PowerUp Coaching — Living Knowledge Base

PowerUp AI Mastermind Recap — 2026-03-19

13 min read

PowerUp AI Mastermind — March 19, 2026

Guest: Michael Simmons — Paradigm Collision, AI Curation, and the Future of Thought Leadership

“Ideas are the core currency of thought leadership, not content. Content is the catalyst that spreads an idea.” — Michael Simmons

This Week in 30 Seconds

  • Michael Simmons — 100M-view thought leader and writer shares his full journey from 60-hour articles to AI-first knowledge architecture
  • The blockbuster strategy — why spending dozens of hours on fewer articles outperforms volume, and how Michael discovered this from a Harvard professor’s media research
  • Paradigm collision — Michael’s core framework: 400+ categorized paradigms, parallel AI agents, and an elimination tournament to surface counterintuitive, actionable insights
  • The AI curator — an emerging knowledge entrepreneur archetype: someone who has high-quality conversations with AI and publishes what surfaces, transparently
  • Voice and authenticity — Michael’s surprising resolution to the voice problem: AI-first authorship, where editorial judgment replaces voice mimicry
  • Lou’s Hot Take vision — how the mastermind recaps might evolve to include an AI-generated perspective expansion that members didn’t think of themselves
  • Personalization prompt — Ri Ca shared Rich Schefren’s framing: “knowing what you know about me…” shifts AI from generic expert to context-aware advisor

Michael Simmons — The Journey to AI-First Writing

Michael Simmons has been thinking about knowledge and ideas his entire adult life. He co-founded a web development company in high school, wrote a book in college, toured 500 colleges with his wife in an RV, and stumbled into content creation when a friend invited him to write for Forbes. What happened next is worth telling in detail, because it’s the origin story for everything he shared today.

When he started on Forbes, Michael didn’t try to match the platform’s volume norms. He’d discovered a Harvard professor named Anita Elberse whose research showed that in every media category — books, music, films, TV — the winning strategy was quality concentration, not volume. Blockbuster releases beat the long tail. So Michael started spending dozens of hours per article instead of a few hours. His average Forbes article got 10x the platform average. On Medium, where he later moved, his average article reached over 200,000 readers.

Over time he developed a systematic approach to knowledge creation: 48 months of deep-diving one mental model per month (10,000-word manuals), studying the specific cognitive operations behind insight generation, and A/B testing titles before writing a single word. His click-through rates went from 1% to 10% as a minimum baseline. He built what amounts to a craft science for thought leadership.

When ChatGPT arrived, he assumed this system was the perfect input for AI. He created hundreds of prompts, hired Make.com developers, built automations. The frustrating reality: copy-pasting between prompts doesn’t scale, Make.com breaks and requires constant debugging, and the recursive nature of real learning (an interesting idea mid-process spawns a new direction) doesn’t fit linear pipelines.

Then he tried Claude Code. His assessment: it’s the thing he’d been looking for — a way to create systems for knowledge work, not just individual outputs.

“It has the word code in it. I’m not a programmer — it must not be for me.” — Michael Simmons, describing the barrier he had to overcome (and warning others about it)

His advice to anyone who installed Claude Code and stalled: find someone who can watch you use it, give you a few shortcuts, and help you get unstuck. Michael paid someone $150/week for 90-minute sessions just to get over the initial friction. The investment paid off immediately.

The Paradigm Collision Framework

The intellectual core of the session. Michael’s insight: most thinking happens inside a single paradigm, which means it inherits that paradigm’s blind spots. The way to generate non-obvious insight is to deliberately collide multiple paradigms — and let the collision surface what none of them could see alone.

His implementation: a CSV file of 400+ paradigms organized by discipline and tension (economics schools, behavioral perspectives, ecological economics, market critics, etc.). A Claude Code skill that deploys parallel agents, each looking at a subsection of the paradigm list, assessing relevance to the topic, and generating candidate insights from that perspective. The agents run for about an hour on a given topic.

He demonstrated this live with Jack Dorsey’s Block layoffs: Wall Street saw a leaner, more AI-forward company. The AI research community saw something else. Employees saw something else again. Each lens reveals something; each lens also hides something. The synthesis — the insight that no single perspective could produce — is what Michael is after.

The elimination tournament: the initial run generates 40+ candidate insights. They’re then run through structured rounds with explicit criteria (Is it counterintuitive? Does it change how you think? Is it actionable?) until 5-10 survive. Those are the ones worth developing.

Michael’s underlying principle comes from The Wisdom of Crowds by James Surowiecki: under the right conditions, diverse perspectives produce results that no individual perspective can match. The paradigm encyclopedia is designed to produce that “hybrid vigor” — where the perspectives actively cancel each other’s blind spots.

“Most of the time, these paradigms are invisible to us. We’re just in a paradigm, and we feel like that’s reality.” — Michael Simmons

Michael also mentioned that the paradigm taxonomy itself can be extended. He shared his current categories in chat: Paradigms, Mental Models, Councils, Other Types of Perspectives, Time, AQAL (Ken Wilber), Developmental Perspectives. The instruction: “Pick perspectives that optimize for a wisdom of crowds effect / hybrid vigor.”

Deep Dive: Insight - Paradigm Collision Is the Engine of Non-Obvious Insight — how to build an insight-generation machine that systematically outthinks single-perspective analysis.

Ideas as the Core Currency of Thought Leadership

Michael made a distinction that reframes how most coaches and consultants think about their content strategy. The people with durable thought leadership brands — Kevin Kelly, Cal Newport, Mel Robbins — each built on one or a small number of powerful ideas. The books and articles were distribution mechanisms. The idea was the asset.

Most content strategy advice optimizes for volume, distribution channels, and format. Michael argues this is optimization for the wrong variable. The bottleneck isn’t content production; it’s idea generation. If you have a genuinely new idea — one that changes how people think about something — the content almost writes itself and distributes because people want to share it.

AI makes this more true, not less. With AI, production is no longer the bottleneck. The remaining bottleneck is the quality and originality of the idea. Which means the investment should go into the thinking process, not the writing process.

His blockbuster strategy was always about this: spend more time on fewer ideas, develop them more deeply, and let the quality of the idea carry the content. AI now lets him accelerate the development phase without sacrificing the depth.

Deep Dive: Insight - Ideas Are the Currency of Thought Leadership, Content Is Just the Catalyst — why the investment in idea quality pays compounding returns that content investment doesn’t.

The AI Curator — A New Knowledge Entrepreneur Archetype

One of the most interesting conceptual contributions of the session was Michael’s search for a name. He described an emerging type of knowledge entrepreneur: someone who can have high-quality conversations with AI, surface the ideas that emerge from those conversations, and publish them with transparent editorial attribution.

It’s not ghostwriting (where AI mimics your voice). It’s not summarization (where you condense others’ work). It’s something between researcher and curator — the value isn’t in the writing, it’s in the quality of the conversation, the judgment about what’s worth publishing, and the editorial skill in presenting it.

Donald proposed “cairator” in chat. Lou riffed on “Curaitor.” The name doesn’t exist yet, which is itself a signal: the archetype is new.

Michael’s resolution to the voice problem connects here. He spent years trying to get AI to match his voice, finding it frustrating and fragile. His current experiment: stop trying. Instead, write articles that are transparently AI-generated, curated by him. He argues this is more honest than AI-mimicked voice, and the market may respond better to genuine AI-assisted synthesis than to AI that’s pretending to sound human.

“The irreplaceable human role in AI podcast creation isn’t creation — it’s curation. Editorial judgment and distribution instinct.” — Michael Simmons

Mazie captured it cleanly: “I like that you can create a multidimensional article with thoughts and concepts that go way beyond my own perspective.” Don: “The approach gives me a framework to start thinking about how the ideas arise and how to grow them.”

Lou’s Hot Take Vision

Lou connected Michael’s paradigm collision approach to an evolution he’s considering for the mastermind recaps. Rather than producing standard session summaries, he wants to experiment with running each session’s most interesting ideas through multi-perspective analysis — generating a “Hot Take” section that includes a perspective the participants didn’t think of themselves.

Michael confirmed the vision: AI-assisted idea curation as an editorial layer on top of the group conversation. The question Lou is working through: is there a reliable prompt architecture for genuine perspective expansion, or does it require the kind of curated paradigm encyclopedia Michael has spent years building?

Don observed something resonant: “Now I wonder how AI has changed my voice vs. the other way around.” The relationship between human and AI thinking may be genuinely bidirectional at this point.

The Personalization Prompt — Ri Ca’s Contribution

Ri Ca shared a prompt framing she learned from Rich Schefren that the group responded to strongly: “Knowing what you know about me…” — used as a prefix before any question to Claude. The effect: Claude shifts from operating as a generic expert to operating as a context-aware advisor who draws on everything it already knows about you.

This is a small syntactic shift with a surprisingly large output difference. Tried it during the session and confirmed it works.

Deep Dive: Insight - Knowing What You Know About Me - The Personalization Prompt — the framing shift that activates Claude’s contextual memory and produces more personally relevant answers.

Closing Threads — Consciousness, Curiosity, and What’s Next for Michael

The final stretch of the session turned philosophical. Don observed that “the future is not fixed and predetermined” and that “the most critical action is to initiate — nothing develops at rest.” Michael’s work on paradigm collision is itself an enactment of this: there’s no fixed endpoint to the insight generation process, just better and better conditions for emergence.

Michael mentioned he’s a new father (4-month-old congratulated by the group) and that his next development phase involves building the RAG database behind his paradigm encyclopedia — he flagged Lou as the person he wants to work with on that architecture.

Donald made the observation that “everyone can now have their own PhD research assistant” — Michael’s response: the irreplaceable thing is still the human, the intellectual curiosity, the editorial judgment. You can’t drink beer with the assistant, Don pointed out. Not yet, Donald replied.

Community Corner

Michael Simmons was, as promised, exactly the right guest for this moment. Lou’s framing — “a guy who journeyed from skeptic to advanced practitioner” — turned out to be underselling it. Michael has been thinking systematically about knowledge creation longer than most of this group has been thinking about AI. The result is a methodology that the group can borrow from immediately.

Mazie’s comment near the end of the session — “I’m finding that people crave deep conversation” — felt like a summary of the whole session. Two hours of depth on idea generation, paradigm diversity, and what it means to be a knowledge entrepreneur in an AI world. Nobody wanted it to end.

Books mentioned:

  • The Wisdom of Crowds by James Surowiecki — see above
  • Why Greatness Cannot Be Planned — mentioned by Michael Simmons; relates to non-linear paths to breakthrough ideas and the limits of goal-setting as an innovation strategy
  • Colin & Samir Podcast — recommended by Michael for creators thinking about the creator economy

Try This Before Next Session

Run a 5-lens perspective expansion on one idea or topic you’re currently working on.

  1. Choose a topic you’ve been writing about, teaching, or thinking through
  2. Open Claude and say: “I’m going to give you a topic. I want you to analyze it from five different paradigms or stakeholder perspectives. For each one, tell me: (a) what this perspective sees that others miss, and (b) what this perspective hides or gets wrong. Then synthesize: what insight emerges from the collision of these five perspectives that none of them could produce alone? Topic: [your topic].”
  3. Read the synthesis. Is there something there that surprises you? That challenges your current framing?
  4. If yes: develop that. If no: ask Claude to try with five more perspectives from different disciplines

The goal isn’t a finished piece of content. It’s to experience the difference between single-perspective and multi-perspective thinking — and to find the idea worth pursuing.

Open Threads

  • What’s the right name for the “AI curator” role — someone whose skill is having high-quality AI conversations and surfacing the ideas that emerge?
  • How does the paradigm collision approach adapt for coaches who work in one specific domain — does the paradigm encyclopedia need to be domain-specific to be useful?
  • Lou’s Hot Take vision — how do you consistently find the insight that a group didn’t think of, rather than just synthesizing what they did? Is there a reliable prompt architecture?
  • If “ideas are the currency,” what’s the equivalent of a diversified portfolio — how many signature ideas should a coach be building around simultaneously?
  • Michael and Lou: connect on knowledge graph / RAG architecture for Michael’s next development phase

Next session: April 2, 2026 (Lou returns from Mexico)

← Previous · Next →

Derived Artifacts

Graph View

Backlinks