2025-06-05 AI Mastermind

Table of Contents

Session Overview

This session was a hybrid event — part masterclass open to prospective members, part pitch for the newly-launched AI Leaders Mastermind. Lou delivered his most advanced current thinking on AI prompting to an audience that included both experienced members (from the former Collective Intelligence Agency group) and new attendees.

The core teaching centered on what Lou calls “process prompts” — a category above meta prompts in the prompting hierarchy that produces dynamic, interview-driven AI systems rather than fixed-use-case outputs. The distinction he drove home: most people build prompts that solve specific problems; process prompts build systems that can dynamically solve any version of a class of problem. Lou demonstrated this live using Claude and ChatGPT, showing both a writing style profiler built through conversation with the AI, and an “infinite prompt meta-generator” based on Michael Simmons’ infinite prompting framework.

The second half of the session was devoted to introducing the AI Leaders Mastermind structure: two tiers (Explorers and Leaders), small intimate group (12-20 members), weekly live calls, fly-on-the-wall sessions, and a referral-based growth model. Lou described his philosophy of keeping groups small and closely-knit to maximize the quality of peer support.

High-Signal Moments

  • Lou introduces the “process prompts” term for the first time — AI systems that infer variables dynamically and think before executing, rather than fixed-instruction prompts
  • Demonstration of a multi-tiered writing style profiler that goes beyond “analyze my writing” to include psycholinguistics, Big Five personality inference, and stylometric signals — all generated through conversation with the AI
  • Introduction of “infinite prompting” — each reasoning step written to a separate artifact, producing dramatically deeper thinking per step; takes 10 minutes per run but produces qualitatively superior output
  • Lou’s key reframe: “I’m talking to someone who’s going to architect an asset for me” — the mental model shift from prompt-user to system-designer
  • The meta-prompting principle: have the AI generate the process prompt, not just the output; teach it your framework for thinking about the task, not just the task
  • Session closes with Lou announcing the mastermind structure, emphasizing the “givers community” philosophy and the anti-guru, accessible-pricing philosophy

Open Questions

  1. What is the right use case selection criteria for when to invest in a process prompt versus using a simpler specific prompt?
  2. How do infinite prompts scale for clients who are not technical or patient with longer AI run times?
  3. What happens when the AI’s “best practices” within a process prompt conflict with the coach’s own methodology?
  4. How does Lou’s process prompt approach interact with Claude’s project context and memory features?
  5. Can process prompts be packaged into shareable templates, or are they inherently too personal to the creator’s judgment model?

Suggested Follow-Through

  1. For Lou: Finalize the Mastermind onboarding materials, telegram setup, and initial welcome sequence
  2. For members: Watch the session replay and identify one recurring task to rebuild as a process prompt using the demonstrated methodology
  3. For Lou: Create the “style profiler” and “content brief generator” as shareable templates for mastermind members
  4. Practice assignment: Have each member attempt one “meta-conversation” with Claude or ChatGPT — describe a goal, ask what the AI needs to know to accomplish it well, then codify the AI’s recommendations into a system prompt
  5. For Lou: Document the “infinite prompting” workflow as a step-by-step guide that members can follow without needing to read the original Simmons article

Additional Resources

Ideas from Chat

  • Don Back: “Building our custom thinking machines for multiple uses rather than trying to build something one-off” — a strong one-liner framing the shift from task prompting to reusable AI systems
  • Don Back: “I couldn’t generate this level of specification for a contract writer to create an article. This is deep, deep, deep.” — member reaction highlighting that process prompts exceed what was previously possible even with human specialists