"Write a professional email to our program officer updating them on Q1 enrollment (up 15%) and requesting a meeting to discuss next quarter’s targets"
"Summarize this report"	"Summarize the three main recommendations in this report and list the evidence supporting each one"
"Help me with this proposal"	"Review this grant proposal abstract for clarity and strength of argument. Identify the weakest section and suggest specific improvements”

Provide context

Tell the AI what it needs to know to do the task well:

• Who the audience is — a funder, a board, a colleague, a client
• What tone you want — formal, conversational, plain language
• What format you need — bullet points, a table, prose paragraphs, a specific length
• Any constraints — word limits, things to include or avoid

I'm writing a board report for a non-profit board that has limited
technical background. Summarize our Q1 AI usage data (attached) in
plain language. Focus on: how many staff are using the tools, what
kinds of tasks they're using AI for, and one specific example of
time saved. Keep it to 3 short paragraphs.

Define the output format

If you have a specific structure in mind, say so explicitly:

Create a comparison table of the three grant opportunities listed
below. Columns: funder name, deadline, maximum award, eligibility
requirements, and whether we qualify based on our org profile.

Iterate

Your first prompt is rarely your last. Follow up with:

• “Make this shorter”
• “Add more detail about the budget section”
• “Change the tone to be less formal”
• “Give me three alternative versions of the opening paragraph”

Think of it as a conversation, not a single command.

Patterns that work well

Role assignment

Giving the AI a role can sharpen its responses significantly:

You are an experienced grant reviewer for community health foundations.
Review the following abstract and identify three strengths and three
weaknesses from a funder's perspective.

Step-by-step reasoning

For complex analysis or calculations, ask the AI to show its work:

Walk me through analyzing this budget gap step by step.
Show your reasoning at each stage.

Few-shot examples

Show the AI what you want by providing examples before the task:

Rewrite these sentences in active voice:

Original: "The program was delivered by our team."
Rewritten: "Our team delivered the program."

Original: "Outcomes were measured by program staff."
Rewritten: "Program staff measured outcomes."

Now rewrite: "The evaluation was completed by our external evaluator."

Structured extraction

Good for pulling specific information out of documents:

From the attached meeting transcript, extract:
1. All decisions made (as a bullet list)
2. All action items (person responsible + what they'll do)
3. Any open questions that need follow-up

Use cases by role

Non-profit and community organizations

• Draft grant LOIs and proposals from program descriptions
• Turn program data into board reports and funder updates
• Create plain-language summaries of policy or regulatory documents
• Develop FAQ documents for your community from internal resources
• Prepare talking points for advocacy meetings or funder presentations
• Draft intake forms, referral letters, and follow-up communications

Faculty and researchers

• Generate discussion questions and assignment prompts from course materials
• Summarize research papers and identify themes across multiple sources
• Draft and refine grant abstracts
• Create rubrics and assessment criteria
• Get feedback on manuscript sections

Administrators and staff

• Summarize meeting notes and extract action items
• Draft internal communications, announcements, and policy updates
• Analyze open-ended survey responses for themes
• Create onboarding materials and process documentation
• Build templates for recurring communications

What AI won’t do reliably

• Accurate citations — AI can fabricate plausible-sounding references. Always verify citations independently.
• Current information — Models have knowledge cutoffs. Don’t rely on them for current events, new regulations, or recent research without verification.
• Precise numbers — AI is not a calculator. Use it to structure analysis, not to do arithmetic.
• Confidential research — Don’t paste sensitive personal information, student records, or health data into AI conversations unless your institution has specifically approved it.

---

Need help?

• Get help: Contact us
• About campusGenAI: Main site

Your interface may look slightly different depending on your institution’s deployment.