2. Make funny birthday cards for niche

I want to build out a task that makes funny birthday cards for any niche. I would input a description of my niche audience or topic, and the final output would be a collection of completed images (in jpeg), with funny text and image. I think we could generate the image with any of our AI generators, eg Dalle, and provided we prompt something that looks hard drawn the end result would be good. But what we really want to spend time on is the writing of the puns for the card. These should be funny and relevant to the niche/topic in question. We would therefore want a series of prompts which researches the topic in question, looks for possible words and phrases that are relevant to the niche, and then generates a list of slogans for that niche. Since this is such an important step we likely want to run a few searches to collect possible names, and then a few LLM prompts that import a summary of these search results and use the research to generate slogan ideas. We are likely to need some custom prompts to generate "play-on" words. The slogans are important. We should likely run these prompts a few times, and then have another AI review and choose the best slogans. And then it's just a case of inserting the text and images into a layout. I have a tool called Bannerbear which lets me build template image structures, so I will likely make 2-3x of these before we formalize the task, and then we will just slot "card-text-line-1" and "card-text-line-2" and "doodle-image-url" into this template to give us our cards. These can be 1000x1000 to start (Bannerbear lets me scale up easily). Research is very important with this one: before we start the whole process, we should probably search Google images (for "{niche topic} birthday cards") and see what current top-selling images in the niche look like, and then from there we can begin to grab some relevant niches.

Sub Agent #1 | Sub Agent #2 | Sub Agent #3

CLAUDE:- final-output-and-component-outputs

Thank you for the detailed description of your project. Let me address your questions one by one.

a) The final output of this project would be a set of funny birthday card images in JPEG format, tailored to a specific niche or topic. Each card would have a humorous slogan or pun relevant to the niche, along with a doodle-style image generated by an AI tool like DALL-E. The images would be sized at 1000x1000 pixels initially, with the flexibility to scale up using the Bannerbear tool.

b) To break down the task into its core components, we can identify the following:

1. Research Component:
   - Perform Google image searches for "{niche topic} birthday cards" to gather inspiration and understand current top-selling designs in the niche.
   - Collect relevant keywords, phrases, and ideas related to the niche topic.

2. Slogan Generation Component:
   - Use the research findings to generate a list of potential slogans or puns for the niche topic.
   - Employ custom prompts to generate "play-on" words and phrases.
   - Run multiple iterations of slogan generation prompts to obtain a diverse set of options.
   - Utilize an AI reviewer to select the best slogans from the generated list.

3. Image Generation Component:
   - Use an AI image generator like DALL-E to create doodle-style images relevant to the niche topic.
   - Ensure the generated images have a hand-drawn appearance to match the desired style.

4. Card Layout Component:
   - Create 2-3 template image structures using the Bannerbear tool.
   - Define placeholders for "card-text-line-1", "card-text-line-2", and "doodle-image-url" within the templates.

5. Final Assembly Component:
   - Combine the selected slogans and generated images into the pre-defined Bannerbear templates.
   - Insert the chosen slogans into the "card-text-line-1" and "card-text-line-2" placeholders.
   - Incorporate the generated doodle image by providing its URL to the "doodle-image-url" placeholder.
   - Generate the final birthday card images in JPEG format, initially sized at 1000x1000 pixels.

The core components can be summarized as:
1. Research findings (keywords, phrases, inspiration)
2. Generated slogans or puns
3. Generated doodle-style images
4. Bannerbear templates with placeholders
5. Final assembled birthday card images in JPEG format

By breaking down the task into these specific components, the AI agent can focus on each aspect individually and ensure a structured approach to generating the desired funny birthday cards for any given niche or topic.

GPT4:- final-output-and-component-outputs

a) The final output of this task would be a collection of JPEG images of funny birthday cards tailored to a specific niche or topic. Each card would include an image generated using an AI generator, such as Dalle, and funny text or a slogan relevant to the specific niche or topic.

b) The task can be broken down into several components:

1. **Research & Information Collection**: Using an AI to search for information on the niche topic, look for relevant words, phrases, or slang, see what already exists in terms of birthday cards in that particular niche and understand the layout. The output at this stage can be a compiled list or summary of the research findings.

2. **Content Generator - Slogans/Puns**: Using AI to generate possible funny text/slogans based on the summary of research findings. The output of this would be a list of relevant slogans or puns related to the niche topic.

3. **Content Review**: AI will review and choose the best slogans from the generated list. The output here would be the selected slogans that will go into the final cards.

4. **Image Generator**: Based on the provided prompts and relevant to the niche topic, an AI generator (like Dalle) will create the images. The output here would be the generated doodle images in the form of URLs or downloadable files.

5. **Card Layout & Design**: Using Bannerbear or a similar tool, a template for the card layout will be created. The slots in the template (i.e., "card-text-line-1," "card-text-line-2," "doodle-image-url") will be filled with relevant text and images generated from previous steps. The output at this stage would be the final JPEG of the birthday card.

6. **Iteration**: The process would be repeated (steps 2-5) to create a collection of different birthday cards.

Thus, the overall process involves AI research, slogan and pun generation, slogan selection, AI image generation, and card synthesis.

claude 3 - extract final output

extract component outputs

To break down the task into its core components, we can identify the following:

Claude - research-ideas-and-input-variables

Thank you for the detailed task description. Let me address your questions one by one:

a) To clarify the final output of the task, I would ask the following questions:
1. How many birthday card designs do you want to generate for each niche/topic?
2. What dimensions should the final birthday card images be? (You mentioned 1000x1000 as a starting point, but it would be good to confirm the desired final dimensions)
3. In what format should the final output be delivered? (e.g., a zip file containing all the JPEG images, a Google Drive folder, etc.)
4. Are there any specific design elements or styles you want to incorporate into the birthday cards? (e.g., minimalist, vintage, hand-drawn, etc.)

b) Before launching the task completion process, the user should provide the following inputs:
1. The niche or topic for which the birthday cards will be created (e.g., "fitness enthusiasts," "dog lovers," "software developers," etc.)
2. Any specific keywords or phrases related to the niche that should be considered when generating the slogans and puns.
3. The desired number of birthday card designs for the given niche/topic.
4. The dimensions and format of the final output images.
5. Any specific design preferences or requirements (if applicable).

c) I have a suggestion for segmenting the task of generating funny and relevant slogans for the birthday cards:
1. Research and data collection: This silo would focus on gathering information about the niche/topic, including relevant keywords, phrases, and concepts. It would involve conducting targeted searches on search engines, social media platforms, and other relevant sources to compile a comprehensive list of terms and ideas related to the niche.
2. Slogan generation: This silo would take the information gathered in the research phase and use it to generate a large pool of potential slogans and puns. It would involve using LLMs and custom prompts to create a variety of funny, witty, and relevant slogans based on the niche-specific data.
3. Slogan evaluation and selection: This silo would focus on reviewing and selecting the best slogans from the generated pool. It would involve using AI to assess the slogans based on factors such as relevance, humor, and originality, and then choosing the top candidates for each birthday card design.

By segmenting the slogan generation process in this way, you can ensure that each step is given the necessary attention and resources, resulting in higher-quality and more targeted slogans for the birthday cards.

GPT - research-ideas-and-input-variables

a) Here are some questions that might help to clarify exactly what the final output should look like:

1. What size should the greetings cards be? You mentioned they could start at 1000x1000, but are there any optimum dimensions for the cards?
2. How many slogans should each card contain? Only one or multiple? 
3. Should the cards include any additional elements beyond the funny text and niche-relevant images?
4. Would you like to include any specific branding elements in the cards, such as a logo, color scheme, or a particular style of typography?
5. How many cards do you want to generate per each niche? 
6. Should all cards for one niche have the same slogan, or would you like different slogans for each?
7. Do you prefer a specific style of humor or would any type be acceptable as long as it fits the niche? 
   
b) Before the task is initiated, the following inputs would be necessary:

1. A clear description or definition of the niche audience or topic for the cards. 
2. Any specific instructions regarding the humor or tone to be used in the slogans. For example, the humor could be sarcastic, punny, slapstick, self-deprecating, etc.
3. Potential words, phrases, or themes that are crucial to the niche or that you would like to see included in the card.
4. Details and login information (if needed) for Bannerbear, so the AI knows where to insert the text and images.
5. Preferences for card aesthetics (font type, color scheme, layout etc) so the finished cards fit into the look and feel you're going for.
6. Any pre-made templates in Bannerbear you might want to use. If available, this would be a great input to speed up the card design process.

template-ideas

Based on the project description and the importance of collecting templates to refine the definition of a successful output, here are some suggestions for templates and resources the AI Agent could gather:

1. Successful Birthday Card Images:
   - Perform a Google Image search for "funny birthday cards" or "{niche topic} birthday cards" to find top-selling and popular designs.
   - Visit online greeting card marketplaces like Etsy, Zazzle, or Redbubble to view bestselling birthday cards in various niches.
   - Collect a set of successful birthday card images that align with the desired style (humorous slogans, doodle-style images) and niche relevance.

2. Slogan and Pun Examples:
   - Search for websites or blog posts that compile lists of funny birthday slogans, puns, or jokes for inspiration.
   - Look for niche-specific humor websites or forums where people share jokes and puns related to the target niche or topic.
   - Gather a collection of successful and engaging slogans or puns that resonate with the intended audience.

3. Doodle-Style Image References:
   - Explore platforms like Dribbble, Behance, or Pinterest to find examples of doodle-style illustrations or graphics.
   - Search for "doodle art" or "hand-drawn illustrations" to get a sense of the desired visual style for the birthday card images.
   - Collect a set of doodle-style image references that can serve as inspiration for the AI-generated images.

4. Birthday Card Design Templates:
   - Visit websites that offer birthday card design templates, such as Canva or greeting card manufacturer websites.
   - Look for templates that showcase effective layouts, color schemes, and typography for birthday cards.
   - Gather a few design templates that can serve as a starting point for structuring the generated birthday card images.

5. Niche-Specific Content:
   - Research popular blogs, websites, or social media accounts related to the target niche or topic.
   - Collect examples of niche-specific content, such as articles, photos, or memes, to gain insights into the language, humor, and preferences of the target audience.
   - Use this content to inform the slogan generation process and ensure relevance to the niche.

By collecting these templates and resources, the AI Agent can gain a clearer understanding of what constitutes a successful output for the birthday card generation task. The Agent can analyze the gathered examples to identify common patterns, styles, and elements that resonate with the target audience. This analysis will help refine the definition of a successful output and guide the generation process to create funny, relevant, and visually appealing birthday cards for the specific niche or topic.

claude 3 - searches

Here are 5 specific keyphrases to search on Google to find helpful templates and refine the definition of a successful output for the funny birthday card generation task:

1. "funny dog lover birthday cards etsy" - This search will help find popular and successful birthday card designs specifically tailored to the "dog lover" niche audience, available on the Etsy marketplace.

2. "humorous birthday slogans for engineers" - This search aims to uncover examples of funny and relevant slogans or puns that cater to the "engineer" niche, which can serve as inspiration for generating niche-specific humor.

3. "hand-drawn doodle style birthday illustrations dribbble" - By including the design platform Dribbble in the search, we can find high-quality examples of doodle-style illustrations specifically designed for birthday cards, helping to establish the desired visual style.

4. "canva funny birthday card templates" - Searching for templates on the popular design platform Canva will provide examples of effective layouts, color schemes, and typography for creating visually appealing and humorous birthday cards.

5. "funny birthday memes for teachers" - This search will help discover niche-specific humor and content related to the "teacher" audience, providing insights into the language and preferences of the target group, which can inform the slogan generation process.

By searching for these specific keyphrases, we can quickly find relevant templates and examples that will help refine the definition of a successful output for the funny birthday card generation task. The results will provide insights into popular designs, humorous slogans, visual styles, and niche-specific content that resonate with the target audience, guiding the AI Agent in creating highly relevant and engaging birthday cards.

subagent-summary

A) FULL TASK SUMMARY: Generate a set of funny birthday card images tailored to a specific niche or topic, with humorous slogans and AI-generated doodle-style images.

B) FINAL TASK OUTPUT: A collection of JPEG images, each sized at 1000x1000 pixels, containing funny birthday card designs specific to a given niche or topic. The images will have relevant and humorous slogans or puns, along with AI-generated doodle-style illustrations.

C) COMPONENTS OVERVIEW:
1. Research findings (keywords, phrases, inspiration)
2. Generated slogans or puns
3. Generated doodle-style images
4. Bannerbear templates with placeholders
5. Final assembled birthday card images in JPEG format

D) OVERVIEW OF SUBAGENTS:
1. Research Agent: Conducts research on the given niche or topic to gather relevant keywords, phrases, and inspiration for the birthday card designs.
2. Slogan Generator Agent: Generates humorous slogans or puns based on the research findings, using natural language processing and language models.
3. Image Generator Agent: Creates AI-generated doodle-style images relevant to the niche or topic, ensuring a hand-drawn appearance.
4. Template Designer Agent: Designs and sets up template image structures using the Bannerbear tool, defining placeholders for text and images.
5. Card Assembly Agent: Combines the generated slogans, images, and templates to produce the final birthday card images in JPEG format.

E) SUBAGENT 1: Research Agent
- Name: Niche Researcher
- Final Output: A JSON object containing relevant keywords, phrases, and inspiration for the given niche or topic.
- Accomplishment: The Niche Researcher will utilize Google's Custom Search API to perform image searches for "{niche topic} birthday cards". It will analyze the top search results, extracting relevant keywords, phrases, and design elements using natural language processing and computer vision techniques. The subagent will compile these findings into a structured JSON object.
- Sequence:
  1. Receive the niche or topic as input.
  2. Perform Google image searches using the Custom Search API.
  3. Analyze the top search results using NLP and computer vision libraries.
  4. Extract relevant keywords, phrases, and design inspiration.
  5. Structure the findings into a JSON object.
  6. Output the JSON object containing the research results.

F) SUBAGENT 2: Slogan Generator Agent
- Name: Pun Master
- Final Output: A list of humorous slogans or puns related to the given niche or topic.
- Accomplishment: The Pun Master will take the research findings from the Niche Researcher as input. It will utilize language models like GPT-3 to generate a variety of humorous slogans or puns based on the provided keywords and phrases. The subagent will employ custom prompts to encourage the generation of "play-on" words and witty phrases. It will run multiple iterations to obtain a diverse set of options and then use an AI-based reviewer to select the best slogans.
- Sequence:
  1. Receive the research findings JSON object as input.
  2. Extract relevant keywords and phrases from the research.
  3. Utilize GPT-3 with custom prompts to generate humorous slogans and puns.
  4. Run multiple iterations to obtain a diverse set of options.
  5. Employ an AI-based reviewer to select the best slogans.
  6. Output the list of selected slogans or puns.

G) SUBAGENT 3: Image Generator Agent
- Name: Doodle Designer
- Final Output: A collection of AI-generated doodle-style images relevant to the given niche or topic.
- Accomplishment: The Doodle Designer will use the research findings and generated slogans as inspiration to create doodle-style images. It will leverage AI image generation tools like DALL-E, providing specific prompts to generate hand-drawn-like illustrations that complement the humorous slogans. The subagent will generate multiple image variations and select the most suitable ones based on relevance and aesthetic appeal.
- Sequence:
  1. Receive the research findings and generated slogans as input.
  2. Use the input to formulate specific prompts for AI image generation.
  3. Utilize tools like DALL-E to generate doodle-style images.
  4. Generate multiple image variations for each slogan.
  5. Select the most suitable images based on relevance and aesthetics.
  6. Output the collection of generated doodle-style images.

H) SUBAGENT 4: Template Designer Agent
- Name: Layout Architect
- Final Output: Bannerbear templates with placeholders for text and images.
- Accomplishment: The Layout Architect will design 2-3 template image structures using the Bannerbear tool. It will define placeholders for "card-text-line-1", "card-text-line-2", and "doodle-image-url" within the templates. The subagent will ensure that the templates have an appealing layout and are compatible with the generated slogans and images.
- Sequence:
  1. Analyze the generated slogans and images to determine layout requirements.
  2. Design 2-3 template image structures using the Bannerbear tool.
  3. Define placeholders for text and image elements within the templates.
  4. Ensure the templates have an appealing layout and are compatible with the content.
  5. Output the Bannerbear templates with defined placeholders.

I) SUBAGENT 5: Card Assembly Agent
- Name: Birthday Card Composer
- Final Output: A collection of final birthday card images in JPEG format, sized at 1000x1000 pixels.
- Accomplishment: The Birthday Card Composer will take the generated slogans, doodle-style images, and Bannerbear templates as input. It will combine these elements by inserting the slogans into the designated text placeholders and incorporating the images using the provided image URLs. The subagent will ensure proper alignment and formatting of the elements within the templates. It will then render the final birthday card images in JPEG format, initially sized at 1000x1000 pixels.
- Sequence:
  1. Receive the generated slogans, doodle-style images, and Bannerbear templates as input.
  2. Insert the slogans into the designated text placeholders within the templates.
  3. Incorporate the doodle-style images using the provided image URLs.
  4. Ensure proper alignment and formatting of the elements within the templates.
  5. Render the final birthday card images in JPEG format, sized at 1000x1000 pixels.
  6. Output the collection of final birthday card images.

These subagents work together to generate a set of funny birthday card images tailored to a specific niche or topic, with humorous slogans and AI-generated doodle-style illustrations.

subagent-summary-amended-for-actual-tasks

Thank you for providing the updated list of available tasks. Based on these tasks, I will now revise the subagents and their specific actions to generate the funny birthday card images tailored to a specific niche or topic.

SUBAGENT 1: Niche Researcher
- Final Output: A JSON object containing relevant keywords, phrases, and inspiration for the given niche or topic.
- Sequence:
  1. Receive the niche or topic as input from the user interface.
  2. Use Task 15 (Search Google) to perform searches for "{niche topic} birthday cards" and collect the search results.
  3. Use Task 1 (GPT4) to analyze the search results and extract relevant keywords, phrases, and design inspiration.
  4. Structure the findings into a JSON object.
  5. Use Task 9 (Save content to server) to save the JSON object to the server.
  6. Output the URL of the saved JSON object.

SUBAGENT 2: Pun Master
- Final Output: A list of humorous slogans or puns related to the given niche or topic.
- Sequence:
  1. Receive the URL of the research findings JSON object from the Niche Researcher.
  2. Use Task 10 (Access text files already saved) to retrieve the JSON object.
  3. Use Task 1 (GPT4) with custom prompts to generate humorous slogans and puns based on the research findings.
  4. Run multiple iterations using Task 1 (GPT4) to obtain a diverse set of options.
  5. Use Task 1 (GPT4) as an AI-based reviewer to select the best slogans.
  6. Use Task 9 (Save content to server) to save the list of selected slogans as a text file.
  7. Output the URL of the saved text file containing the slogans.

SUBAGENT 3: Doodle Designer
- Final Output: A collection of AI-generated doodle-style images relevant to the given niche or topic.
- Sequence:
  1. Receive the URL of the research findings JSON object and the URL of the generated slogans text file.
  2. Use Task 10 (Access text files already saved) to retrieve the research findings and slogans.
  3. Use Task 1 (GPT4) to formulate specific prompts for AI image generation based on the research findings and slogans.
  4. Use Task 22 (Create square image with Dalle3) to generate multiple doodle-style images for each slogan.
  5. Use Task 1 (GPT4) to select the most suitable images based on relevance and aesthetics.
  6. Output the URLs of the generated doodle-style images.

SUBAGENT 4: Layout Architect
- Final Output: Bannerbear templates with placeholders for text and images.
- Sequence:
  1. Analyze the generated slogans and images to determine layout requirements.
  2. Use Task 26 (Bannerbear - image creator) to design 2-3 template image structures with placeholders for text and images.
  3. Output the URLs of the Bannerbear templates.

SUBAGENT 5: Birthday Card Composer
- Final Output: A collection of final birthday card images in JPEG format, sized at 1000x1000 pixels.
- Sequence:
  1. Receive the URLs of the generated slogans, doodle-style images, and Bannerbear templates.
  2. Use Task 26 (Bannerbear - image creator) to insert the slogans and images into the designated placeholders within the templates.
  3. Use Task 26 (Bannerbear - image creator) to render the final birthday card images in JPEG format, sized at 1000x1000 pixels.
  4. Output the URLs of the final birthday card images.

These subagents utilize the available tasks to generate funny birthday card images tailored to a specific niche or topic, with humorous slogans and AI-generated doodle-style illustrations.

random-autocomplete

To achieve the end goal of generating a set of funny birthday card images tailored to a specific niche or topic, the most relevant tasks might be:

1. Research the niche or topic to gather relevant keywords, phrases, and themes.
2. Generate a list of potential slogans or puns based on the research findings.
3. Use an AI image generation tool like DALL-E to create doodle-style images related to the niche or topic.
4. Design a template for the birthday card layout using the Bannerbear tool.
5. Integrate the generated slogans and images into the birthday card template.
6. Export the final birthday card designs as 1000x1000 pixel JPEG images.

The overall sequence of actions would be:

1. Input niche or topic description > Research relevant keywords, phrases, and themes > Generate a list of potential slogans or puns > Select the best slogans or puns
2. Input niche or topic description > Generate doodle-style images using an AI tool like DALL-E > Select the most appropriate images
3. Design birthday card template using Bannerbear > Integrate selected slogans and images into the template > Export final designs as 1000x1000 pixel JPEG images

To complete this task, the Agent could follow these steps:

1. Use web scraping tools or APIs to gather information about the specified niche or topic from various sources, such as Google Images, Wikipedia, or relevant websites.
2. Employ natural language processing (NLP) techniques to analyze the gathered information and extract relevant keywords, phrases, and themes.
3. Use the extracted keywords, phrases, and themes to generate a list of potential slogans or puns using an AI language model like GPT-3 or a custom-trained model.
4. Develop a ranking system to select the best slogans or puns based on factors such as relevance, humor, and originality.
5. Utilize an AI image generation tool like DALL-E to create doodle-style images related to the niche or topic, using the extracted keywords and themes as prompts.
6. Implement a selection process to choose the most appropriate images based on factors such as relevance, visual appeal, and consistency with the overall theme.
7. Use the Bannerbear API to design a template for the birthday card layout, incorporating placeholders for the selected slogans and images.
8. Integrate the selected slogans and images into the birthday card template using the Bannerbear API or a custom-built integration script.
9. Export the final birthday card designs as 1000x1000 pixel JPEG images using the Bannerbear API or a custom export function.

By following this sequence of actions and leveraging the appropriate APIs, libraries, and AI tools, the Agent should be able to successfully generate a set of funny, niche-specific birthday card images as desired.

claude 3 - mermaid-for-all

<!DOCTYPE html>
<html>
<head>
  <title>Mermaid Diagram Example</title>
  <script src="https://cdnjs.cloudflare.com/ajax/libs/mermaid/9.4.3/mermaid.min.js"></script>
</head>
<body>
  <div class="mermaid">
graph TD
A[User Input: Niche or Topic] --> B{Niche Researcher}
B --> C[Search Google for niche birthday cards]
C --> D[Analyze search results and extract keywords, phrases, and inspiration]
D --> E[Structure findings into a JSON object]
E --> F[Save JSON object to server]
F --> G{Pun Master}
G --> H[Retrieve research findings JSON object]
H --> I[Generate humorous slogans and puns based on research findings]
I --> J[Run multiple iterations to obtain diverse options]
J --> K[Select the best slogans using AI-based reviewer]
K --> L[Save selected slogans as a text file]
L --> M{Doodle Designer}
M --> N[Retrieve research findings and slogans]
N --> O[Formulate prompts for AI image generation]
O --> P[Generate multiple doodle-style images for each slogan]
P --> Q[Select the most suitable images based on relevance and aesthetics]
Q --> R{Layout Architect}
R --> S[Analyze generated slogans and images to determine layout requirements]
S --> T[Design Bannerbear templates with placeholders for text and images]
T --> U{Birthday Card Composer}
U --> V[Insert slogans and images into designated placeholders within templates]
V --> W[Render final birthday card images in JPEG format, sized at 1000x1000 pixels]
W --> X[Output URLs of final birthday card images]
  </div>

SubAgent #1

claude 3 - subagent1

subagent1-refined

A) SUBAGENT SUMMARY: The Pun Master subagent generates humorous slogans and puns related to a given niche or topic based on research findings.

B) FINAL TASK OUTPUT: A text file containing a list of 10-20 humorous slogans or puns, each being 5-15 words in length, saved on the server with a URL for accessing the file.

C) SUBAGENT INPUT: The URL of a JSON object containing relevant keywords, phrases, and inspiration for the given niche or topic, obtained from the Niche Researcher subagent.

E) SUBAGENT TASK SUMMARY:
1. Retrieve research findings JSON object using the input URL and Task 10 (Access text files already saved).
2. Use Task 1 (GPT4) to analyze the research findings and generate a list of 50-100 potential slogans and puns based on the keywords, phrases, and inspiration found in the JSON object.
3. Use Task 1 (GPT4) to refine and select the best 20-30 slogans and puns from the generated list, ensuring they are humorous, relevant to the niche, and within the 5-15 word limit.
4. Use Task 1 (GPT4) to further narrow down the list to the top 10-20 slogans and puns.
5. Use Task 9 (Save content to server) to save the final list of slogans and puns as a text file on the server.
6. Output the URL of the saved text file containing the slogans and puns.

F) SILOS:
Silo 1: Research Analysis and Slogan Generation
- Input: URL of the research findings JSON object
- Task 10 to retrieve the JSON object
- Task 1 to analyze findings and generate initial list of slogans and puns
- Output: List of 50-100 potential slogans and puns

Silo 2: Slogan Refinement and Selection
- Input: List of 50-100 potential slogans and puns
- Task 1 to refine and select the best 20-30 slogans and puns
- Task 1 to further narrow down to the top 10-20 slogans and puns
- Output: Final list of 10-20 humorous and relevant slogans and puns

Silo 3: Saving and Outputting Results
- Input: Final list of 10-20 slogans and puns
- Task 9 to save the list as a text file on the server
- Output: URL of the saved text file containing the slogans and puns

subagent1refined-repeating

Subagent1 task-ID-numbers

subagent1-output-variables

Here is the updated task list with the output tokens added:

G) FULL TASKS (INCLUDING REPEATING TASKS):

1. Use Task 10 (Access text files already saved) to retrieve the research findings JSON object using the input URL [research-findings-json].
2. Use Task 1 (GPT4) to analyze the research findings [research-findings-json] and generate a list of 50-100 potential slogans and puns based on the keywords, phrases, and inspiration found in the JSON object. Output: [generated-slogans-puns-list].
3. Use Task 1 (GPT4) to refine and select the best 20-30 slogans and puns from the generated list [generated-slogans-puns-list], ensuring they are humorous, relevant to the niche, and within the 5-15 word limit. Output: [refined-slogans-puns-list].
4. Use Task 1 (GPT4) to further narrow down the list [refined-slogans-puns-list] to the top 10-20 slogans and puns. Output: [top-slogans-puns-list].
5. Use Task 9 (Save content to server) to save the final list of slogans and puns [top-slogans-puns-list] as a text file on the server. Output: [slogans-puns-text-file-url].
6. Output the URL of the saved text file containing the slogans and puns [slogans-puns-text-file-url].

H) OUTPUT TOKENS:

[research-findings-json]: The JSON object containing the research findings retrieved from the server.
[generated-slogans-puns-list]: The initial list of 50-100 potential slogans and puns generated based on the research findings.
[refined-slogans-puns-list]: The refined list of 20-30 slogans and puns selected from the generated list.
[top-slogans-puns-list]: The final list of 10-20 top slogans and puns selected from the refined list.
[slogans-puns-text-file-url]: The URL of the text file containing the final list of slogans and puns, saved on the server.

subagent1refined-repeating-json-core

G) TASKFLOW SUMMARY

Title task: Retrieve research findings JSON object
Description of the task: Use Task 10 (Access text files already saved) to retrieve the research findings JSON object using the input URL provided.
Input(s) involved: [input_research_findings_url]
Output token name: [research_findings_json]

Title task: Generate potential slogans and puns
Description of the task: Use Task 1 (GPT4) to analyze the research findings and generate a list of 50-100 potential slogans and puns based on the keywords, phrases, and inspiration found in the JSON object.
Input(s) involved: [research_findings_json]
Output token name: [generated_slogans_list]

Title task: Refine and select best slogans and puns
Description of the task: Use Task 1 (GPT4) to refine and select the best 20-30 slogans and puns from the generated list, ensuring they are humorous, relevant to the niche, and within the 5-15 word limit.
Input(s) involved: [generated_slogans_list]
Output token name: [refined_slogans_list]

Title task: Further narrow down the slogans list
Description of the task: Use Task 1 (GPT4) to further narrow down the list to the top 10-20 slogans and puns.
Input(s) involved: [refined_slogans_list]
Output token name: [final_slogans_list]

Title task: Save final slogans list to server
Description of the task: Use Task 9 (Save content to server) to save the final list of slogans and puns as a text file on the server.
Input(s) involved: [final_slogans_list]
Output token name: [saved_slogans_url]

Title task: Output saved slogans URL
Description of the task: Output the URL of the saved text file containing the slogans and puns.
Input(s) involved: [saved_slogans_url]
Output token name: [output_slogans_url]

subagent1mermaid

SubAgent #2

claude 3 - subagent2

subagent2-refined

A) SUBAGENT SUMMARY: The Pun Master subagent generates humorous slogans and puns related to a given niche or topic based on research findings.

B) FINAL TASK OUTPUT: A text file containing a list of selected humorous slogans or puns related to the given niche or topic, saved on the server with a URL for accessing the file.

C) SUBAGENT INPUT:
- URL of the research findings JSON object from the Niche Researcher subagent.

E) SUBAGENT TASK SUMMARY:
1. Retrieve research findings:
   - Input: URL of the research findings JSON object
   - Task 10 (Access text files already saved): Retrieve the JSON object from the provided URL
   - Output: Research findings JSON object

2. Generate slogans and puns:
   - Input: Research findings JSON object
   - Task 1 (GPT4) with custom prompts: Generate humorous slogans and puns based on the research findings
   - Output: List of generated slogans and puns

3. Select best slogans:
   - Input: List of generated slogans and puns
   - Task 1 (GPT4) as an AI-based reviewer: Select the best slogans from the generated list
   - Output: List of selected slogans

4. Save selected slogans:
   - Input: List of selected slogans
   - Task 9 (Save content to server): Save the list of selected slogans as a text file on the server
   - Output: URL of the saved text file containing the slogans

F) SILOS:
Silo 1: Retrieval and Generation
- Retrieve research findings (Task 10)
- Generate slogans and puns (Task 1 with custom prompts)

Silo 2: Selection and Saving
- Select best slogans (Task 1 as AI-based reviewer)
- Save selected slogans (Task 9)

The Pun Master subagent operates in two main silos. The first silo focuses on retrieving the research findings and generating a diverse set of slogans and puns using GPT-4 with custom prompts. The second silo involves selecting the best slogans from the generated list using GPT-4 as an AI-based reviewer and saving the selected slogans as a text file on the server.

By breaking down the subagent into these silos, the workflow becomes more modular and efficient, allowing for better organization and potential parallelization of tasks within each silo.

subagent2refined-repeating

A) SUBAGENT SUMMARY: The Pun Master subagent generates humorous slogans and puns related to a given niche or topic based on research findings.

B) FINAL TASK OUTPUT: A text file containing a list of selected humorous slogans or puns related to the given niche or topic, saved on the server with a URL for accessing the file.

C) SUBAGENT INPUT:
- URL of the research findings JSON object from the Niche Researcher subagent.

F) SILOS:
Silo 1: Retrieval and Generation
- Retrieve research findings (Task 10)
- Generate slogans and puns (Task 1 with custom prompts)

Silo 2: Selection and Saving
- Select best slogans (Task 1 as AI-based reviewer)
- Save selected slogans (Task 9)

G) FULL TASKS (INCLUDING REPEATING TASKS):
1. Retrieve research findings (Task 10):
   - Input: URL of the research findings JSON object
   - Output: Research findings JSON object

2. Generate slogans and puns - Iteration 1 (Task 1 with custom prompts):
   - Input: Research findings JSON object
   - Output: List of generated slogans and puns - Iteration 1

3. Generate slogans and puns - Iteration 2 (Task 1 with custom prompts):
   - Input: Research findings JSON object
   - Output: List of generated slogans and puns - Iteration 2

4. Generate slogans and puns - Iteration 3 (Task 1 with custom prompts):
   - Input: Research findings JSON object
   - Output: List of generated slogans and puns - Iteration 3

5. Combine generated slogans and puns (Task 1):
   - Input: Lists of generated slogans and puns from Iterations 1, 2, and 3
   - Output: Combined list of generated slogans and puns

6. Select best slogans (Task 1 as AI-based reviewer):
   - Input: Combined list of generated slogans and puns
   - Output: List of selected slogans

7. Save selected slogans (Task 9):
   - Input: List of selected slogans
   - Output: URL of the saved text file containing the slogans

The full task list includes three iterations of generating slogans and puns using GPT-4 with custom prompts. This ensures a diverse set of options to choose from. The generated slogans and puns from each iteration are then combined into a single list before proceeding to the selection and saving steps.

Subagent2 task-ID-numbers

subagent2-output-variables

Here's the updated task list with the output tokens added:

A) SUBAGENT SUMMARY: The Pun Master subagent generates humorous slogans and puns related to a given niche or topic based on research findings.

B) FINAL TASK OUTPUT: A text file containing a list of selected humorous slogans or puns related to the given niche or topic, saved on the server with a URL for accessing the file.

C) SUBAGENT INPUT:
- URL of the research findings JSON object from the Niche Researcher subagent.

2. Generate slogans and puns:
   - Input: Research findings JSON object [research-findings]
   - Task 1 (GPT4) with custom prompts: Generate humorous slogans and puns based on the research findings
   - Output: List of generated slogans and puns [generated-slogans]

3. Select best slogans:
   - Input: List of generated slogans and puns [generated-slogans]
   - Task 1 (GPT4) as an AI-based reviewer: Select the best slogans from the generated list
   - Output: List of selected slogans [selected-slogans]

4. Save selected slogans:
   - Input: List of selected slogans [selected-slogans]
   - Task 9 (Save content to server): Save the list of selected slogans as a text file on the server
   - Output: URL of the saved text file containing the slogans [slogans-url]

F) SILOS:
Silo 1: Retrieval and Generation
- Retrieve research findings (Task 10)
- Generate slogans and puns (Task 1 with custom prompts)

Silo 2: Selection and Saving
- Select best slogans (Task 1 as AI-based reviewer)
- Save selected slogans (Task 9)

G) FULL TASKS (INCLUDING REPEATING TASKS):
1. Retrieve research findings (Task 10):
   - Input: URL of the research findings JSON object
   - Output: Research findings JSON object [research-findings]

2. Generate slogans and puns - Iteration 1 (Task 1 with custom prompts):
   - Input: Research findings JSON object [research-findings]
   - Output: List of generated slogans and puns - Iteration 1 [generated-slogans-1]

3. Generate slogans and puns - Iteration 2 (Task 1 with custom prompts):
   - Input: Research findings JSON object [research-findings]
   - Output: List of generated slogans and puns - Iteration 2 [generated-slogans-2]

4. Generate slogans and puns - Iteration 3 (Task 1 with custom prompts):
   - Input: Research findings JSON object [research-findings]
   - Output: List of generated slogans and puns - Iteration 3 [generated-slogans-3]

5. Combine generated slogans and puns (Task 1):
   - Input: Lists of generated slogans and puns from Iterations 1, 2, and 3 [generated-slogans-1], [generated-slogans-2], [generated-slogans-3]
   - Output: Combined list of generated slogans and puns [combined-slogans]

6. Select best slogans (Task 1 as AI-based reviewer):
   - Input: Combined list of generated slogans and puns [combined-slogans]
   - Output: List of selected slogans [selected-slogans]

7. Save selected slogans (Task 9):
   - Input: List of selected slogans [selected-slogans]
   - Output: URL of the saved text file containing the slogans [slogans-url]

H) OUTPUT TOKENS:
- [research-findings]: Research findings JSON object
- [generated-slogans-1]: List of generated slogans and puns - Iteration 1
- [generated-slogans-2]: List of generated slogans and puns - Iteration 2
- [generated-slogans-3]: List of generated slogans and puns - Iteration 3
- [combined-slogans]: Combined list of generated slogans and puns
- [selected-slogans]: List of selected slogans
- [slogans-url]: URL of the saved text file containing the slogans

subagent2refined-repeating-json-core

G) TASKFLOW SUMMARY

1. Retrieve research findings:
   - Title task: Retrieve research findings from the provided URL
   - Description of the task: Access the research findings JSON object from the URL provided by the Niche Researcher subagent using Task 10 (Access text files already saved).
   - Input(s) involved: [URL of the research findings JSON object]
   - Output token name: [Research findings JSON object]

2. Generate slogans and puns - Iteration 1:
   - Title task: Generate humorous slogans and puns based on research findings (Iteration 1)
   - Description of the task: Use GPT-4 with custom prompts to generate a list of humorous slogans and puns related to the given niche or topic, based on the retrieved research findings.
   - Input(s) involved: [Research findings JSON object]
   - Output token name: [List of generated slogans and puns - Iteration 1]

3. Generate slogans and puns - Iteration 2:
   - Title task: Generate humorous slogans and puns based on research findings (Iteration 2)
   - Description of the task: Use GPT-4 with custom prompts to generate another list of humorous slogans and puns related to the given niche or topic, based on the retrieved research findings.
   - Input(s) involved: [Research findings JSON object]
   - Output token name: [List of generated slogans and puns - Iteration 2]

4. Generate slogans and puns - Iteration 3:
   - Title task: Generate humorous slogans and puns based on research findings (Iteration 3)
   - Description of the task: Use GPT-4 with custom prompts to generate a third list of humorous slogans and puns related to the given niche or topic, based on the retrieved research findings.
   - Input(s) involved: [Research findings JSON object]
   - Output token name: [List of generated slogans and puns - Iteration 3]

5. Combine generated slogans and puns:
   - Title task: Combine the generated slogans and puns from all iterations
   - Description of the task: Use GPT-4 to combine the lists of generated slogans and puns from Iterations 1, 2, and 3 into a single list for further processing.
   - Input(s) involved: [List of generated slogans and puns - Iteration 1], [List of generated slogans and puns - Iteration 2], [List of generated slogans and puns - Iteration 3]
   - Output token name: [Combined list of generated slogans and puns]

6. Select best slogans:
   - Title task: Select the best slogans from the combined list
   - Description of the task: Use GPT-4 as an AI-based reviewer to select the best slogans from the combined list of generated slogans and puns.
   - Input(s) involved: [Combined list of generated slogans and puns]
   - Output token name: [List of selected slogans]

7. Save selected slogans:
   - Title task: Save the selected slogans as a text file on the server
   - Description of the task: Use Task 9 (Save content to server) to save the list of selected slogans as a text file on the server and obtain a URL for accessing the file.
   - Input(s) involved: [List of selected slogans]
   - Output token name: [URL of the saved text file containing the slogans]

subagent2mermaid

SubAgent #3

claude 3 - subagent3

subagent1-refined

A) SUBAGENT SUMMARY: The Doodle Designer subagent generates doodle-style images using AI based on the provided research findings and slogans.

B) FINAL TASK OUTPUT: A collection of AI-generated doodle-style images in PNG format, sized at 1024x1024 pixels, relevant to the given niche or topic. The output will be a JSON object containing the URLs of the generated images.

C) SUBAGENT INPUT:
- URL of the research findings JSON object containing relevant keywords, phrases, and inspiration for the given niche or topic.
- URL of the text file containing the generated slogans or puns related to the given niche or topic.

E) SUBAGENT TASK SUMMARY:
1. Receive the URL of the research findings JSON object and the URL of the generated slogans text file as input.
2. Use Task 10 (Access text files already saved) to retrieve the research findings JSON object.
3. Use Task 1 (GPT4) to extract relevant keywords, phrases, and inspiration from the research findings JSON object.
4. Use Task 10 (Access text files already saved) to retrieve the generated slogans text file.
5. Use Task 1 (GPT4) to formulate specific prompts for AI image generation based on the extracted keywords, phrases, inspiration, and slogans.
6. For each slogan:
   a. Use Task 22 (Create square image with Dalle3) to generate multiple doodle-style images based on the formulated prompts.
   b. Use Task 9 (Save content to server) to save the generated images to the server.
   c. Collect the URLs of the saved images.
7. Use Task 1 (GPT4) to select the most suitable images for each slogan based on relevance and aesthetics.
8. Create a JSON object containing the URLs of the selected doodle-style images.
9. Use Task 9 (Save content to server) to save the JSON object containing the image URLs to the server.
10. Output the URL of the saved JSON object containing the selected doodle-style image URLs.

F) SILOS:
Silo 1: Research Findings Processing
- Retrieve research findings JSON object (Task 10)
- Extract relevant keywords, phrases, and inspiration (Task 1)

Silo 2: Slogan Processing
- Retrieve generated slogans text file (Task 10)

Silo 3: Prompt Formulation
- Formulate specific prompts for AI image generation (Task 1)

Silo 4: Image Generation
- For each slogan:
  - Generate multiple doodle-style images (Task 22)
  - Save generated images to the server (Task 9)
  - Collect URLs of saved images

Silo 5: Image Selection
- Select the most suitable images for each slogan (Task 1)

Silo 6: Output Generation
- Create JSON object containing selected image URLs
- Save JSON object to the server (Task 9)
- Output URL of saved JSON object

subagent1refined-repeating

A) SUBAGENT SUMMARY: The Doodle Designer subagent generates doodle-style images using AI based on the provided research findings and slogans.

F) SILOS:
Silo 1: Research Findings Processing
- Retrieve research findings JSON object (Task 10)
- Extract relevant keywords, phrases, and inspiration (Task 1)

Silo 2: Slogan Processing
- Retrieve generated slogans text file (Task 10)

Silo 3: Prompt Formulation
- Formulate specific prompts for AI image generation (Task 1)

Silo 4: Image Generation
- For each slogan:
  - Generate multiple doodle-style images (Task 22)
  - Save generated images to the server (Task 9)
  - Collect URLs of saved images

Silo 5: Image Selection
- Select the most suitable images for each slogan (Task 1)

Silo 6: Output Generation
- Create JSON object containing selected image URLs
- Save JSON object to the server (Task 9)
- Output URL of saved JSON object

G) FULL TASKS (INCLUDING REPEATING TASKS):
1. Use Task 10 (Access text files already saved) to retrieve the research findings JSON object.
2. Use Task 1 (GPT4) to extract relevant keywords, phrases, and inspiration from the research findings JSON object.
3. Use Task 10 (Access text files already saved) to retrieve the generated slogans text file.
4. Use Task 1 (GPT4) to determine the number of slogans in the text file.
5. For each slogan:
   a. Use Task 1 (GPT4) to formulate a specific prompt for AI image generation based on the extracted keywords, phrases, inspiration, and the current slogan.
   b. Use Task 22 (Create square image with Dalle3) to generate 3 doodle-style images based on the formulated prompt.
   c. For each generated image:
      i. Use Task 9 (Save content to server) to save the generated image to the server.
      ii. Collect the URL of the saved image.
6. Use Task 1 (GPT4) to select the most suitable image for each slogan based on relevance and aesthetics.
7. Create a JSON object containing the URLs of the selected doodle-style images.
8. Use Task 9 (Save content to server) to save the JSON object containing the image URLs to the server.
9. Output the URL of the saved JSON object containing the selected doodle-style image URLs.

Subagent1 task-ID-numbers3

subagent3-output-variables

Here's the updated task list with the output tokens added:

H) OUTPUT TOKENS:
- [research-findings-url]: URL of the research findings JSON object
- [research-keywords]: Extracted relevant keywords, phrases, and inspiration from the research findings
- [slogans-url]: URL of the generated slogans text file
- [num-slogans]: Number of slogans in the text file
- [slogan-prompt-1] to [slogan-prompt-N]: Formulated prompts for AI image generation for each slogan
- [slogan-1-image-1-url] to [slogan-N-image-3-url]: URLs of the generated doodle-style images for each slogan
- [slogan-1-selected-image-url] to [slogan-N-selected-image-url]: URLs of the selected doodle-style images for each slogan
- [output-json-url]: URL of the saved JSON object containing the selected doodle-style image URLs

A) SUBAGENT SUMMARY: The Doodle Designer subagent generates doodle-style images using AI based on the provided research findings and slogans.

E) SUBAGENT TASK SUMMARY:
1. Receive the URL of the research findings JSON object and the URL of the generated slogans text file as input.
2. Use Task 10 (Access text files already saved) to retrieve the research findings JSON object. [research-findings-url]
3. Use Task 1 (GPT4) to extract relevant keywords, phrases, and inspiration from the research findings JSON object. [research-keywords]
4. Use Task 10 (Access text files already saved) to retrieve the generated slogans text file. [slogans-url]
5. Use Task 1 (GPT4) to formulate specific prompts for AI image generation based on the extracted keywords, phrases, inspiration, and slogans. [slogan-prompt-1] to [slogan-prompt-N]
6. For each slogan:
   a. Use Task 22 (Create square image with Dalle3) to generate multiple doodle-style images based on the formulated prompts. [slogan-1-image-1-url] to [slogan-N-image-3-url]
   b. Use Task 9 (Save content to server) to save the generated images to the server.
   c. Collect the URLs of the saved images.
7. Use Task 1 (GPT4) to select the most suitable images for each slogan based on relevance and aesthetics. [slogan-1-selected-image-url] to [slogan-N-selected-image-url]
8. Create a JSON object containing the URLs of the selected doodle-style images.
9. Use Task 9 (Save content to server) to save the JSON object containing the image URLs to the server. [output-json-url]
10. Output the URL of the saved JSON object containing the selected doodle-style image URLs.

F) SILOS:
Silo 1: Research Findings Processing
- Retrieve research findings JSON object (Task 10) [research-findings-url]
- Extract relevant keywords, phrases, and inspiration (Task 1) [research-keywords]

Silo 2: Slogan Processing
- Retrieve generated slogans text file (Task 10) [slogans-url]

Silo 3: Prompt Formulation
- Formulate specific prompts for AI image generation (Task 1) [slogan-prompt-1] to [slogan-prompt-N]

Silo 4: Image Generation
- For each slogan:
  - Generate multiple doodle-style images (Task 22) [slogan-1-image-1-url] to [slogan-N-image-3-url]
  - Save generated images to the server (Task 9)
  - Collect URLs of saved images

Silo 5: Image Selection
- Select the most suitable images for each slogan (Task 1) [slogan-1-selected-image-url] to [slogan-N-selected-image-url]

Silo 6: Output Generation
- Create JSON object containing selected image URLs
- Save JSON object to the server (Task 9) [output-json-url]
- Output URL of saved JSON object

G) FULL TASKS (INCLUDING REPEATING TASKS):
1. Use Task 10 (Access text files already saved) to retrieve the research findings JSON object. [research-findings-url]
2. Use Task 1 (GPT4) to extract relevant keywords, phrases, and inspiration from the research findings JSON object. [research-keywords]
3. Use Task 10 (Access text files already saved) to retrieve the generated slogans text file. [slogans-url]
4. Use Task 1 (GPT4) to determine the number of slogans in the text file. [num-slogans]
5. For each slogan:
   a. Use Task 1 (GPT4) to formulate a specific prompt for AI image generation based on the extracted keywords, phrases, inspiration, and the current slogan. [slogan-prompt-1] to [slogan-prompt-N]
   b. Use Task 22 (Create square image with Dalle3) to generate 3 doodle-style images based on the formulated prompt. [slogan-1-image-1-url] to [slogan-N-image-3-url]
   c. For each generated image:
      i. Use Task 9 (Save content to server) to save the generated image to the server.
      ii. Collect the URL of the saved image.
6. Use Task 1 (GPT4) to select the most suitable image for each slogan based on relevance and aesthetics. [slogan-1-selected-image-url] to [slogan-N-selected-image-url]
7. Create a JSON object containing the URLs of the selected doodle-style images.
8. Use Task 9 (Save content to server) to save the JSON object containing the image URLs to the server. [output-json-url]
9. Output the URL of the saved JSON object containing the selected doodle-style image URLs.

subagent3refined-repeating-json-core

G) TASKFLOW SUMMARY

1. Title task: Retrieve research findings JSON object
   Description of the task: Retrieve the research findings JSON object containing relevant keywords, phrases, and inspiration for the given niche or topic using Task 10 (Access text files already saved).
   Input(s) involved: URL of the research findings JSON object
   Output token name: research_findings_json

2. Title task: Extract relevant information from research findings
   Description of the task: Use Task 1 (GPT4) to extract relevant keywords, phrases, and inspiration from the research findings JSON object.
   Input(s) involved: research_findings_json
   Output token name: extracted_keywords_phrases_inspiration

3. Title task: Retrieve generated slogans text file
   Description of the task: Retrieve the generated slogans text file containing the slogans or puns related to the given niche or topic using Task 10 (Access text files already saved).
   Input(s) involved: URL of the generated slogans text file
   Output token name: generated_slogans_text

4. Title task: Determine number of slogans
   Description of the task: Use Task 1 (GPT4) to determine the number of slogans in the generated slogans text file.
   Input(s) involved: generated_slogans_text
   Output token name: number_of_slogans

5. Title task: Formulate image generation prompt for each slogan
   Description of the task: Use Task 1 (GPT4) to formulate a specific prompt for AI image generation based on the extracted keywords, phrases, inspiration, and the current slogan.
   Input(s) involved: extracted_keywords_phrases_inspiration, generated_slogans_text, current_slogan_index
   Output token name: image_generation_prompt

6. Title task: Generate doodle-style images for each slogan
   Description of the task: Use Task 22 (Create square image with Dalle3) to generate 3 doodle-style images based on the formulated prompt for each slogan.
   Input(s) involved: image_generation_prompt
   Output token name: generated_images

7. Title task: Save generated images to the server
   Description of the task: Use Task 9 (Save content to server) to save each generated image to the server.
   Input(s) involved: generated_images
   Output token name: saved_image_urls

8. Title task: Select the most suitable image for each slogan
   Description of the task: Use Task 1 (GPT4) to select the most suitable image for each slogan based on relevance and aesthetics.
   Input(s) involved: saved_image_urls, generated_slogans_text
   Output token name: selected_image_urls

9. Title task: Create JSON object containing selected image URLs
   Description of the task: Create a JSON object containing the URLs of the selected doodle-style images.
   Input(s) involved: selected_image_urls
   Output token name: image_urls_json

10. Title task: Save JSON object to the server
    Description of the task: Use Task 9 (Save content to server) to save the JSON object containing the image URLs to the server.
    Input(s) involved: image_urls_json
    Output token name: saved_json_url

11. Title task: Output URL of saved JSON object
    Description of the task: Output the URL of the saved JSON object containing the selected doodle-style image URLs.
    Input(s) involved: saved_json_url
    Output token name: doodle_designer_output

subagent3mermaid

prompts-required

Based on the components identified, here are the unique prompts that need to be written for the AI agent workflow:

Task: Niche Topic Research
Output of the task: A list of relevant keywords, phrases, and ideas related to the niche topic, along with a summary of current top-selling birthday card designs in the niche.
Description: Write a prompt that instructs the AI to perform Google image searches for "{niche topic} birthday cards" and gather information about current top-selling designs, popular themes, and relevant keywords or phrases associated with the niche. The prompt should guide the AI to summarize its findings, providing a list of key insights and inspiration for generating slogans and designs.

Task: Slogan Generation
Output of the task: A list of humorous slogans or puns tailored to the niche topic.
Description: Write a prompt that utilizes the research findings from the previous task to generate a diverse set of funny slogans or puns related to the niche topic. The prompt should encourage the AI to use "play-on" words and phrases, and to generate multiple iterations to obtain a variety of options. Additionally, the prompt should instruct the AI to review and select the best slogans from the generated list.

Task: Image Generation Prompt
Output of the task: Descriptions or keywords to guide the AI image generator in creating relevant doodle-style images.
Description: Write a prompt that instructs the AI to generate descriptions or keywords based on the niche topic and the generated slogans. These descriptions or keywords will be used as input for the AI image generator (e.g., DALL-E) to create doodle-style images that match the desired hand-drawn appearance and are relevant to the niche topic and slogans.

Task: Bannerbear Template Creation
Output of the task: Instructions for creating 2-3 template image structures in Bannerbear.
Description: Write a prompt that guides the user in creating 2-3 template image structures using the Bannerbear tool. The prompt should instruct the user to define placeholders for "card-text-line-1", "card-text-line-2", and "doodle-image-url" within the templates, ensuring that the final birthday card images can be easily assembled by inserting the generated slogans and doodle images into the appropriate placeholders.

Task: Final Assembly Instructions
Output of the task: Step-by-step instructions for combining the generated slogans, images, and Bannerbear templates to create the final birthday card images.
Description: Write a prompt that provides clear instructions on how to assemble the final birthday card images using the generated slogans, doodle images, and Bannerbear templates. The prompt should guide the user in inserting the chosen slogans into the "card-text-line-1" and "card-text-line-2" placeholders, incorporating the generated doodle image by providing its URL to the "doodle-image-url" placeholder, and generating the final birthday card images in JPEG format, initially sized at 1000x1000 pixels.

These unique prompts will help guide the AI agent through the various tasks required to generate funny birthday cards for any given niche or topic, ensuring a structured and efficient workflow.