Learning Platform
Artificial Intelligence10 min read

A Practical Guide to Effective Prompting

August 21, 2026Download
Table of Contents
Documentation

Generative AI can be a useful helper for everyday work, from drafting emails and proposals to summarizing documents, brainstorming ideas, and organizing information. When used well, AI can save time on repetitive or time-consuming tasks, so you have more space to focus on the work that matters most: making decisions, building relationships, supporting communities, and applying your own judgment, experience, and expertise.

This guide offers a balanced introduction to prompting, the instructions you give AI tools. While these tools can improve efficiency and confidence, they can also produce inaccurate information, reflect bias, raise privacy concerns, and fall short when cultural understanding, lived experience, or human leadership are needed. The goal is to help you use AI thoughtfully, safely, and practically. In the right situations, it can be used as a support, not a replacement for your own knowledge and decision-making.

What is Generative AI?

Generative AI is a type of artificial intelligence that can create original-looking content such as text, images, summaries, plans, and code. It works by learning patterns from large amounts of data and using those patterns to predict what is likely to come next based on the prompt you provide.

At the same time, AI does not think, understand, or make judgments in the way people do. It lacks lived experience, cultural understanding, wisdom, and intent. Also, it cannot guarantee that the information it produces is accurate. AI can generate responses that sound confident but are incomplete, biased, or incorrect, so important facts should always be verified independently.

When Not to Use AI

Use extra caution, or avoid AI entirely, when the task requires legal, medical, financial, or professional judgment; involves sensitive, sacred, personal, or confidential information; or depends on emotional intelligence and human leadership, such as conflict resolution, performance reviews, or sensitive communications.

Never enter private community knowledge, sacred cultural information, personal data, or content involving ceremonies, language, regalia, or community members unless you are certain it is appropriate and permitted.

A helpful rule of thumb is: use AI to accelerate your work, not replace your judgment.

You remain responsible for the final output. AI should support your thinking, save time on routine tasks, and help you focus on the work that matters most, while decisions, accountability, and care stay with people.

What is a Prompt?

  • A prompt is the instruction you give an AI tool or app. It ranges from a simple question to detailed multi-paragraph specifications with examples and formatting requirements.
  • Think of a prompt as a brief to a new colleague: clearer instructions yield better results.

The difference between vague and well-constructed prompts is the difference between generic responses and precise, actionable outputs that save real time.

Prompting Foundations

  • AI models don't read minds. They respond based entirely on your input. A well-structured prompt means the difference between generic paragraphs and tailored recommendations, inaccurate summaries and precise ones, ten minutes of refinement and getting it right the first time.
  • Prompting isn't about memorizing tricks. It's about developing a repeatable habit of giving clear, complete instructions. It’s a skill that improves with practice and translates into better, faster work.

The Four Core Elements for Prompting

Every effective prompt combines one or more of these elements. Understanding them helps you construct better requests consistently. Not every prompt requires all four elements, but knowing when and how to use each dramatically improves your results.

Element 1: Instruction

The instruction is your core directive. It’s the task you want performed. It should be clear, specific, and action-oriented using direct verbs.

Weak instructions

Strong instructions

"Communications plan"

"Write a one-page communications plan for a community job fair taking place in two weeks."

"Help with email"

"Draft a professional follow-up email to a vendor who hasn't responded in two weeks."

Pro tip: Start with action verbs like Summarize, List, Compare, Draft, Explain, Classify, Rewrite, and Analyze.

Element 2: Context

Context provides the background information AI needs for relevant responses. It creates parameters that are specific, actionable, and targeted. Without it, models default to generic assumptions.

Without context

With context

"Write a project update."

This is too vague—what project? For whom? What's the status?

"Write a project update for the implementation of a new client management system. The project is slightly behind schedule because staff needed more time for onboarding and data migration. The audience is the CEO and CTO. Include current status, completed tasks, delays, risks, next steps, and any decisions needed from leadership."

Context answers the implicit questions: Who is this for? What's the situation? What constraints exist?

Element 3: Input Data

Input data is the specific material you want processed. It’s the text to summarize, data to analyze, lists to sort, or documents to review.

Best practices:

  • Clearly separate data from instructions using labels or line breaks
  • Use delimiters (quotes, dashes) to maintain clarity
  • Help AI distinguish between what to do and what to do it to
  • Place data after instructions for better processing

Element 4: Output Data

Output indicators specify how you want responses formatted. Its length, structure, tone, or other constraints on the deliverable.

Format control

“Respond in a numbered list" produces organized structure.

Tone setting

"Use professional tone for a presentation" sets the tone and style.

Length control

"Keep response under 100 words" manages brevity.

Structure definition

Define a precise format like:

  • "Present as a table with columns for Name, Role, Status"
  • "List the steps in numbered order."
  • "Present your response as a bulleted list."
  • "Write a single paragraph of no more than 100 words."
  • "Format the response as a professional email with a subject line, greeting, body, and sign-off."
Effective Prompt Example

Compare Gemini, ChatGPT, and Microsoft Copilot for use by an Indigenous-owned business. Include features, best use cases, pricing tiers, workplace integrations, data security, privacy, admin controls, and key limitations.

Present the comparison as a table with columns for Tool Name, Main Features, Best For, Integrations, Data Security/Privacy, Key Limitations, and Suitability for Indigenous Organizations.

Clearly state where features depend on subscription type, region, or admin settings. Then provide a 150-word recommendation for which tool is most suitable for a small organization that is Microsoft 365-based and handles sensitive community data.

Working with Documents and Larger Content

How to feed the right information into any AI tool: Paste, attach, or focus. The right approach depends on the content.

Short content

“Copy text from your document, paste it in, and ask your question.”

Longer documents

“Attach the file directly. Most tools (Copilot, ChatGPT, Gemini, Claude) support Word, PDF, and text files.”

Very long documents

“Direct the AI to specific sections or pages rather than asking it to process the entire document at once.”

What Good Prompts Accomplish

  • Reduces ambiguity: AI doesn't guess—you've been explicit about needs
  • Sets clear expectations: Format, length, tone, and depth stated upfront
  • Provides sufficient context: Model has background to produce relevance
  • Focuses on the task: One clearly defined deliverable, not open-ended fishing
  • Enables a useful first draft: Even if refinement is needed, the initial response is workable

Prompting Methods

Few-Shot Prompting

Include examples of desired input-output patterns before presenting your actual task. Examples teach the model your expected format and style.

Example

When to use:

  • The task requires specific formatting
  • You need consistent outputs

Tips for success:

  • Use 2-5 examples to establish a pattern
  • Keep format consistent across examples
  • Choose representative inputs
  • Place the actual request last

Role or Persona Prompting

Assigning the AI a specific role, profession, or persona before giving it a task. This frames the model's response within the perspective and expertise of that role, drawing on patterns associated with that role's knowledge, vocabulary, and priorities.

Example prompts

Why it works:

A prompt framed with a role or persona will produce more domain-relevant output than the same question asked without a role. The role provides implicit context that guides both content and tone.

Tips for success:

  • Be specific about the role's seniority or specialization, when relevant
  • Combine role prompting with other elements, like instruction and output format, for the best results
  • Use role prompting to control tone and register as well as content

Context Engineering

Adding relevant background information to your prompt so the AI can produce a response that is tailored to your specific situation rather than a generic answer. Context eliminates guesswork. The more relevant information you provide, the less the AI has to fill in with generic assumptions.

Example

Why it works:

This produces a targeted, relevant proposal that requires minimal editing.

Chain-of-Thought Prompting

Instructing the AI to work through a problem step by step rather than jumping directly to a final answer. This technique significantly improves accuracy on tasks that involve reasoning, calculation, or multi-part analysis. When the AI is asked to reason explicitly, it is less likely to skip steps or make errors.

Example

Interview-Style Prompting

Interview-style prompting is a way of prompting an AI model in which the model asks you targeted questions before producing a final answer. It is useful when you have a goal but need help clarifying the context, audience, constraints, or details first.

Example

Further Reading

Artificial Intelligence

AI Readiness: Making Responsible Decisions Before Adoption

  • Document
Artificial Intelligence

How to Evaluate AI Tools

  • Document
Artificial Intelligence

How Different Types of AI Work

  • Document
Artificial Intelligence

What is AI? What Can it Do?

  • Document