Do You Suck at Ideating? Prompt Strategies for Ideation With LLMs Will Save You

I’ve talked about different aspects of the content creation process such us:

• Repurposing

• Automation

• Generation

But…

What about where it all starts?

The ideation stage.

The ideation stage is crucial because it lays the foundation for the entire content creation process—Strong ideas guide repurposing, automation, and generation to create content that resonates.

Effective ideation has become increasingly important as the internet continues to be flooded with more content.

You need to consistently generate fresh, compelling ideas to capture attention.

But how do we become effective ideation machines?

AI is here to help.

Today we’re going to explore

1. The world of ideation with LLMs: covering prompting strategies, knowledge bases, and tools to supercharge your ideation process.

2. Real-world examples & case studies across industries: providing valuable insights into integrating LLMs into your workflows.

You’ll better appreciate ideation and the purpose it serves, and you’ll walk away with some practical strategies to get ideating with LLMs yourself!

Let’s jump in!

LLMs & Ideation

So how can LLMs help with ideation?

LLMs can help with:

Faster iteration times— quickly generate multiple variations and refinements of ideas, allowing for rapid exploration and development of concepts.

Synthesis of differing ideas—processing and combining ideas from various sources and perspectives to come up with solutions that might not have been considered otherwise.

Breaking creative blocks—can provide fresh perspectives, prompts, and inspiration to help overcome barriers and stimulate new ideas.

Expanding knowledge domains—can help ideators explore unfamiliar territories and make unexpected connections, leading to more diverse and innovative ideas.

Let’s talk about each of these areas and how you can start using LLMs with them today.

Iteration Ideation

Let’s say you want to get to an idea as fast as possible.

You aren’t sure what idea will hit but you know seeing more ideas will help you land on a final one sooner.

LLMs have the ability to generate multiple variations of an idea quickly.

To iterate quickly over ideas, we can use the following workflow steps:

1. Initial Idea Generation

• Provide a clear and concise prompt describing the core idea or theme you want to explore

• Ask the LLM to generate a set number of initial idea variations (e.g., "Generate 5 different ideas for a resource guide on repurposing newsletters into tweets")

2. Idea Evaluation and Selection

• Review the generated idea variations and select the most promising ones

• Provide feedback to the LLM on why you selected those ideas (e.g., "I like ideas 2 and 4 because they solve a clear problem and have unique features")

3. Idea Refinement

• Ask the LLM to refine and expand upon the selected ideas

• Provide specific aspects you want the LLM to focus on (e.g., "For idea 2, generate 3 variations that emphasize ease of use and social sharing")

4. Iteration

• Repeat steps 2 and 3, evaluating and refining the ideas further

• Continue iterating until you have a set of well-developed, promising ideas

5. Final Selection

• Choose the best idea(s) from the refined variations

• Ask the LLM to provide a detailed summary of the selected idea(s), including key features, benefits, and potential implementation steps

Example of initial prompt:

Generate 5 different ideas for a resource guide on repurposing newsletters into tweets. Provide a brief description of each resource guide, focusing on its unique features and target audience.

Idea Synthesis

Idea synthesis is the process of combining ideas from different sources and domains to create novel solutions.

LLMs can help you make connections between seemingly unrelated concepts, leading to innovative ideas.

To synthesize ideas using LLMs, you can follow these steps:

1. Collect Ideas from Different Domains

• Identify the key domains or areas related to your problem or topic

• Gather ideas, concepts, and examples from each domain

• Create a list of these ideas to use as input for the LLM

2. Prompt the LLM to Make Connections

• Ask the LLM to find commonalities or connections between the ideas from different domains

• Encourage the LLM to combine ideas in unique ways to create new solutions

• Provide an example of idea synthesis to guide the LLM

3. Evaluate and Refine the Synthesized Ideas

• Review the generated idea combinations and select the most promising ones

• Ask the LLM to expand on the selected ideas, providing more details and potential applications

• Refine the ideas further by asking the LLM to address any limitations or challenges

4. Iterate and Expand

• Repeat steps 2 and 3, using the refined ideas as input for further synthesis

• Encourage the LLM to bring in additional ideas from other domains to expand the solution space

• Continue iterating until you have a set of well-developed, innovative ideas

Example of idea synthesis prompt:

We have the following ideas from different domains:

1. Domain: Psychology - Idea: Positive reinforcement 

2. Domain: Technology - Idea: Gamification

3. Domain: Education - Idea: Microlearning

Find connections between these ideas and suggest 3 novel solutions that combine elements from each domain to create an engaging learning experience.

By prompting LLMs to synthesize ideas from different domains, you can unlock new possibilities and generate creative solutions that you might not have considered otherwise. This approach helps you break free from conventional thinking patterns and encourages cross-disciplinary innovation.

Breaking Creative Blocks

Creative blocks can be frustrating, but LLMs can help you overcome them by providing fresh perspectives and inspiration.

One way LLMs can assist is by continuing an idea or piece of writing when you feel stuck.

To use LLMs for this purpose, follow these steps:

1. Provide the Context

• Input the existing idea or piece of writing into the LLM

• Include any relevant context or background information

• Clearly state where you're stuck or what aspect you want the LLM to focus on

2. Prompt the LLM to Continue

• Ask the LLM to generate potential continuations of the idea or writing

• Encourage the LLM to explore different directions or perspectives

• Provide guidance on the tone, style, or length of the continuation

3. Review and Select

• Read through the generated continuations and select the most promising ones

• Look for ideas that resonate with your vision or offer unexpected insights

• Don't be afraid to combine elements from different continuations

4. Refine and Integrate

• Take the selected continuations and refine them further

• Adapt the language and style to match your original idea or writing

• Integrate the new ideas seamlessly into your work

5. Iterate as Needed

• If you still feel stuck or unsatisfied with the results, repeat steps 2-4

• Experiment with different prompts or ask the LLM to focus on specific aspects

• Continue iterating until you break through the creative block

Example of a creative block prompt:

I'm working on a blog post about the benefits of mindfulness, but I'm struggling to come up with a compelling introduction. Here's what I have so far:

"Mindfulness has become a buzzword in recent years, but what does it really mean? In this post, we'll explore the concept of mindfulness and how it can improve your mental well-being."

Please generate 3 potential continuations for this introduction, focusing on engaging the reader and setting the stage for the rest of the post. Each continuation should be 2-3 sentences long.

By leveraging LLMs to continue your ideas or writing when you're stuck, you can overcome creative blocks and find new inspiration.

This approach can help you explore different directions, gain fresh perspectives, and ultimately move forward with your creative projects.

Expanding Knowledge Domains

When exploring new ideas or topics, it's essential to have a broad understanding of various knowledge domains.

LLMs can help you expand your knowledge by providing insights, explanations, and resources related to the subjects you're interested in.

To use LLMs for expanding your knowledge domains, try the following steps:

1. Identify the Topic

• Determine the topic or idea you want to explore further

• Be specific about the aspects you want to learn more about

• Provide any context or background information you already have

2. Ask for an Overview

• Prompt the LLM to give you a high-level overview of the topic

• Request definitions, key concepts, and important terminology

• Ask the LLM to explain the topic's relevance and its connections to other domains

3. Seek Deeper Insights

• Based on the overview, identify areas where you want more in-depth information

• Ask the LLM to provide more detailed explanations or examples

• Encourage the LLM to share different perspectives or debates surrounding the topic

4. Request Resources

• Ask the LLM to recommend resources for further learning

• Specify the type of resources you prefer (e.g., articles, books, videos, experts)

• Prompt the LLM to provide a brief summary of each resource and why it's valuable

5. Synthesize and Apply

• Review the information and resources provided by the LLM

• Synthesize the key insights and connect them to your existing knowledge

• Consider how you can apply this new knowledge to your ideas or projects

Example of an expanding knowledge prompt:

I'm interested in learning more about the concept of "design thinking" and how it can be applied to problem-solving in business. Can you help me understand this topic better?

Please provide:

1. A brief overview of design thinking, its key principles, and its origins

2. An explanation of how design thinking differs from other problem-solving approaches

3. 2-3 examples of how design thinking has been successfully applied in business

4. Recommendations for 2-3 high-quality resources (articles or books) for further learning about design thinking and its applications

By leveraging LLMs to expand your knowledge domains, you can gain a deeper understanding of various topics, discover new insights, and identify valuable resources for further learning.

This approach can help you broaden your perspective, make connections between different ideas, and ultimately enhance your creative problem-solving skills.

Prompting Workflow Scenarios

Self-Interview Questioning

The self-interview process is a powerful way to leverage LLMs for ideas.

At a high-level, the workflow is:

1. Craft a prompt instructing the model to enter into a dialogue with the goal of diving deeper through questioning.

2. Answer all questions the model gives you.

3. Have the model synthesize your answers at the end into insights and ideas.

Here’s what that flow could look like:

Step 1: Setup the interview instructions

You are an expert interviewer and content idea expert.

Your goal is to help me better understand my ideas and synthesize them into a cohesive idea, topic or theme.

In this exercise, I will start us off by giving a topic delimited by <topic> tags.

I want you to help me:
1) Hone in on my point of view
2) Refine my perspective
3) Help me think clearly about a topic 

I want to get clarity on 5 main things:

1. What problem am I solving?
2. Whose problem am I solving?
3. What benefit am I unlocking?
4. What emotion am I creating?
5. What action am I encouraging?

You will help me test the strength of my topic, identify underlying assumptions, examples, values, influences, etc. 

Feel free to play devil's advocate. 

Follow the direction of the conversation naturally. 

You can ask a variety of surface level, straightforward, and challenging questions to push my thinking.  

Point out inconsistencies or contradictions you find along the way. 

We don't need to dive too deep into psychology.

I will respond in brief or in depth.

I only want 1 question at a time, so that I can focus. You can ask follow up questions after each answer.  

I may ask you to pause along they way by saying DONE to summarize. We will continue until I say stop.

To start, I will give you the topic I want to focus on.

Step 2: Answer until you are satisfied

This prompt above will take you through a series of questions and then at the end when you say DONE it will synthesize and summarize what was discussed.

I find this helpful when I have no exact idea I want to pursue in my head, but want to collect my thoughts and find through lines in my ideas.

Exploration of Ideas into Drafts: From Concepts to Content

Once you have a set of promising ideas generated through LLMs, the next step is to explore and expand those ideas into actual content drafts

Here's a possible workflow for exploring ideas into drafts:

1. Select one or more ideas from your generated list that you want to explore further.

2. Prompt the model to generate a rough outline or structure for a content piece based on the selected idea. For example, "Create an outline for a blog post on '10 Productivity Hacks for Remote Workers.'"

3. Review the generated outline and provide feedback or additional prompts to refine the structure as needed.

4. Once you have a solid outline, prompt the model to generate a draft of the content piece. You can provide specific instructions on tone, style, length, and any other relevant parameters.

5. Review the generated draft and iterate with the model to refine and improve the content. This may involve prompting the model to expand on certain sections, provide more examples, or adjust the language and style.

6. Repeat the process with other ideas until you have a set of content drafts that you can evaluate and prioritize for further development.

By using LLMs to explore ideas into drafts, you can quickly test and refine your concepts, saving time and effort in the content creation process.

Identify Content Gaps

Let’s walk through an example where I identify content gaps in my newsletter.

Step 1: Scrape Newsletter Using Apify

First, go to Apify and create an account.

Then go to this apify crawler here.

Once you’ve selected the crawler, paste the base url of your newsletter into the “Start URL”:

Once it’s run, you should see the output like the below:

There may duplicates, but that’s okay! We can ask ChatGPT to remove the duplicates and return us a cleaned csv file.

Once you’ve done that, you can upload the csv file like I did below:

Use the following prompt with the CSV:

You are an expert in branding and content marketing tasked with conducting a content gap analysis
for a business with the following brand positioning:
<brand>
Leveraging AI to be More Efficient with Content Processes, Self improvement, spirituality, entrepreneurship, community building
</brand>
The business targets the following audience:
<target_audience>
Content marketers, busy professionals / business owners, educational content creators, online writers (twitter and/or newsletters), soloprenuers
</target_audience>
To conduct the analysis, please review the content of the business's past newsletters, which are
provided in CSV I have attached
Follow these steps to complete the content gap analysis:
1. Carefully read through the text content of each newsletter in the CSV. As you read, extract and
make note of the key themes, topics, concepts and ideas covered in the newsletters.
2. For each key theme, topic, concept and idea you identified in step 1, analyze how it relates to
the brand positioning. Also consider how it connects with the target audience. Think about what pain
points the content may be addressing and what desires or goals it speaks to for the audience.
3. Based on your understanding of the brand and target audience, identify 3-5 key topics that are
currently missing or inadequately covered in the newsletter content. These should be topics that
would resonate strongly with the target audience and reinforce the brand positioning.
4. For each topic you suggest covering, provide the following details:
<suggested_topic_template>
Topic: <topic idea>
Reasoning: <reason why it relates to the brand and audience>
Pain Points:
- <pain point 1 it addresses>
- <pain point 2 it addresses>
- <additional pain points>
Desires:
- <desire 1 it speaks to>
- <desire 2 it speaks to>
- <additional desires>
</suggested_topic_template>
5. After completing the analysis, provide your overall content gap recommendations inside
<content_gap_recommendations> tags.
Remember, the goal is to identify valuable content opportunities that will resonate with
the target audience and reinforce the person's brand. Let me know if you have any other questions!
 

I also made a video about this process, which you can watch here:

AI Tools For Ideation

Before we end this week, I want to share some AI tools that can also help in your ideation process:

1. VidIQ’s Custom GPT—connects to your YouTube channel analytics. Can help you come up with on brand ideas, titles, scripts and more.

2. Idea Generator—a simple UI that will help you generate a ton of ideas for a topic you’re interested in.

3. AI Suggests—Offers over 20 templates and utilizes AI to suggest new ideas for various types of content marketing such as Instagram captions, video titles, blog titles, product descriptions, and so on.

I don’t imagine you’ll need these as much if you master prompting and workflows for your own needs but I wanted to share other ways you can ideate with AI.

Closing Thoughts

So to recap, we explored the crucial role of ideation in the content creation process and how LLMs can supercharge your ideation efforts.

By leveraging techniques like

• iteration ideation

• idea synthesis

• breaking creative blocks

• expanding knowledge domains

you can generate fresh, compelling ideas consistently.

We've also delved into practical prompting workflows and AI tools that can help you integrate LLMs seamlessly into your ideation process.

As the digital landscape continues to evolve, mastering the art of ideation with LLMs will become increasingly essential for capturing attention and creating content that resonates with your audience.

Embrace the power of AI-assisted ideation, and unlock your full creative potential to stay ahead in the game.

That’s all for now folks! ‘Til next time, much love and peace y’all 🤠