Quartz 4

Workshop - Gemini Co-Prompting

18 min read

Summary for 002-gemini-collaborative-co-prompting-session.mp4

Segment 1

Overarching Principles

  • AI as an Augmentative Partner: Utilize AI not to replace human creativity or critical thinking, but to augment it. AI handles the formulaic, time-consuming tasks, freeing up cognitive load for high-value strategic work.
  • Systemize Your Expertise: Document and structure your unique processes and workflows. This allows you to delegate the initial drafting and organization to AI, dramatically increasing efficiency and consistency.
  • Voice-to-Task Workflow: The most efficient way to create structured content is to convert unstructured, spoken thoughts (brain dumps, meeting transcripts) into polished, actionable documents using AI as an intermediary.
  • Iterative Prompting over Manual Editing: Instead of manually rewriting AI-generated drafts, provide conversational feedback to the AI to refine the output. This “teaches” the AI your style and saves significant editing time.
  • Develop a “Spiky Point of View”: To establish thought leadership, articulate a unique and specific perspective. This involves identifying commonly held industry beliefs you challenge and sharing unconventional insights from your experience.

Frameworks

1. Functional Technology Framework

A three-pronged approach for assessing and planning technology initiatives within a business.

  • Pillar 1: Derisk: Proactively identify and create projects to mitigate all areas of technological and operational risk to the business.
  • Pillar 2: Unclog: Identify and remove all barriers to sustainable growth. This is diagnosed by asking, “If the business were to grow 10x, what would break?” This pillar focuses on eliminating obstacles in people, processes, and technology.
  • Pillar 3: Scale: Identify and execute projects specifically designed to grow the business and expand its capabilities.

2. AI-Powered Project Planning Workflow

A method for rapidly converting client conversations or internal brain dumps into structured, actionable project plans or proposals.

  1. Capture: Record the audio of a discovery meeting, client call, or a personal “brain dump” on project requirements.
  2. Transcribe: Use an automated transcription service (e.g., Otter.ai) to convert the audio into a raw text document.
  3. Upload & Synthesize: Upload the text transcript into a large language model like Gemini CLI / Maestro.
  4. Prompt & Structure: Instruct the AI to analyze the transcript and populate a pre-defined project plan template with the relevant information (e.g., project name, description, “Why is it important,” ideal outcome, etc.).

3. Custom AI Persona (The “Feature Creator”)

A technique for creating a specialized AI assistant within a Gemini CLI / Maestro “Project” by using the “Custom Instructions” feature. This ensures consistent, high-quality output for recurring, complex tasks.

  • Define the Persona: Assign a specific role to the AI (e.g., “As an Agile Feature Guide…”).
  • Set the Tone: Dictate the desired communication style (e.g., “playful tone with approachable communication, ensuring clarity and friendliness”).
  • Establish the Process: Provide a clear, step-by-step sequence of tasks the AI must perform (e.g., 1. Scan requirements, 2. Build features, 3. Build stories, 4. Size stories, etc.).
  • Provide Core Data: Embed essential, reusable information into the instructions, such as team member roles, hourly rates, and weekly capacity, to enable the AI to perform calculations like cost and delivery estimates.

Actionable Flight Plan

  1. Set Up Your Content Creation Engine:

    • Use a transcription tool like Otter.ai to record client meetings, personal thoughts, and any spoken content.
    • Create dedicated “Projects” within Gemini CLI / Maestro for different content types (e.g., LinkedIn Articles, Client Proposals, Project Planning).

  2. Define Your AI’s Persona & Voice:

    • In each Gemini CLI / Maestro Project, use the “Custom Instructions” to define how the AI should respond.
    • Specify the desired professional tone (e.g., relatable, inspiring, practical, actionable).
    • Provide stylistic rules (e.g., “Always write in the active voice,” “Use short, scannable paragraphs,” “Vary sentence length for interest and readability”).
    • Upload past articles, emails, or documents you’ve written by hand to give the AI a corpus of your work to learn from.

  3. Generate Your First Technical Project Plan:

    • Record a brain dump or a client discovery call where you discuss project needs.
    • Transcribe the recording using Otter.ai.
    • Upload the transcript to Gemini CLI / Maestro.
    • Prompt Gemini CLI / Maestro: “Using the attached transcript, fill out the following project plan structure for each initiative discussed: [Paste the column headers from the Technology Project Planner: Project Name, Category, Team, Description, Why is it Important, Ideal Outcome, Best Result, Worst Result].”

  4. Create a Specialized “Feature Creator” Project:

    • Create a new Project in Gemini CLI / Maestro named “Feature Creator” or similar.
    • Copy the detailed custom instructions shared by George Reyna (starting with “As an Agile Feature Guide…”) into the project’s custom instructions.
    • Customize the prompt with your specific team members, roles, hourly rates, and weekly capacity.
    • To use it, start a new chat within the project and feed it the requirements for a new feature (e.g., “Build me a computer that can mine bitcoin”).

  5. Develop Your Thought Leadership Content:

    • Use the “Develop Your Unique Perspective” worksheet to answer the reflection questions.
    • Record your answers using a transcription app.
    • Upload the transcript to your “Thought Leadership” project in Gemini CLI / Maestro.
    • Prompt Gemini CLI / Maestro: “Based on the attached transcript, generate 5 potential LinkedIn article titles and outlines that reflect a ‘spiky point of view’.”
    • Select your favorite outline and prompt Gemini CLI / Maestro: “Craft an article based on the first outline. Write it in my voice.”
    • Refine the draft by giving conversational feedback (e.g., “Make this crisper,” or “Add a personal story to the second paragraph.”).

Segment 2

Overarching Principles

  • Iterative Collaboration: Treat AI as a collaborative partner, not a one-shot content generator. The best results come from a back-and-forth process of drafting, feedback, and refinement.
  • You Are the Editor-in-Chief: The user’s role is to provide strategic direction, stylistic guidance, and final approval. The AI executes, but the user’s judgment and edits are what make the content unique and effective.
  • Train Your AI Ghostwriter: Consistently providing feedback and, most importantly, showing the AI your final, human-edited versions creates a powerful feedback loop that trains the model on your specific voice, tone, and preferences over time.
  • Maximize Asset Leverage: Don’t start from a blank page. Use the AI to repurpose, adapt, and expand on existing assets like audio transcripts, slide decks, previous training materials, and email sequences.
  • Context is King: Utilize long, continuous chat threads for a series of related tasks (e.g., creating multiple articles from a content plan). This allows the AI to maintain context and apply learnings from one task to the next.

Frameworks

  1. The “Good, But…” Feedback Loop Framework

    • Generate: Get an initial draft from the AI based on a prompt.
    • Critique: Provide specific, qualitative feedback. Examples:
      • “This feels a little too long-winded.”
      • “This is way better but now it’s too choppy.”
      • “This part seems a little off.”
    • Guide with Models: Direct the AI’s style by referencing well-known archetypes. Example:
      • “Can we have a mix of both? Think Seth Godin or Tim Ferriss.”
    • Provide the Gold Standard: After you perform your final manual edits, paste the perfected version back to the AI. Example:
      • “Here is the final version of the article I posted. Please note the changes I made and flow this into future articles you generate.”

  2. The “Asset Transformation” Framework

    • Upload Source Material: Upload an existing asset (PDF, DOCX, TXT) like a workshop slide deck, an audio transcript, or an old email sequence.
    • Define the New Format: Give a clear instruction on what the new output should be.
    • Set Transformation Parameters: Provide specific constraints and requirements for the new version. Example:
      • Prompt: “I have attached a PDF of slides from our sister company… I want to make all the updates we need to the branding and language to frame this for CTOs instead of CMOs… Can you provide me a slide-by-slide breakdown on what content needs to change based on these parameters?”

  3. The “Agenda & Run-of-Show” Framework

    • Brain Dump Agenda Items: List all the topics and segments you want to cover in a meeting, workshop, or webinar.
    • Request Time Allocation: Ask the AI to assign a duration for each item based on a total time for the session.
    • Request Timestamping: Ask the AI to convert the time allocations into a specific run-of-show with start and end times. Example:
      • Prompt: “Can you give me this to me with timestamps starting at 11:00 am?”

Actionable Flight Plan

  1. Establish Your “Personal Brand” Project:

    • Create a dedicated “Project” in Gemini CLI / Maestro.
    • Upload 3-5 high-quality examples of your writing (blog posts, LinkedIn articles, your professional bio) to the “Project Knowledge” to give the AI a baseline for your voice.

  2. Generate a Content Strategy:

    • Start a new chat within your project.
    • Prompt the AI to generate a list of 5-10 article titles and outlines based on your core expertise and the documents you uploaded.

  3. Co-Create and Train (The First Article):

    • Ask Gemini CLI / Maestro to write the first article from the list.
    • Review the draft and provide specific feedback using the “Good, But…” Framework. Go through at least two rounds of feedback.
    • Perform a final, manual gut-check and edit the article to your satisfaction.
    • Paste the final, published version back into the chat and instruct Gemini CLI / Maestro to “learn from these changes” for future articles.

  4. Batch and Accelerate:

    • In the same chat, prompt Gemini CLI / Maestro to “move on to #2” from the content strategy list.
    • Observe if the new draft is closer to your style. Provide minimal feedback if needed.
    • Repeat this process for the remaining articles in your content plan.

  5. Repurpose and Multiply:

    • Find an existing asset (e.g., a transcript of you talking from Otter.ai, an old slide deck).
    • Upload the asset to a new chat in your project.
    • Prompt Gemini CLI / Maestro to transform it into a new format (e.g., “Turn this transcript into a LinkedIn article,” or “Create a 5-part email sequence based on these slides”).

Segment 3

Overarching Principles

  • Codify Expertise into a System: Deconstruct complex processes (like starting a business) into a series of discrete, sequential, and documented steps. This transforms personal knowledge into a repeatable, scalable framework.
  • Leverage AI as a Guided Workflow Engine: Instead of using AI for one-off creative tasks, program it with custom instructions and a knowledge base to act as an interactive guide, leading a user through your codified process.
  • Structured AI Interaction for Consistent Outcomes: By feeding the AI numbered documents and instructing it to process them sequentially, you create a structured, predictable, and robust workflow that ensures all necessary steps are completed in the correct order.
  • Accelerate Development Cycles: Systematizing ideation and planning through a guided AI workflow can dramatically reduce the time it takes to move from a high-level concept to a fully articulated, actionable plan, condensing tasks that might take months or years into a much shorter timeframe.
  • Use AI for Emotional and Relational Intelligence: AI can be used as a “translator” or “sounding board” to reframe emotionally charged or difficult communications into more diplomatic, strategic, and effective language, helping to manage stakeholder relationships and navigate complex conversations.

Frameworks

The “Blueprint” Workflow Framework

This is a structured, end-to-end system for business or product creation using an AI assistant guided by a pre-defined knowledge base.

  1. Knowledge Base (The Inputs): A collection of separate, numbered documents, with each document detailing a specific stage of the business creation process. The numbered sequence defines the workflow path.

    • 00. The Idea
    • 01. Your Avatar
    • 02. The Message
    • 03. The Offer
    • 04. The Hooks
    • 05. The Lead Magnet
    • 06. Site Pages
    • 07. Auth Amplify
    • 08. Emails
    • 09. Enroll Script
    • 10. Webinars
    • 11. Roadmap
    • 12. Get Traffic
    • 13. Retarget & Metrics

  2. AI Role & Rules (The Process): Custom instructions that define the AI’s persona and operational rules within the project.

    • Persona: An “AI assistant.”
    • Core Task: To guide users through a series of stages based on the uploaded documents.
    • Rule 1: Sequential Approach: Process each document in the exact numerical order listed. Do not skip steps or make assumptions about future steps.
    • Rule 2: Clarify First: If any instruction is unclear, ask for clarification before proceeding.

  3. Articulated Vision (The Output): A comprehensive, fully-documented business and marketing plan that has been methodically developed by following the structured workflow.

Actionable Flight Plan

  1. Deconstruct Your Process: Identify a complex, multi-step process you are an expert in (e.g., starting a business, launching a product, planning a marketing campaign).
  2. Document Each Step: Break down the process into discrete, logical stages. Create a separate text document for each stage, detailing the objectives, tasks, and considerations for that stage. Number these documents sequentially (e.g., 01_Discovery.txt, 02_Strategy.txt, etc.).
  3. Create an AI Project: In a tool like Gemini CLI / Maestro.ai, initiate a new “Project.”
  4. Upload Knowledge Files: Upload all the numbered documents from Step 2 into the project’s “Project Knowledge” or context section.
  5. Set Custom Instructions: Define the AI’s role. Instruct it to act as a guide and to process the uploaded files sequentially, one by one, walking you through each step.
  6. Initiate and Execute the Workflow: Start a chat and allow the AI to guide you through your own process, from the first document to the last, to flesh out a new idea or project.
  7. Refine Communication with AI:
    • For external communication (e.g., emails, proposals), paste your draft into the AI.
    • Provide context about the recipient and your goal (e.g., “This is for a non-technical CEO; make it simpler and focus on business outcomes”).
    • Ask the AI to rewrite the text in a specific tone (e.g., “more compassionate,” “more diplomatic,” “more persuasive”).
  8. Develop Your Writing Voice:
    • Provide the AI with a piece of your own writing.
    • Provide examples of writing styles you admire from other authors.
    • Ask the AI to rewrite your piece in a hybrid style that blends your voice with elements of the others, helping you discover new ways to articulate your ideas.

Segment 4

Overarching Principles

  • Personalization at Scale: Transition from generic, one-size-fits-all templates to personalized outreach for multiple prospects efficiently.
  • Problem-Centric Qualification: Initiate contact by asking a question about a specific, likely pain point to immediately qualify the prospect’s need and demonstrate relevance.
  • Iterative Refinement: Use a series of follow-up prompts to progressively refine and specialize the AI’s output, rather than expecting a perfect result from a single command.
  • Accelerate, Don’t Abdicate: The goal of AI is to speed up the creative and personalization process, not to eliminate the human element. The user still reviews, refines, and adds the final human touch.

Frameworks

The A/B/C Test Variation Framework

This framework generates multiple distinct email versions from a single base template, each targeting a different pain point.

  1. Isolate Core Challenges: Identify 2-3 specific, high-impact problems your service solves (e.g., Outdated PMS, Cybersecurity, Guest Experience).
  2. Generate Variations: Instruct the AI to create a unique email variation for each challenge.
  3. Deploy and Measure: Send the different variations to similar prospects and track which one receives the best response rate.

The “Open-Ended Question” Email Structure

This is a framework for structuring a cold outreach email to maximize engagement.

  1. Hook (Open-Ended Question): Start with a short question directly related to the prospect’s potential pain point.
    • Example: “How much time does your staff spend wrestling with your property management system each day?”
  2. Relate (Value Proposition): State who you are and establish immediate authority by showing you solve this exact problem for similar companies.
    • Example: “As a fractional CTO for mid-sized resorts, I’ve seen firsthand how outdated PMS can drain productivity…”
  3. Prove (Solution Bullets): List 2-3 tangible ways you address the problem.
    • Example: “Implementing modern, cloud-based PMS solutions”
  4. Outcome (The Result): Briefly state the positive business outcome of your work.
    • Example: “The result? Happier guests, more efficient staff, and a stronger bottom line.”
  5. Call to Action (Low-Friction Hook): End with another engaging, low-commitment question that invites further discussion.
    • Example: “Curious how a PMS upgrade could impact your efficiency and guest satisfaction scores?”

Actionable Flight Plan

  1. Draft Base Template: Write a foundational email that introduces your service, expertise, and target audience.
  2. Identify 3 Key Challenges: From your base template, extract the three most common and critical problems you solve for your clients.
  3. Generate Variations: Use the following prompt, inserting your base email:

    Prepare 3 variations of these emails. Start each email with a short, open-ended question with one of these specific challenges. End with an open-ended question that hooks the reader - a busy CEO - like, “Interested in discussing further?”

  4. Refine with Personalization: Provide the AI with a list of specific prospects and use a follow-up prompt to tailor the generated templates:

    Now personalize the first message template for the following individuals, focusing on their specific roles and potential challenges their resorts might face:

    • John Smith, Founder and CEO, Hilton Head Resort
    • Sally Raphael, Regional General Manager, Paradise Point
    • Ursula George, Chief Operating Officer, Kilauea Falls Resort
  5. Develop Follow-Up Sequence: Instruct the AI to build a multi-touch campaign based on the initial emails:

    Now Develop 3 follow-up emails and include a placeholder for my calendly link in #2 and #3.

  6. Review and Execute: Proofread the personalized, multi-touch sequences, add final human refinements, and deploy your outreach campaign.

Segment 5

Overarching Principles

  1. Tool Stacking for Maximum Efficiency: No single AI tool is a silver bullet. The most effective strategy involves “stacking” different tools based on their unique strengths—using web-connected AI like Perplexity for real-time research and data gathering, and using large-context AI like Gemini CLI / Maestro for deep analysis of provided documents.
  2. Automated Research as a Force Multiplier: Leverage AI to automate the tedious aspects of market research, lead qualification, and competitive analysis. AI can rapidly gather, synthesize, and structure public data from sources like LinkedIn, news articles, and financial reports, saving significant time.
  3. AI as a Content Idea Engine: AI search tools can do more than just answer questions; they can inspire new content. Features like Perplexity’s suggested follow-up questions provide ready-made prompts for blog posts, social media content, and email newsletters that directly address audience curiosity.
  4. Community Engagement as a Long-Term Strategy: Beyond data-driven tools, establishing authority in niche communities (like Facebook Groups) is a powerful method for generating inbound leads. This involves consistently providing value and positioning oneself as a go-to expert over time.

Frameworks

1. The Research & Synthesize AI Workflow

This framework outlines a process for using different AIs to gather and analyze information efficiently.

  • Phase 1: Data Scoping
    • Identify Information Need: Clearly define the question you need to answer.
    • Determine Data Source: Is the required information static (e.g., in documents you possess) or dynamic (requiring real-time web access)?
  • Phase 2: Tool Selection & Execution
    • For Dynamic Data: Use a web-connected AI (e.g., Perplexity) to search, gather, and aggregate current information from the internet.
    • For Static Data: Use a large-context-window AI (e.g., Gemini CLI / Maestro) to upload documents and perform deep analysis, summarization, or data extraction.
  • Phase 3: Output Structuring
    • Command the Format: Explicitly instruct the AI to organize the final output into a practical format (e.g., “put it in a table,” “create a bulleted list,” “write a summary paragraph”).

2. The Community Authority Lead Generation Model

This framework details a non-salesy, value-driven approach to attracting clients from online communities.

  • Step 1: Identify & Infiltrate
    • Find and join the specific online communities (Facebook Groups, Slack channels, forums) where your ideal prospects are most active.
  • Step 2: Provide Consistent Value
    • Regularly answer questions, share helpful insights, and post valuable content (tips, “cheat sheets,” summaries of new trends) without an immediate sales pitch.
  • Step 3: Establish Authority
    • Through consistent, high-quality contributions, position yourself as a knowledgeable and trustworthy expert in your field.
  • Step 4: Attract Inbound Leads
    • As your reputation grows, community members will naturally begin to reach out to you directly for paid help, creating a sustainable and organic lead flow.

Actionable Flight Plan

  1. Conduct Prospect Research with Perplexity: Use Perplexity AI to research target companies or roles.
    • Prompt Example: “Based on public LinkedIn data, list all company board affiliations for CTOs in the SaaS industry. For any publicly traded companies, find their current market capitalization and present the results in a markdown table.”
  2. Automate News & Trend Monitoring: Use Perplexity to create quick summaries of important industry news to stay informed and generate content.
    • Prompt Example: “Summarize the top 5 announcements from the latest Apple product event, citing your sources.”
  3. Generate Content from AI-Suggested Questions: Look for the “Related” or “Follow-up” questions generated by Perplexity after a search and use them as titles or topics for your next blog post or email.
  4. Enrich Lead Lists with Multiple Sources: Use a data provider like Apollo.io, Seamless.AI, or ZoomInfo to build initial lead lists. Manually or programmatically cross-reference this data with LinkedIn to verify that contacts are still in their listed roles.
  5. Join Niche Communities: Identify and join 2-3 active Facebook Groups, LinkedIn Groups, or Slack communities relevant to your target customer.
  6. Schedule Community Engagement: Block 30 minutes twice a week on your calendar to actively participate in your chosen communities by answering questions and sharing valuable insights.
  7. Explore the Gemini CLI / Maestro Prompt Library: Visit Anthropic’s official Gemini CLI / Maestro Prompt Library to discover new, pre-built prompts and learn advanced techniques for getting better results from the AI.

Graph View

Backlinks