A short time ago, I had a fun conversation with Clyde ISD’s Saicy Lytle and Mike Neal. They are the hosts of the Ruff Draft podcast. We had a chance to discuss a blog entry I wrote about bot stacking, something I have had a lot of fun exploring using Boodle Box AI. Before I share more, I wanted to share a little more about the Ruff Draft podcast, a podcast educators may find of interest.

Ruff Draft Podcast

The hosts describe Ruff Draft in this way:

Welcome to the Ruff Draft Podcast, where education meets innovation! Hosted by Clyde CISD, this podcast spotlights the latest in educational technology, classroom creativity, and inspiring stories from our district. Whether you’re a teacher looking for tech tips, a student curious about new tools, or a community member passionate about learning, we’ve got something for you. Join us as we celebrate the Bulldog spirit and explore exciting ways to enhance education! 

In Episode 33: Bot Stacking, they were kind enough to discuss one of my blog entries, Bot Stacking Your Way to a #GenAI Dream Team. Some of the show note links:

• Bot Stacking Your Way to a GenAI Dream Team
• Stages of the AI in Education Journey
• SHINE: Lighting the Way to Responsible AI Adoption
• What Happens to Deleted AI Chats?
• OpenAI Privacy Illusion

Bot Stacking Podcast Description

The hosts describe it in this way:

In this episode, we dive into the world of bot stacking—the strategic use of multiple AI tools to boost creativity, productivity, and personalization in education. We’re joined by Miguel Guhlin from TCEA, author of the original blog post that inspired this conversation, to explore how educators can act as conductors, orchestrating specialized AI bots like ChatGPT, Perplexity, DALL·E, and others to tackle everything from lesson planning to student engagement.

Together, we break down the benefits, share real-world examples, and tackle the ethical must-knows of building your own GenAI dream team. Whether you’re new to AI or already experimenting, this episode will give you tools, ideas, and laughs to level up your classroom AI game.

The thoughts, opinions, and views expressed in this podcast are those of the individuals featured in the episode and do not necessarily reflect the official policies, positions, or views of Clyde CISD.

We had a fun chat, and I had the opportunity to listen to myself stagger, like a man walking on sand dunes, through wonderful points raised by the hosts. I hope we’ll get a chance for a second conversation.

Google Notebook LM

Explore the topic via the blog entry above, but you might also find this a fun way to dig into the content:

Outline

I. Executive Summary

This briefing document synthesizes key insights from Miguel Guhlin’s “Bot-Stacking Your Way to a GenAI Dream Team” blog post on TCEA TechNotes and related podcast questions, along with other TCEA resources on AI implementation. The central theme is “bot stacking,” an advanced strategy for leveraging multiple Generative AI (GenAI) models in sequence to achieve more nuanced, accurate, and educationally sound results than a single AI tool could provide. This approach emphasizes a “human-in-the-loop” model, where educators act as “orchestra conductors,” directing AI tools to amplify their impact rather than replacing their expertise. The sources also touch upon practical applications in education, ethical considerations, and the future evolution of GenAI in learning environments.

II. Key Concepts and Themes

A. What is “Bot Stacking”?

Bot stacking is a strategy for combining multiple AI models in a strategic workflow, where each bot contributes its unique strengths to achieve a specific educational goal. It’s analogous to a “collaborative lesson plan” or a “multi-strategy approach” in teaching. Instead of relying on a single AI for all tasks, educators leverage specialized GenAI models in sequence, often within a single platform like BoodleBox AI, to create a “suite of bots, each with their own specialty.”

• Quote: “It is a strategy for combining multiple AI models to enhance teaching, learning, and educational leadership.” – “Bot Stacking Your Way to a #GenAI Dream Team”
• Quote: “Instead of asking one AI (e.g. ChatGPT, Gemini, Claude) to handle everything from research to assessment, you create a strategic workflow.” – “Bot Stacking Your Way to a #GenAI Dream Team”
• Analogy: Guhlin compares it to his teacher mentor Ms. Gonzaga’s multi-layered approach to writing workshops, contrasting it with a “one-activity-at-a-time” method. “Her orchestration of a lesson didn’t rely on a single strategy. It seamlessly blended questioning techniques, visual aids, collaborative activities, and formative assessments. Each element served a specific purpose. This multi-strategy approach resulted in a far more powerful effect than any single approach could achieve.” – “Bot Stacking Your Way to a #GenAI Dream Team”

B. The “GenAI Dream Team” – Specialized Bots for Educational Tasks

Guhlin categorizes various AI models based on their specialized capabilities, highlighting the importance of knowing “your team.” The ideal setup allows these models to work together seamlessly, eliminating the need for constant copying-and-pasting between different platforms.

• Visual Designers (Instructional Material Creators): Flux 1.1 Pro, Ideogram V3, DALL-E 3, Imagen 3, Stable Diffusion 3. These are used for creating engaging visual learning materials like realistic images, infographics, artistic visuals, and photorealistic content.
• Deep Thinkers (Curriculum Specialists): Gemini 2.5 Pro, Claude 4.0 Opus, OpenAI o3, LLAMA-4-Maverick. These reasoning models are “wonderful to ‘bot stack'” for complex learning experiences, unit plans, discussion prompts, STEM problem sets, and privacy-conscious materials.
• Researchers (Professional Development Partners): Perplexity, Semantic Scholar. Essential for finding current educational research, teaching strategies, policy updates, and academic papers with proper citations.
• All-Rounders (Everyday Teaching Assistants): GPT-4.1, GPT-4o, Claude 4.0 Sonnet. Versatile models for daily tasks like polishing lesson plans, creating worksheets, exit tickets, and brainstorming activities.
• Specialists (Educational Experts): Mistral Codestral (for coding) and PromptBot (for crafting better AI prompts).

C. The “Human-in-the-Loop” Approach

A core tenet of bot stacking is that AI tools are amplifiers, not replacements, for human expertise. The educator remains the “orchestra conductor” and “instructional leader,” directing the process and critically evaluating AI outputs.

• Quote: “You, as the orchestra conductor, remain the instructional leader. You direct the process and ensure alignment with learning objectives.” – “Bot Stacking Your Way to a #GenAI Dream Team”
• Quote: “This ‘human-in-the-loop approach’ produces resources that are more nuanced, accurate, and educationally sound than relying on a single AI model.” – “Bot Stacking Your Way to a #GenAI Dream Team”
• Emphasis on Critical Thinking: “effective AI use requires critical thinking and ethical consideration.” Educators must “Check your own biases,” “Verify the accuracy,” “Look for corroborating educational research,” and “Ensure alignment with curriculum standards and best practices.” – “Bot Stacking Your Way to a #GenAI Dream Team”

D. Practical Workflows and Frameworks

The article provides concrete examples of bot stacking in action and how it aligns with pedagogical frameworks:

1. Curriculum Development Project Workflow:Perplexity: Research evidence-based practices for teaching a specific topic.
2. Gemini 2.5 Pro: Process and categorize the research findings.
3. Claude 4.0 Opus: Create a detailed intervention plan based on the analyzed strategies.
4. Creating Accessible Learning Materials Workflow:PromptBot: Get advice on structuring prompts for differentiated content.
5. Claude 4.0 Sonnet: Generate varied reading materials at different grade levels.
6. Ideogram V3: Create visual supports like diagrams.

• Integration with PRISM Framework: Using different bots for Patterns, Reasoning, Ideas, Situation, and Methods.
• Alignment with SOLO Taxonomy: Stacking bots to create activities progressing from Unistructural to Extended Abstract levels.

E. The CRAFT Method for Effective Prompts

To maximize AI output quality, Guhlin advocates for the CRAFT method:

• C – Clarify your objective.
• R – Role-play for the AI (assign a persona).
• A – Add context and constraints.
• F – Format the output.
• T – Test and Tweak.

F. Ethics, Privacy, and Safety

The sources highlight ongoing concerns and the need for guardrails:

• Guardrails/Policies: Schools need policies before encouraging GenAI tool use. (Podcast Question 12)
• Privacy Concerns: Awareness of potential “red flags” when using GenAI tools in education. (Podcast Question 13)
• Data Safeguarding: BoodleBox AI is mentioned for “safeguarding data” and FERPA compliance, with emphasis on deleted chats actually being removed.
• Bias: Educators are reminded to “Check your own biases (remember, AI wants to please you so maintain professional distance).”

G. Future of GenAI in Education

• Evolution: GenAI tools are expected to evolve significantly in the next year or two, especially in education. (Podcast Question 14)
• Essential Tool: Guhlin poses whether AI will become “as essential as Google Workspace or Microsoft 365 in schools.” (Podcast Question 16)
• Student Use: The potential for bot stacking for students is considered. (Podcast Question 10)
• Creativity/Originality Concerns: Addresses the perennial concern about AI replacing human creativity or originality. (Podcast Question 11)

III. Practical Advice for Educators

• Start Small: Identify a “time-consuming, known process or task” and begin with a simple stack of 2-3 complementary AI models (e.g., Perplexity → Claude 4.0 Sonnet).
• Document and Reflect: Keep track of effective prompts and workflows, and use frameworks like PRISM to analyze and improve.
• Focus on Relationships: The ultimate goal is to “Focus more energy on what matters, such as building relationships with students and creating rich learning experiences.”
• Explore Platforms: Platforms like BoodleBox AI are highlighted for their ability to integrate various AI models, manage knowledge banks, and offer “Coach Mode” for prompt improvement, creating a unified environment for bot stacking.

IV. Conclusion

Bot stacking represents a sophisticated and effective approach to integrating GenAI into education. By strategically combining specialized AI tools under the direction of an informed educator, schools can unlock significant enhancements in teaching, learning, and productivity. This method prioritizes human agency, critical thinking, and ethical considerations, ensuring that technology serves to amplify, rather than diminish, the vital role of educators in the learning process. As AI continues to evolve, understanding and implementing bot stacking will be crucial for navigating the future of education.

Frequently Asked Questions

1. What is “bot stacking” in the context of Generative AI (GenAI)?

Bot stacking is a strategic approach to using multiple Generative AI models in tandem, rather than relying on a single AI tool for all tasks. It involves creating a workflow where different AI models, each with their own specialized capabilities, work sequentially or collaboratively to achieve a specific educational or professional goal. The core idea is to leverage the unique strengths of various AI tools, much like a teacher orchestrates different strategies and resources in a lesson plan, to produce more nuanced, accurate, and effective results. This method often involves feeding the output of one AI model as context or input to another, creating a “dream team” of specialized bots.

2. How does bot stacking enhance workflow and productivity, particularly in an educational setting?

Bot stacking significantly improves workflow and productivity by optimizing the use of AI tools for specific tasks. Instead of trying to force one AI to handle everything (e.g., research, content creation, visual design), bot stacking allows users to direct specialized models for each part of a process. For instance, in curriculum development, one AI might gather research, another analyze it, and a third transform it into grade-appropriate materials. This eliminates the need for manual copying and pasting between different AI platforms, saving time and effort. By leveraging each bot’s strengths, the overall output is more refined, accurate, and tailored to specific educational needs, leading to higher quality results with greater efficiency.

3. What are some examples of specialized AI models and their roles within a “bot stacking” framework for educators?

In a bot stacking framework, AI models are categorized by their specialized capabilities to form an educational AI dream team. Examples include:

• Visual Designers (Instructional Material Creators): Models like Flux 1.1 Pro, Ideogram V3, DALL-E 3, Imagen 3, and Stable Diffusion 3 are used for creating engaging visual learning materials such as realistic images, infographics, artistic visuals, or photorealistic content.
• Deep Thinkers (Curriculum Specialists): Models such as Gemini 2.5 Pro, Claude 4.0 Opus, OpenAI o3, and LLAMA-4-Maverick excel at complex tasks like processing large texts, developing comprehensive unit plans, creating detailed problem sets, or handling privacy-conscious materials.
• Researchers (Professional Development Partners): Perplexity and Semantic Scholar are ideal for finding current educational research, evidence-based practices, and academic papers with proper citations.
• All-Rounders (Everyday Teaching Assistants): GPT-4.1, GPT-4o, and Claude 4.0 Sonnet are versatile for daily tasks like polishing lesson plans, generating exit tickets, or brainstorming engagement activities.
• Specialists (Educational Experts): Mistral Codestral is invaluable for computer science teachers for coding tasks, while PromptBot helps educators craft better prompts for other AI models.

4. What is the CRAFT method, and how does it contribute to effective bot stacking?

The CRAFT method is a framework for writing effective prompts when interacting with AI models, crucial for successful bot stacking. It stands for:

• C – Clarify: Define the specific learning goal or desired outcome clearly.
• R – Role-play: Assign the AI a specific persona (e.g., “You are a 4th-grade math specialist”).
• A – Add: Provide context and constraints (grade level, curriculum standards, student needs).
• F – Format: Specify how the output should be presented (lesson plan, rubric, bulleted list).
• T – Test: Evaluate the AI’s response and refine the prompt as needed through iteration.

By using the CRAFT method, educators can ensure that the initial input to each bot in a stack is precise and well-structured, leading to more accurate, relevant, and high-quality outputs, thereby maximizing the benefits of bot stacking.

5. How can bot stacking be integrated with established educational frameworks like PRISM and SOLO Taxonomy?

Bot stacking’s power is amplified when aligned with pedagogical frameworks:

• PRISM Framework (Patterns, Reasoning, Ideas, Situation, Methods): Different AI models can be used to address each component. For instance, Perplexity or Semantic Scholar for “Patterns” (identifying research patterns), Claude 4.0 Opus for “Reasoning” (analyzing connections), GPT-4.1 for “Ideas” (generating applications), Gemini 2.5 Pro for “Situation” (analyzing context), and Claude 4.0 Sonnet for “Methods” (creating assessment tools). This ensures a diverse range of outputs and perspectives across the framework.
• SOLO Taxonomy (Structure of Observed Learning Outcomes): Bot stacking can help create materials progressing through the SOLO levels. GPT-4o can be used for “Unistructural” (basic identification), Claude 4.0 Sonnet for “Multistructural” (connecting multiple concepts), Claude 4.0 Opus for “Relational” (synthesis activities), and OpenAI o3 for “Extended Abstract” (transfer learning challenges). This structured approach helps design differentiated learning experiences that cater to varying levels of cognitive complexity.

6. What ethical considerations and “guardrails” should schools consider when encouraging the use of GenAI tools?

When encouraging GenAI tools, schools should prioritize ethical use, privacy, and safety. Key guardrails and policies include:

• Critical Thinking and Ethical Consideration: Emphasizing that effective AI use requires critical evaluation and awareness of potential biases.
• Data Privacy: Ensuring the chosen AI tools comply with privacy regulations like FERPA, and that user data, especially student information, is safeguarded and not used for training models without consent.
• Transparency: Understanding how AI tools handle deleted chats and data retention.
• Human Oversight: Reinforcing that AI tools amplify human expertise rather than replacing it. Teachers remain the instructional leaders, responsible for assessing AI outputs for accuracy, bias, and alignment with curriculum standards.
• Verification of Accuracy: Educating users to always verify factual claims and corroborate information from AI with reliable educational research.

7. What advice is given to educators hesitant to adopt multiple AI tools and how can they start building their AI practice?

For hesitant educators, the advice is to start small and incrementally build their AI practice:

1. Identify a Time-Consuming Task: Choose a specific, repetitive task that consumes significant prep time, like creating differentiated reading materials or formative assessments.
2. Select 2-3 Complementary AI Models: Begin with a simple “stack,” such as using Perplexity for research and then Claude 4.0 Sonnet for content generation.
3. Document the Process: Keep a record of effective prompts and workflows to build a personal library of successful strategies.
4. Reflect and Refine: Use frameworks like PRISM to analyze results and continuously improve the approach.

The emphasis is on an iterative, experimental process, much like trying out different perspectives to solve a problem. This helps educators gain comfort and expertise without feeling overwhelmed.

8. How is the “human-in-the-loop” approach emphasized in bot stacking, and why is it crucial?

The “human-in-the-loop” approach is central to bot stacking and is crucial because it ensures that AI tools amplify human intelligence and expertise rather than replacing them. As the “orchestra conductor,” the human user directs the process, making critical decisions at each stage of the AI workflow. This involves:

• Defining Objectives: Clearly outlining the specific educational goal.
• Selecting and Directing Bots: Choosing the right specialized AI models for each task and providing precise prompts.
• Injecting Insights: Including personal knowledge and prompts to customize and refine AI-generated content.
• Critical Evaluation: Continuously assessing the AI’s outputs for accuracy, bias, relevance, and alignment with curriculum standards and student needs.

This active involvement ensures that the final resources are nuanced, accurate, educationally sound, and reflect the pedagogical expertise of the educator, maintaining the focus on building relationships with students and creating rich learning experiences.