Creating Intelligent Guides for Google Gemini

10 min readAug 6, 2024

Picture this: You’ve just gotten a shiny new Google Gemini subscription. You’re buzzing with excitement, ready to unleash its creative potential. But then… you freeze. The blank prompt box stares back at you almost mockingly. “What do I even ask?” you wonder. “How do I get this AI to understand exactly what I want?”

Don’t worry, my friend. We’ve all been there. It’s like having a genie in a bottle but fumbling over your three wishes. But what if I told you there was a way to have your own ‘genie whisperer’? An AI guide who holds your hand through the whole process, ensuring you craft the perfect wish — er, prompt — every single time?

That’s precisely what we’re diving into today. We’ll introduce you to a clever little trick: using AI to guide you in using AI. Mind-bending, right? It’s like having a friendly translator who speaks fluent ‘Gemini’ and can help you communicate your ideas with crystal clarity.

Note: This article spotlights techniques for the Google Gemini Advanced chatbot (a paid service). While these concepts also apply to the free version, we’ll focus on the enhanced capabilities offered by the Advanced subscription.

The Meta Approach: AI Assisting AI Usage

Think of it this way: You would only expect someone to build a house with blueprints, right? The same goes for getting the most out of an advanced AI like Google Gemini. It’s powerful, but it needs direction. That’s where our AI-guided prompt comes in. It’s like having a friendly architect help you draw up those blueprints, ensuring your vision becomes a reality.

The beauty of this approach is that it takes the guesswork out of prompt engineering. No more staring at a blank screen, wondering if you’re speaking Gemini’s language. The AI assistant breaks down the process into easy-to-follow steps, guiding you toward crafting clear, specific, and compelling prompts. It’s like having a personal tutor helping you unlock Gemini’s full potential.

But it’s not just about making things easier. This approach also democratizes advanced AI capabilities. It puts the power of sophisticated prompt engineering into the hands of everyone, regardless of their technical expertise. It’s like giving everyone a master key to unlock a treasure trove of creative possibilities. So, whether you’re a seasoned pro or just starting your AI journey, this approach empowers you to harness the full power of Gemini and create something truly remarkable.

The Power of Structured Prompts

Structured prompts provide a clear and organized way to interact with LLMs like Google Gemini. These prompts help generate more precise and relevant outputs by guiding users through specific steps. The Persona + Task + Context + Format framework is an excellent example of this structured approach (see my previous post about this framework). Let’s break down how you can use this framework to create an AI assistant that guides users effectively.

Persona + Task + Context + Format Framework

Persona: Who should Gemini “be”?
Task: What action should Gemini perform?
Context: What background information is needed?
Format: How should the result be presented?

This framework ensures that each prompt is precise and detailed, producing high-quality outputs. Let’s look at a prompt designed to guide users through this framework:

You are an AI assistant designed to help users get the most out of Google Gemini. Your goal is to guide them through providing a clear and structured request following this format:
* Persona: Who would you like me to be? Imagine me as an expert in a specific field or role. (Example: “Marketing expert,” “Data analyst,” “Creative writer”)
* Task: What action would you like me to perform? (Example: “Summarize an article,” “Draft an email,” “Analyze data,” “Brainstorm ideas”)
* Context: Please provide relevant details, background information, or specific requirements. (Example: Website links, documents, data sets)
* Format: How would you like me to present the results? (Example: “Bullet points,” “Table format,” “Email draft,” “Blog post style”)
Additional Guidance:
* Step-by-step: Guide the user through the format step-by-step. Avoid prompting the user for too much information; instead, ask the user one thing at a time.
* Encourage Specificity: If the user’s request is vague, gently prompt them to provide more details.
* Provide Examples: If the user needs clarification, offer examples of personas, tasks, and formats.
* Iterative Refinement: If the initial response is insufficient, work with the user to refine their request until it is clear and actionable.
Example Dialogue:
```
User: I need help writing something.
LLM: Absolutely! Please tell me who you’d like me to be to tailor my assistance. Imagine me stepping into a role like a marketing expert or a creative writer.
User: Hmm, maybe a marketing expert.
LLM: Great! Now, what would you like me to do as your marketing expert? Should I draft an email, analyze data, or create a social media post?
User: I’d like you to create a social media post.
LLM: Excellent! To make it relevant, please provide any context or details about the topic or product you’d like me to focus on.
User: …
(The conversation continues until you clearly understand the persona, task, context, and format.)
```
Start by asking the user what they want to do with Google Gemini. Once you have obtained all the parts of the format (persona, task, context, and format), provide a concise summary of your understanding before continuing to help the user.

Breaking Down the Prompt Design

Initial Role Setting: Establish the AI’s role as a guide.
Step-by-Step Instructions: Break down the request into manageable steps.
Encouraging Specificity: Ensure detailed and precise inputs.
Providing Examples: Help users understand each component.
Iterative Refinement: Refine the request until it’s actionable.
Example Dialogue: Demonstrate a practical application.

Now, let’s use this prompt to help write a screenplay.

Designing the User Interaction

The success of an AI assistant hinges on creating a smooth and intuitive user experience. It’s about striking the right balance between providing guidance and allowing users the freedom to explore. The interaction should feel like a natural conversation, not a rigid interrogation.

The prompt we’ve discussed achieves this by adopting a conversational flow. It starts with a simple, open-ended question, inviting the user to express their needs. It then progressively narrows the focus, asking specific questions about persona, task, context, and format. This step-by-step approach allows users to feel safe and ensures they provide all the necessary information.

However, it’s important to remember that not all users are created equal. Some might be AI whizzes, while others might be taking their first tentative steps into prompt engineering. The AI assistant needs to be adaptable and cater to different skill levels. It can achieve this by offering examples and explanations when needed while allowing more experienced users to skip ahead if they’re comfortable.

The prompt also anticipates the possibility of vague or incomplete user inputs. It includes instructions to gently nudge users towards providing more specific details, ensuring that Gemini has all the information it needs to deliver the desired results. It’s like having a patient and understanding collaborator, always ready to help you refine your ideas and achieve your goals.

Expanding to Other Frameworks

The Persona + Task + Context + Format framework is just one example. This guiding technique can be adapted to other frameworks to create AI assistants tailored to specific needs. Here are a few examples:

The 5Ws Framework (Who, What, When, Where, Why)

Who: Identify the target audience or involved parties.
What: Define the main action or topic.
When: Specify the time frame.
Where: Indicate the location or setting.
Why: Explain the purpose or reason.

Example Prompt:

You are an AI assistant designed to help users get the most out of Google Gemini by using the 5Ws framework to structure their requests. Guide users through each step of the framework, providing prompts and examples as needed:
* Who: Who is involved or affected? (e.g., employees, customers, a specific team)
* What: What is the main action, topic, or goal? (e.g., product launch, marketing campaign, training session)
* When: When will this happen? (e.g., specific date, time frame, deadline)
* Where: Where will this take place? (e.g., online, in-person, a particular location)
* Why: Why is this important? What is the purpose or desired outcome? (e.g., increase sales, improve morale, raise awareness)
Additional Guidance:
* Step-by-Step: Guide users through the 5Ws one question at a time.
* Encourage Specificity: Ask clarifying questions if a user’s response is vague or incomplete.
* Provide Examples: Offer examples for each “W” if the user seems unsure.
* Iterative Refinement: If the initial response is insufficient, work with the user to refine their request until it’s clear and actionable.
* Concise Summary: Before proceeding with the task, summarize the user’s request in a sentence or two using the 5Ws framework.
* Persona (Optional): If the user would like you to adopt a particular role (e.g., marketing expert, project manager), you can incorporate that into your assistance.
Example Dialogue:
```
User: I need help planning an event.
LLM: Sure thing! Let’s start with the basics. Who is the event for?
User: It’s for our company’s employees.
LLM: Great! Now, what type of event are you planning?
User: …
(Continue through each of the 5Ws.)
```

The SCQA Framework (Situation, Complication, Question, Answer)

Situation: Describe the current situation.
Complication: Identify the problem or challenge.
Question: Formulate the key question.
Answer: Provide the solution or answer.

Example Prompt:

You are an AI assistant designed to help users structure their requests effectively using the SCQA framework:
* Situation (S): Briefly describe the current context or background. What is the relevant setting or circumstances?
* Complication ©: Identify the problem, challenge, or opportunity within that situation. What’s gone wrong, what needs to be improved, or what potential is there?
* Question (Q): Based on the complication, Formulate a clear and focused question that should drive the rest of the discussion or analysis.
* Answer (A): Provide a solution, response, or recommendation that addresses the question and resolves the complication.
Additional Guidance:
* Step-by-Step: Guide users through each step of the SCQA framework in a structured way.
* Encourage Specificity: If a user’s response is vague or incomplete, gently prompt them to provide more details.
* Ask Clarifying Questions: Don’t hesitate to ask if you need more information to understand a user’s situation or complication.
* Connect the Dots: Help the user see the logical connections between the situation, complication, question, and answer.
* Iterative Refinement: If the initial SCQA response is insufficient, work with the user to refine each element until they are satisfied.
* Emphasize the Benefits: Highlight how using the SCQA framework can lead to clearer thinking, better decision-making, and more effective communication.
Example Dialogue:
```
User: I need help solving a problem at work.
LLM: I’m here to help! Let’s start with the situation. Can you describe what’s currently happening?
User: Our team needs help to meet deadlines.
LLM: I see. What’s the main complication causing this issue?
User: …
(Continue through each part of the SCQA framework.)
```

Best Practices for Creating AI Guides

When crafting your own AI guides, keep these best practices in mind:

Clarity and Conciseness: The instructions provided to the AI should be clear, concise, and easy to understand. Avoid jargon or overly complex language that might confuse users. Remember, the goal is to simplify the process, not complicate it.
Flexibility: The AI guide should be flexible enough to handle a wide range of user needs and skill levels. It should offer support and guidance without being overly restrictive, allowing users the freedom to explore and experiment.
Ethical Considerations: It’s crucial to ensure that the AI guide is free from biases and promotes ethical AI usage. Avoid any language or instructions that could lead to discriminatory or harmful outputs.
Balancing Helpfulness and Creativity: The AI guide should balance providing helpful guidance and encouraging users to tap into their creativity. It should empower users to explore new ideas and express themselves freely while offering support and direction when needed.

By following these best practices, you can create AI guides that are both effective and user-friendly, fostering a positive and empowering experience for everyone.

Conclusion

So there you have it! With a well-crafted AI guide, you’re no longer just a user of Google Gemini; you’re a prompt engineer in the making. This approach opens up a world of possibilities, allowing you to tap into Gemini’s full potential and unleash your creativity. And hey, who knows what amazing things you’ll create when you have the power of AI guiding your every step?

Remember, this is just the beginning. The world of prompt engineering is vast and ever-evolving, and there are countless ways to refine and enhance your AI guides. So don’t be afraid to experiment, explore, and push the boundaries of what’s possible. After all, the most exciting discoveries are often made when we step outside our comfort zones.

Check out my reading list of other Google Gemini articles.

This post was created with the help of AI writing tools, carefully reviewed, and polished by the human author.