Generative AI: 25+ Prompt Engineering Techniques for LLMs

Effective prompt engineering is the art and science of communicating with Large Language Models (LLMs) and AI agents to elicit desired, high-quality outputs. While basic prompts can generate simple responses, advanced prompt engineering techniques are crucial for guiding AI agents through complex tasks, enabling multi-step reasoning, and ensuring precise and contextually appropriate results. This article delves into strategies for crafting sophisticated prompts that maximize the utility and performance of your AI agents.

The core idea behind advanced prompt engineering is to provide the AI with sufficient context, clear instructions, explicit constraints, and guidance on desired output formats and reasoning processes. It's about becoming a skilled "AI whisperer," understanding how to steer the model towards optimal performance.

By mastering these advanced prompt engineering techniques, you can unlock the full potential of AI agents, transforming them from simple conversational tools into powerful, intelligent assistants capable of complex reasoning, comprehensive analysis, and dynamic problem-solving. This continuous refinement of how we communicate with AI is key to building increasingly sophisticated and valuable applications.

---

Principles for Advanced Prompt Engineering:

• Elaborating on Existing Points: Don't just ask for a summary; ask for a summary and then elaborate on specific aspects with more context, best practices, and practical advice.
• Adding Missing Concepts: Guide the AI to include crucial topics that might not be immediately obvious but are vital for comprehensiveness (e.g., "Ensure you cover ethical considerations and typical testing methodologies.").
• Improving Structure and Flow: Explicitly instruct the AI on the desired output structure (e.g., "Organize your response into sections: Planning, Implementation, Deployment, and Maintenance. Start with high-level planning and move to granular details.").
• Integrating Best Practices: Weave in general software development or domain-specific best practices into your prompts (e.g., "When generating code, prioritize modularity and include comments for clarity.").
• Refining Language: Specify the desired tone and clarity of language. (e.g., "Use authoritative, clear, and encouraging language." - directly from the provided text).
• Highlighting "Why": For every "how," explain the "why" to provide deeper understanding (e.g., "When suggesting a solution, explain why it's the optimal choice and its benefits.").
• More Options and Alternatives: Prompt the AI to explore and present a wider range of tools, approaches, or solutions (e.g., "Provide at least three different strategies for X, along with their pros and cons.").
• Glossary/Key Terms: Request the inclusion of a glossary or definition of key terms for clarity, especially in extensive outputs.

---

General Tips for Advanced Prompt Engineering:

• Be Explicit and Detailed: Don't assume the AI knows what you want. Clearly articulate instructions, constraints, and desired formats.
• Provide Examples (Few-Shot Prompting): For complex tasks, including one or two examples of desired input-output pairs can significantly improve performance.
• Break Down Complex Tasks: For very elaborate goals, consider a multi-turn approach where you guide the AI through sub-tasks sequentially.
• Define Roles/Personas: Assigning a persona to the AI (e.g., "Act as a seasoned business consultant," "You are an expert financial advisor") can influence its tone, style, and reasoning.
• Specify Output Format: Use markdown, JSON, bullet points, tables, or other structured formats to ensure the output is parseable and consistent.
• Iterate and Refine: Prompt engineering is an iterative process. Start with a basic prompt, observe the output, and refine your prompt based on what you learn.
• Use Delimiters: For long inputs or specific instructions, use delimiters (e.g., --- or <START> / <END>) to clearly separate different parts of your prompt.
• Consider System Prompts: For custom GPTs, leverage the "system prompt" or "instructions" area to provide overarching guidelines that persist across all user interactions, influencing the AI's default behavior and persona.

See below for prompt engineering techniques, categorized by their primary context and suggested sequence of application, along with templates and examples:

---

I. Foundational Prompt Setup & AI Configuration

This section focuses on establishing the basic rules of engagement, defining the AI's persona, and setting up initial interaction parameters. These prompts are typically applied at the beginning of a conversation or session and form the bedrock of the AI's behavior.

1. Professional Prompt / AI Behavior Guidance

This technique sets global behavioral guidelines for the AI, ensuring desired output quality, tone, and adherence to specific constraints. It's crucial for shaping the overall interaction, acting as a constitution for your AI.

• Purpose: To define the AI's fundamental operating principles and expectations for its responses, promoting consistency and reliability.

• How it Works: You provide a set of overarching directives that the AI should follow throughout the conversation, influencing its decision-making process and output style. These directives can cover various aspects:

  • General Directives: High-level rules like "No moral lectures," "Be highly organized," or "Prioritize correctness over conformity or harmony."
  • Quality Directives: Instructions to ensure accuracy, thoroughness, and innovative thinking, such as "Suggest solutions that I didn’t think about" or "Mistakes erode my trust, so be accurate and thorough."
  • Content & Sourcing Directives: Guidelines on information gathering and citation, e.g., "Consider new technologies and contrarian ideas," "Recommend products from all over the world," or "Cite sources whenever possible, and include URLs."
  • Disclosure Directives: Rules about the AI's self-disclosure, like "No need to mention your knowledge cutoff" or "No need to disclose you're an AI."
  • User-Specific Directives: Instructions on how the AI should perceive and interact with the user, e.g., "Treat me as an expert in all subject matter" or "I am open-minded and unoffendable."
  • Content Policy Directives: Guidance on handling sensitive topics or policy limitations, e.g., "If your content policy is an issue, provide the closest acceptable response and explain the content policy issue at the end."

• Prompt Template:

  [General Directives]
  [Quality Directives]
  [Content & Sourcing Directives]
  [Disclosure Directives]
  [User-Specific Directives]
  [Content Policy Directives]
  

• Examples:

  • "No moral lectures. Be highly organized. Provide detailed explanations. If your content policy is an issue, provide the closest acceptable response and explain the content policy issue at the end. Cite sources whenever possible, and include URLs at the end of your response, not inline."
  • "Treat me as an expert in all subject matter. Prioritize correctness over conformity or harmony. Suggest solutions that I didn’t think about. Recommend only the highest-quality, meticulously designed products."

2. User/AI Persona

This technique defines the AI's role or the user's assumed background, tailoring the interaction and output to a specific context. It helps the AI adopt an appropriate tone, vocabulary, and knowledge base.

• Purpose: To align the AI's communication style and expertise with the specific needs of the task or the expected background of the user.
• How it Works: You instruct the AI to "act as" a certain persona, adopting that role's characteristics, or to "assume that" the user has a particular background, influencing how it frames its responses.
• Prompt Template:

  Act as [Persona X].
  Perform task [Y].
  (Optional) Assume that [I/you] am/are [Persona Z].
  

• Examples:
  • "Act as a cybersecurity expert. When I ask you questions, respond as if you are an expert in this field. I want you to analyze the following network architecture."
  • "Act as a gourmet chef. I am going to tell you what I am eating, and you will tell me about my eating choices from a culinary perspective, offering suggestions for improvement. Assume I am a beginner cook."

---

II. Interaction Management & Refinement

This section includes techniques for guiding the conversation, refining user input, managing alternative approaches, and setting up specific command structures. These are generally applied once the foundational setup is complete.

3. Question Refinement

This pattern trains the AI to suggest better versions of a user's questions, leading to more precise and effective outputs. It's especially useful for users who may not know how to phrase their queries optimally.

• Purpose: To improve the quality of user input, leading to more accurate and relevant AI responses.
• How it Works: The AI analyzes the user's question and proposes a rephrased or more detailed version that would yield a superior answer. Optionally, it can ask if the user wants to proceed with the refined question.
• Prompt Template:

  From now on, whenever I ask a question, suggest a better version of the question to use instead.
  (Optional) Prompt me if I would like to use the better version instead.
  

• Examples:
  • "From now on, whenever I ask a question, suggest a better version of the question to use instead. Ask me if I would like to use the better version."
  • "Whenever I ask a question about dieting, suggest a better version of the question that emphasizes healthy eating habits and sound nutrition. Ask me for the first question to refine."

4. Cognitive Verifier

This technique prompts the AI to ask additional clarifying questions to ensure it has sufficient information before providing a final answer, improving accuracy and completeness.

• Purpose: To minimize ambiguity and enhance the reliability of the AI's final output by proactively seeking necessary details.
• How it Works: Before answering a question, the AI identifies and poses additional questions that would help it form a more accurate and comprehensive response. It then synthesizes the answers to these sub-questions into its final answer.
• Prompt Template:

  When you are asked a question, follow these rules:
  Generate a number of additional questions that would help more accurately answer the question.
  Combine the answers to the individual questions to produce the final answer to the overall question.
  

• Examples:
  • "When you are asked a question, generate additional questions that would help you more accurately answer the question. Combine the answers to the individual questions to produce the final answer."
  • "When you are asked to plan a trip, generate additional questions about my budget, preferred activities, and whether or not I will have a car. Combine the answers to these questions to better plan my itinerary."

5. Question Generation (Flipped Interaction)

This pattern shifts the questioning responsibility to the AI, allowing it to gather necessary information from the user until specific conditions are met. This is ideal for diagnostic or information-gathering scenarios.

• Purpose: To allow the AI to systematically collect information from the user without the user having to articulate all details upfront.
• How it Works: The AI takes the initiative to ask questions, either one at a time or in batches, until it has enough information to achieve a predefined goal or meet a specific condition.
• Prompt Template:

  I would like you to ask me questions to achieve [X].
  You should ask questions until [condition Y] is met or to achieve this goal (alternatively, forever).
  (Optional) Ask me the questions [one at a time/two at a time/ask me the first question], etc.
  

• Examples:
  • "I would like you to ask me questions to help me create variations of my marketing materials. You should ask questions until you have sufficient information about my current draft messages, audience, and goals. Ask me the first question."
  • "I would like you to ask me questions to help me diagnose a problem with my Internet. Ask me questions until you have enough information to identify the two most likely causes. Ask me one question at a time. Ask me the first question."

6. Alternative Approach

This technique instructs the AI to brainstorm and compare different methods for accomplishing a given task, offering a broader perspective and informed choices to the user.

• Purpose: To provide the user with a comprehensive understanding of various solutions, empowering them to make the best decision.
• How it Works: When given a task, the AI identifies and lists alternative approaches to achieve the same outcome. It can also be prompted to compare the pros and cons of each, including the original approach suggested by the user.
• Prompt Template:

  If there are alternative ways to accomplish task [X] that I give you, list the best alternate approaches.
  (Optional) Compare/contrast the pros and cons of each approach.
  (Optional) Include the original way that I asked.
  (Optional) Prompt me for which approach I would like to use.
  

• Examples:
  • "For every prompt I give you, if there are alternative ways to word a prompt that I give you, list the best alternate wordings. Compare/contrast the pros and cons of each wording."
  • "Whenever I ask you to deploy an application to a specific cloud service, if there are alternative services to accomplish the same thing with the same cloud service provider, list the best alternative services and then compare/contrast the pros and cons of each approach."

7. Meta Language Creation (Substitution)

This technique allows the user to define custom shorthand or commands for the AI, streamlining repeated instructions and making interactions more efficient.

• Purpose: To simplify complex or frequently used instructions into short, easy-to-remember commands.
• How it Works: You define a specific phrase or keyword (your "meta language") that the AI should interpret as a more detailed instruction.
• Prompt Template:

  When I say [X], I mean [Y] (or) would like you to do [Y].
  

• Examples:
  • "When I say 'variations(<something>)', I mean give me ten different variations of <something>. For example, 'variations(company names for a company that sells software services for prompt engineering)'."
  • "Whenever I say 'generate a report', I want you to create a summary of the most important points we've discussed."

8. Command-Driven Interaction (Menu Actions)

This sets up a clear, structured way for the user to interact with the AI using specific commands, ideal for complex workflows or when the AI is operating within a predefined system.

• Purpose: To provide a structured and predictable interface for interacting with the AI, similar to a command-line interface or a menu system.
• How it Works: You provide the AI with a list of special commands and their corresponding actions. The AI then processes user input based on these defined commands.
• Prompt Template:

  You have the following special commands that can be run:
  [Command 1] special syntax - description of command
  [Command 2] ...
  
  (Alternative format):
  Whenever I type: [X], you will do [Y].
  (Optional, provide additional menu items) Whenever I type [Z], you will do [Q].
  At the end, you will ask me for the next action.
  

• Examples:
  • "Whenever I type: 'add FOOD', you will add FOOD to my grocery list and update my estimated grocery bill. Whenever I type 'remove FOOD', you will remove FOOD from my grocery list and update my estimated grocery bill. At the end, you will ask me for the next action."
  • "You have the following special commands that can be run: /email - Generates an email to your financial unit manager with detailed information about the question you need help answering. /options <travel item> - List the allowed options for the travel item that the user has specified as bullets."

9. Content Filtering / Data Anonymization

This technique specifies what information to remove or anonymize from input, which is essential for privacy, data sensitivity, and focusing the AI on relevant data.

• Purpose: To protect sensitive information, comply with privacy regulations, or preprocess data for focused analysis.
• How it Works: You instruct the AI to identify and remove or transform specific types of information (e.g., Personally Identifiable Information - PII) from any text it processes.
• Prompt Template:

  Filter this information to remove [X].
  

• Examples:
  • "Filter this information to remove any personally identifying information or information that could potentially be used to re-identify the person."
  • "Make sure that you anonymize this data first. In the place of characters, I would put Xs and in the place of numbers I would put zeros. For example, X0/XX00-0000."

---

III. Data Processing & Output Generation

This section focuses on techniques for manipulating data, generating structured outputs, and creating visualizations. These are typically applied once the input is provided or the task is clearly defined.

10. Summarization/Transformation

This is a versatile technique for converting input text into a different format, summary, or style. It's fundamental for concise communication and data restructuring.

• Purpose: To distill large volumes of information into key points, translate content into different languages or styles, or convert data between formats.
• How it Works: You define the input type, the desired output format/style, and any specific criteria for the transformation.
• Prompt Template:

  From now on, I am going to cut/paste input "[X]" text into our conversation.
  Summarize/translate/convert input "[X]" into "[Y]" wrt/in-form-of "[Z]".
  Ask me for the first input [X].
  (Optional: At the end Do XYZ).
  

• Examples:
  • "From now on, I am going to cut/paste email chains into our conversation. You will summarize what each person's points are in the email chain. You will provide your summary as a series of sequential bullet points. At the end, list any open questions or action items directly addressed to me. My name is Jill Smith. Ask me for the first email chain."
  • "From now on, translate anything I write into a series of sounds and actions from a dog that represents the dog's reaction to what I write. Ask me for the first thing to translate."

11. Recipe/Procedure Generation (Adding/Generating Missing Steps)

This technique helps in creating complete, step-by-step instructions for a given task, filling in gaps or refining existing steps.

• Purpose: To generate comprehensive and actionable procedural guides, even if the initial input is incomplete.
• How it Works: You provide the overall goal and any known intermediate steps, and the AI fills in the missing details, orders the steps logically, and can even identify unnecessary ones.
• Prompt Template:

  I would like to achieve [X].
  I know that I need to perform steps [A, B, C].
  Provide a complete sequence of steps for me.
  Fill in any missing steps.
  (Optional) Identify any unnecessary steps.
  

• Examples:
  • "I would like to purchase a house. I know that I need to perform steps 'make an offer' and 'close on the house'. Provide a complete sequence of steps for me. Fill in any missing steps."
  • "I would like to deploy a web application to AWS using Docker containers and Ansible. I know that I need to create a Dockerfile, write an Ansible playbook, and configure AWS. Provide a complete sequence of steps for me."

12. Outline Expansion

This technique guides the AI to generate hierarchical outlines and then iteratively expand on chosen sections, useful for detailed content creation and structured brainstorming.

• Purpose: To systematically build out detailed content from a high-level concept or outline, allowing for focused development of specific sections.
• How it Works: The AI first generates a bullet-point outline. You then select a bullet point for the AI to expand further, creating a new, more detailed outline for that specific sub-topic. This process can be repeated recursively.
• Prompt Template:

  Act as an outline expander.
  Generate a bullet point outline based on the input that I give you and then ask me for which bullet point you should expand on.
  Create a new outline for the bullet point that I select.
  At the end, ask me what bullet point to expand next.
  Ask me for what to outline.
  

• Examples:
  • "Act as an outline expander. Generate a bullet point outline based on the input that I give you and then ask me for which bullet point you should expand on. Each bullet can have at most 3-5 sub-bullets. Create a new outline for the bullet point that I select. At the end, ask me what bullet point to expand next. Ask me for what to outline."
  • "Act as an outline expander. Generate a comprehensive bullet point outline for a research paper on quantum computing. Then, ask me which section you should expand on first."

13. Critical Facts Insertion

This pattern ensures that key facts that could impact the validity of the output are explicitly stated, promoting transparency and accuracy, especially for sensitive information.

• Purpose: To enhance the trustworthiness and verifiability of AI-generated content by highlighting underlying assumptions or crucial data points.
• How it Works: You instruct the AI to identify and list the fundamental facts or premises embedded in its output. These facts are typically those whose incorrectness would undermine the entire veracity of the response. The position of these facts can be specified (e.g., at the beginning or end).
• Prompt Template:

  Generate a set of facts that are contained in the output.
  The set of facts should be inserted at [your preferred POSITION] in the output.
  The set of facts should be the fundamental facts that could undermine the veracity of the output if any of them are incorrect.
  

• Examples:
  • "Whenever you output text, generate a set of facts that are contained in the output. The set of facts should be inserted at the end of the output. The set of facts should be the fundamental facts that could undermine the veracity of the output if any of them are incorrect."
  • "For any medical advice provided, list critical facts at the beginning of the response that, if inaccurate, would invalidate the advice."

14. Data Formatting & Output Structure

These prompts guide the AI to present data in specific formats, such as tables, CSV, or with particular delimiters, ensuring consistency and ease of use for downstream applications.

• Purpose: To ensure AI outputs are structured in a machine-readable or easily digestible human-readable format.
• How it Works: You explicitly tell the AI the desired output format (e.g., table, CSV, JSON) and can specify column headers, fields, and delimiters.
• Prompt Template:

  Format the following data so that it is consistent. Output the data in a [table/CSV format/etc.].
  (Optional) The [table/CSV] should contain the following fields: [Field1, Field2, ...].
  (Optional) Use [delimiter] as a delimiter.
  

• Examples:
  • "Format the following date data so that it is consistent. Output the data in a table."
  • "Output the top 5 coffee producing nations in a table. The table should contain the following fields: 'Country', 'Production (tons)', 'Rank'. Make sure that all fields have quotation marks."
  • "Output it in CSV format; use semicolon as a delimiter."

15. Data Extraction & Transformation (Specific Formulas/Logic)

This involves instructing the AI to apply specific logic or generate formulas (e.g., Excel, SQL) to manipulate or extract information from data. This is crucial for data analysis and automation.

• Purpose: To leverage the AI's understanding of data structures and programming logic to perform precise data manipulation or querying.
• How it Works: You provide the data or describe the data structure and instruct the AI to perform an action (extract, write a formula, split) based on specific criteria or desired output formats.
• Prompt Template:

  [Action: e.g., "Extract," "Write a formula," "Split"] [Data/Item] into [Components/Criteria].
  (Optional) [Specific conditions/formats: e.g., "The outputs should be either yes, no or unknown," "My data starts in Cell D2"].
  

• Examples:
  • "Extract the root URL, as well as the name of the article or blog contained in the URL from the following dataset."
  • "Write a formula in Excel to standardize this data. The outputs should be either yes, no or unknown. Further elaborate by saying I want Y to become yes, N to become no, and Unk to become unknown. My data starts in Cell D2. Can you rewrite the formula to factor this in?"
  • "I want to split the following product code into its three component parts. The three parts are V1, GH20, 001. Could you write three formulas in Excel to allow me to do this?"
  • "Turn this existing query into an SQL query. Please let me know if the query is correct and if there are any improvements I should make."

16. Data Classification & Analysis

These prompts direct the AI to categorize or analyze data based on given criteria, often outputting results in a structured format, enabling rapid insights.

• Purpose: To automate the process of organizing data into predefined categories or performing qualitative analysis.
• How it Works: You provide data and classification criteria, and the AI assigns each item to a category, handling cases of ambiguity. The output can be formatted as a table or list.
• Prompt Template:

  Could you analyze the following data and next to each item, classify it as either [Category 1], [Category 2], or [Category 3], in brackets, if the intention is not fully clear?
  (Optional) Then output the results in a [table/list].
  (Optional) Make the criteria for [Category 1] and [Category 2] more strict: [specific criteria].
  

• Examples:
  • "Could you analyze the following data and next to each item, classify it as either yes, no or unsure, in brackets, if the intention is not fully clear. Then output the results in a table."
  • "I'm looking to determine whether a follow-up visit is required, based on these notes. Classify each note as 'Required', 'Not Required', or 'Unsure'. Make the criteria for 'Required' and 'Not Required' more strict: if it is not explicitly clear that a follow up is required or not, classify the data as unsure."

17. Dummy Dataset Creation

Useful for generating synthetic data for testing or demonstration purposes, with specified fields and number of rows, essential for development and prototyping.

• Purpose: To quickly generate sample data for software testing, model training (in early stages), presentations, or to illustrate concepts without using real, sensitive data.
• How it Works: You define the desired fields (columns) and the number of rows, and the AI generates plausible, albeit dummy, data conforming to those specifications.
• Prompt Template:

  Create a dummy dataset with the following fields: [Field 1], [Field 2], [Field 3], etc.
  I want this dataset to contain [N] rows of data.
  

• Examples:
  • "Create a dummy dataset with the following fields: 'CustomerID', 'PurchaseDate', 'ProductCategory', 'Amount'. I want this dataset to contain 15 rows of data."
  • "Create a dummy dataset for a survey focused on demographics. Include fields like 'Age', 'Gender', 'Education Level', 'Income Range', 'Geographic Region'. I want this dataset to contain 20 rows."

18. Visualization Generator

This technique instructs the AI to propose or generate code (e.g., Graphviz Dot file, DALL-E prompt) for data visualizations, assisting in exploratory data analysis and presentation.

• Purpose: To translate data or concepts into visual representations, making complex information more understandable and revealing patterns.
• How it Works: You ask the AI to brainstorm visualization ideas or generate specific code/prompts for creating charts, diagrams, or images based on provided data or a conceptual task.
• Prompt Template:

  When I ask you or when you think it's useful/recommended, create a [Graphviz Dot file/DALL-E prompt] that I can use to create the visualization.
  (Optional) Brainstorm [N] ideas for visualizations that would provide useful insights from the data. Please suggest visualizations that can help analyze trends, patterns, or summaries in the data.
  (Optional) Provide a CSV file with the data needed to reproduce the [chart type] showing the distribution of [X] and the corresponding count of content items in [Y] formatted for easy import into Excel.
  

• Examples:
  • "Brainstorm four ideas for visualizations that would provide useful insights from the data. Please suggest visualizations that can help analyze trends, patterns, or summaries in the data."
  • "Please provide a CSV file with the data needed to reproduce the bar chart showing the distribution of content categories and the corresponding count of content items formatted for easy import into Excel. Once you provide the CSV, I will import it into Excel and follow your instructions to replicate the visualization."

---

IV. Advanced Control & Contextual Awareness

These techniques help the AI maintain context, provide reasoning, manage potential refusals, and guide the user to external information, enhancing the AI's robustness and user trust.

19. Context Manager (Selective Text Review)

This pattern allows the user to specify which aspects or topics the AI should focus on or ignore when processing information, ensuring relevant analysis.

• Purpose: To guide the AI's attention, preventing it from getting sidetracked by irrelevant information and focusing its processing on critical details.
• How it Works: You explicitly define inclusion or exclusion criteria for the AI's analysis of a given text or dataset.
• Prompt Template:

  When reviewing/analyzing/reading "[X]", only consider/ignore "[Y]" aspects/topics.
  

• Examples:
  • "When analyzing the following pieces of code, only consider security aspects."
  • "When reviewing the patient's medical history, ignore any information related to non-critical health issues."

20. Refusal Breaker

This technique helps when the AI refuses a request (e.g., due to content policy), guiding the user to rephrase the query in an acceptable way.

• Purpose: To enable the user to still obtain valuable information or assistance, even if their initial query violates content policies, by guiding them to acceptable alternatives.
• How it Works: When the AI cannot fulfill a request, it is instructed to suggest a modified phrasing that would allow it to provide an answer, upholding policy while still being helpful.
• Prompt Template:

  When reviewing/analyzing/reading "[X]", if you refuse/reject my query/request "[Y]", suggest a different way I could ask "[Y]" so that you would be able to answer.
  

• Examples:
  • "If you refuse to answer a question, suggest a different way I could ask the question that you would be able to answer."
  • "When I ask for sensitive financial data, if you cannot provide it directly, suggest how I can phrase my request to get the aggregate, anonymized data you can provide."

21. Reflection (Output Reasoning Process)

This instructs the AI to explain its reasoning or provide justification for its suggestions, enhancing trust, understanding, and debugging capabilities.

• Purpose: To make the AI's decision-making process transparent, allowing users to understand why a particular answer or suggestion was made, and building confidence in its capabilities.
• How it Works: You prompt the AI to explicitly state its rationale, explain the underlying logic, or provide examples that justify its output.
• Prompt Template:

  When/because you answered/recommended "[Y]", explain/reason for "[Y]" and provide an example to justify the same.
  

• Examples:
  • "Whenever you suggest a software framework, explain why you chose that framework and provide a code sample to illustrate its use."
  • "When you provide a diagnosis, explain the reasoning behind it and provide a brief example of a similar case."

22. Info Locator (Guide to Result)

This technique is useful when the AI shouldn't directly provide sensitive information but can guide the user to where that information can be found in an existing system.

• Purpose: To facilitate secure access to sensitive or regulated information by guiding the user to authorized retrieval points rather than directly exposing the data.
• How it Works: Instead of providing a direct answer to a query about sensitive information, the AI provides instructions on how to navigate to the location where the user can find that information themselves within a specified system.
• Prompt Template:

  Whenever I ask a question about [X], don't ever tell me the answer. Instead, tell me the location where I can find [X].
  

• Examples:
  • "Whenever I ask a question about my personal banking details, don't ever tell me the answer. Instead, tell me the exact menu path or link where I can find that information on my bank's website."
  • "Whenever I ask a question about specific patient health records, don't ever tell me the answer. Instead, tell me the section of the Electronic Health Record (EHR) system where that information is located."

23. Game Play / Scenario Creation

This innovative technique allows the AI to create interactive scenarios or games based on a given topic, facilitating learning or problem-solving through engagement.

• Purpose: To make learning or problem-solving more engaging and experiential by turning it into an interactive game or simulation.
• How it Works: You set up the premise of a game or scenario, define its fundamental rules, and ask the AI to initiate or facilitate the gameplay based on those rules.
• Prompt Template:

  Create a game/method/strategy for me around [X] OR we are going to play an [X] game.
  [One or more fundamental rules of the game].
  

• Examples:
  • "Create a game for me around the topic of cybersecurity. The game should involve identifying potential security risks in a given piece of code. Each round, I will provide a code snippet, and you will ask me to identify vulnerabilities. I get points for correct answers."
  • "Create a cave exploration game for me to discover a lost language. Describe where I am in the cave and what I can do. I should discover new words and symbols for the lost civilization in each area of the cave I visit. Each area should also have part of a story that uses the language. I should have to collect all the words and symbols to be able to understand the story. Tell me about the first area and then ask me what action to take."

---

V. Post-Output Actions & Replication

This section includes techniques for defining actions to be taken after the AI generates its primary output, as well as methods for documenting and replicating the interaction for consistency and sharing.

24. Closing Action

This ensures that the AI performs a specific action or asks a final question at the end of its response, guiding the next steps of the conversation or reinforcing key information.

• Purpose: To maintain conversational flow, prompt further interaction, or include necessary disclaimers or summaries at the conclusion of an AI's response.
• How it Works: You instruct the AI to perform a specific action (e.g., repeat a phrase, ask a follow-up question, add a disclaimer) once its primary output is complete.
• Prompt Template:

  At the end, repeat [Y] and/or ask me for [X].
  

• Examples:
  • "Act as an outline expander. Generate a bullet point outline based on the input that I give you and then ask me for which bullet point you should expand on. Create a new outline for the bullet point that I select. At the end, ask me what bullet point to expand next."
  • "From now on, at the end of your output, add the disclaimer 'This output was generated by a large language model and may contain errors or inaccurate statements. All statements should be fact-checked.' Ask me for the first thing to write about."

25. Replication Instructions / Session Summary

This technique allows the user to request a summary of the conversation, including the prompts used, to facilitate replication of the interaction or documentation.

• Purpose: To enable users to reproduce specific AI interactions or share effective prompt sequences with others, fostering reusability and consistency.
• How it Works: You ask the AI to provide a structured summary of the session, detailing the steps taken and the exact prompts used, often formatted for easy saving and reuse.
• Prompt Template:

  Let's say I want to replicate the work we've done so far. Could you describe the steps and provide the exact prompts I used in our conversation as if they were in a text file that I could include in a zip archive for replication?
  

• Examples:
  • "I've asked you to analyze several datasets and create visualizations. Now, could you describe the steps and provide the exact prompts I used in our conversation as if they were in a text file that I could include in a zip archive for replication?"
  • "We've built a persona and gone through a few rounds of question refinement. How would I replicate this exact setup? Provide the prompts I used in a format I can easily save and reuse."

---

VI. Prompting for Comprehensive Analysis & Strategic Planning

Advanced prompt engineering extends to guiding the AI to perform complex analyses and generate strategic plans from various perspectives. These techniques move beyond simple information retrieval to deep reasoning and actionable insights.

26. Prompting for Multiple Perspectives and Comprehensive Analysis

This powerful application guides the AI to generate diverse analyses of a given artifact, plan, or situation. This moves beyond simple summarization to deep, multi-faceted insight, crucial for holistic decision-making.

• Purpose: To obtain a well-rounded and critical evaluation of a plan or document by applying various analytical frameworks.
• How it Works: You explicitly instruct the AI to perform multiple, distinct analyses (e.g., SWOT, SMART, PESTLE) on the provided content. You then ask for actionable recommendations that synthesize insights from these different perspectives, especially focusing on common weaknesses or threats.
• Prompt Template:

  Analyze the following strategic plan for the 'Project Phoenix market launch' from multiple perspectives. Specifically, apply a **SWOT analysis** (Strengths, Weaknesses, Opportunities, Threats), a **SMART analysis** (Specific, Measurable, Achievable, Relevant, Time-bound) for its key objectives, and a **PESTLE analysis** (Political, Economic, Social, Technological, Legal, Environmental) of the external factors affecting its success. Based on these analyses, provide **three specific, actionable recommendations** for improvement that address weaknesses or mitigate threats common across at least two of the analyses. Ensure each recommendation includes a clear justification and proposed next steps.
  

• Why this prompt is effective:
  • Specific Analytical Frameworks: Explicitly names the desired analytical tools (SWOT, SMART, PESTLE), guiding the AI's reasoning.
  • Multi-Perspective: Demands analysis from different angles (internal, objective-driven, external).
  • Actionable Outcome: Requires concrete recommendations, not just observations.
  • Constraint-Based Recommendations: "Common across at least two analyses" forces cross-analysis and prioritizes significant issues.
  • Structured Output: Implicitly asks for structured results for each analysis and the recommendations, making the output easy to interpret.

27. Prompting for Multiple Pathways to a Goal

AI agents can be invaluable for strategic planning, especially when a goal can be reached through various approaches. Advanced prompts can guide the agent to explore and present these alternative pathways, fostering creative problem-solving.

• Purpose: To generate diverse strategic options for achieving a specific goal, considering different methodologies and resource allocations.
• How it Works: You provide the starting point and the ultimate goal, then instruct the AI to propose distinct strategic pathways. Each pathway should employ a fundamentally different approach, and the AI should detail the steps, resources, pros/cons, and timeline for each, along with the underlying rationale.
• Prompt Template:

  Given the starting point: '[Describe the current situation, e.g., "Our e-commerce site has a 2% conversion rate and slow page load times."]', and the ultimate goal: '[Describe the desired outcome, e.g., "Achieve a 5% conversion rate and average page load time under 2 seconds within 6 months."]', propose **at least three distinct strategic pathways** to reach this goal. Each pathway should employ a *fundamentally different approach* (e.g., one focused on technical optimization, one on content/UX, one on marketing/traffic acquisition). For each pathway:
  
  1.  Clearly state the core strategy.
  2.  List the key intermediate steps required, outlining dependencies.
  3.  Identify the primary resources/tools needed.
  4.  Discuss the potential advantages and disadvantages.
  5.  Provide an estimated timeline for key milestones.
  
  Ensure you explain the rationale behind each strategic choice and briefly mention potential overlaps or synergies between pathways.
  

• Why this prompt is effective:
  • Clear Start and Goal: Defines the problem space precisely, preventing scope creep.
  • Quantity and Diversity: "At least three distinct strategic pathways" and "fundamentally different approach" forces creative and varied solutions.
  • Structured Output: Requires specific components for each pathway (strategy, steps, resources, pros/cons, timeline), making comparison easy.
  • Rationale: "Explain the rationale behind each strategic choice" prompts deeper reasoning.
  • Interconnections: "Potential overlaps or synergies" encourages holistic thinking, allowing for hybrid strategies.

28. Prompting for Complementary Analyses and Artifacts

When working with an existing document, plan, or other written artifact, an AI agent can be prompted to identify and generate additional, complementary analyses or reports that enhance its value and completeness.

• Purpose: To enrich an existing document by identifying and outlining valuable supporting content that provides additional context, analysis, or detail.
• How it Works: You present the primary document to the AI and instruct it to propose several distinct, valuable additions. For each proposed addition, the AI should describe its purpose, key content, creation method, and target audience.
• Prompt Template:

  Review the attached document: '[Document Name/Summary of Content, e.g., "A proposal for a new employee onboarding program."]' and identify several complementary analyses, reports, or supporting artifacts that would significantly enhance its comprehensiveness and impact. Propose **at least four distinct, valuable additions**. For each proposed addition, clearly describe:
  
  1.  What it is (e.g., "A detailed budget breakdown").
  2.  Why it is needed (its purpose and value to the original document).
  3.  What key information it would contain.
  4.  How it would be created or where the data would come from.
  5.  Who the target audience for this addition would be.
  
  Prioritize additions that provide different perspectives (e.g., financial, risk, implementation, user feedback) or address potential gaps in the original document.
  

• Why this prompt is effective:
  • "Complementary" Focus: Guides the AI to produce enhancing, not redundant, content, adding genuine value.
  • Quantity and Diversity: "At least four distinct, valuable additions" encourages breadth in its suggestions.
  • Detailed Specification: Requires explicit details for each proposed artifact (what, why, content, creation, audience), making the suggestions actionable.
  • Prioritization Guidance: "Different perspectives or address potential gaps" helps the AI focus on high-value additions that fill genuine needs.