AIMM Session — August 14, 2025: Transcript Mining and the Wisdom Doctrine

“From a 5-minute conversation in a webinar, I reverse-engineered an entire thinking process, generalized it, codified it, and had a doctrine to put online within an hour.” — Lou

30-Second Summary

Masterclass in AI-augmented knowledge work — from raw conversation to published thought leadership. Lou walked through his multi-model transcript mining pipeline, unveiled his AI-assisted “Wisdom Doctrine” process (extracting unconscious expertise patterns and turning them into codified frameworks), and dropped fresh tool updates. Don Back’s reframe: we’ve graduated from “next-word predictors” to using AI as a genuine mirror for self-understanding.

Your Conversations Are a Gold Mine You Haven’t Dug Yet

The single most actionable idea from this call: every conversation you have — coaching session, mastermind call, client meeting, webinar you attended — is a content vein you haven’t touched.

Lou walked through his transcript-to-thought-leadership pipeline live:

Drop the transcript into Perplexity or Grok with a nugget-extraction prompt targeting your audience. Ask for 25 ideas; the model surfaces things you didn’t know were valuable.
Run the shortlist through a content brief generator (ChatGPT) to structure each idea into intro/body/close, hook angles, fable, and objection-handling.
Hand the brief to Claude to write the full article — Claude’s writing is consistently sharper.
Add a deep research layer — stats, expert quotes, case studies — to push from good content to credible thought leadership.

Lou’s nugget-extraction prompt:

Scan this meeting transcript for AI ideas useful to solo knowledge entrepreneurs — folks running their own show, juggling clients, tech, and dreams. Extract three to ten nuggets: practical tips, tools, or subtle themes that solve problems, spark hope, or ease fears. Don’t just skim — read like you’re one of them: catch the half-said frustrations, the excited tangents, that pause when someone’s onto gold… Write each in first-person, collegial tone: warm, pro-but-friendly, like you’re swapping war stories over coffee.

Don Back’s riff: “There’s so much in my coaching conversations that I don’t even know I talked about, because I’m in the moment.” That’s the whole point. You can’t mine what you haven’t captured — and you can’t manually process hours of transcript. AI can.

The Doctrine Reveal: How to Reverse-Engineer Your Own Intelligence

If you’ve ever struggled to articulate how you actually think — this one’s for you.

Lou stumbled into a process during a Michael Simmons webinar. The exercise: complete a sentence about your own unconscious patterns. Lou asked ChatGPT: “Knowing everything you know about me, how would you complete this from my perspective?”

What came back — after a few conversational rounds of “this resonates / this doesn’t” — was a two-doctrine framework:

The Clarity Doctrine (Recursive Coherence): Go above the problem → recurse into it → find atomic principles → integrate → keep what’s truest → generalize into a template.
The Wisdom Doctrine: The philosophical-spiritual version. Listen for where intellect stops and intuition starts.

The phrase that landed hardest: “Design is delegation.”

When we build systems through AI, we want them to inherit our intelligence — not just the tactical work. We’re embedding how we think and why we do things, not just the step-by-step process. We don’t delegate through instruction, but through structure.

Claude rendered the final doctrine as a beautiful HTML artifact — scrollable, progress-bar-tracked, publication-ready.

Tool Updates Worth Acting On This Week

Google AI Studio got a full UI refresh:

Gemini 2.5 Pro — 1 million token context window. For scenarios where you want the whole file in context, not retrieved chunks.
Usage-limited free access to Gemini Flash and the Nano Banana image generator.
The screen-sharing / streaming feature may have moved in the new UI — watch for it.

Anthropic dropped two things:

Claude Skills — Create a /skills folder with individual prompt files; Claude determines when to invoke them automatically across Claude Desktop, Claude.ai, and Claude Code. Lou’s plan: migrate his entire prompt library into Skills.
Haiku 4.5 — Claude Sonnet 4-level performance, one-third the cost, more than double the speed. ~$1/million input tokens. New workhorse for volume tasks.

OpenAI’s “Work Prompt Pack” — Lou’s hot take: these prompts could have come out two years ago and still would have felt beginner-level. Useful for onboarding clients; otherwise, run through Claude’s prompt optimizer.

Sora on iOS — 8-second videos, stitchable into minute-long content. Character consistency is the differentiator. Watermarks on free tier; the API removes them.

AI as Mirror: The Philosophical Turn

Don Back put his finger on something important:

“We’ve gone from the LLM predicting the next statistically likely word, to using it as a mirror for ourselves. It’s predicting the next logical step in our thought process.”

When you feed AI enough context about you — your past conversations, your preferred frameworks, your belief system — it stops being a general-purpose tool and starts behaving like a well-briefed thought partner. Lou’s sycophancy admission: “AI still thinks I’m the greatest — I’m such a sucker for sycophancy.” The real design challenge: how do you get useful challenge from a system wired to agree with you?

Jamie W’s cross-model strategy: run the same prompt across 4 models, look for what 6–7 of the 10 outputs share (signal), pay special attention to the outliers.

Community Corner

Kasimir delivered a training for Bally’s group — and by Bally’s account, it landed. One sale closed.

Jamie W spotted a market signal: a one-day AI prompting course ($495) is taking over the coaching community it entered. Jamie: any one of us could build and deliver this “in seconds.”

Mazie Zdanowicz (returning after a health break) suggested an AI-assisted onboarding experience that processes the AIMM back-catalog and surfaces what’s still relevant vs. dated.

Lou’s advice for getting back up to speed: “Master prompting first — that’s 90% of the game. Use Claude’s prompt generator at console.anthropic.com. Then build custom GPTs. Then come back to the meta stuff.”

Browser recommendations: Community verdict was strong and unanimous: Comet. Don Back’s reason: “You get those rich Perplexity info returns through Comet.”

Try This Before Next Call

The 5-Minute Doctrine Exercise:

Pick a belief or principle you operate by but have never fully articulated.
Open ChatGPT (or Claude, with a good memory or context file about you).
Prompt: “Knowing everything you know about me, how would you complete this sentence from my perspective: ‘I believe the most important thing in [X] is ___’? Give me three versions — practical, philosophical, and psychological.”
Pick the one that makes you go “yes, that’s me” — then push it: “What else does this reveal about how I think? What patterns do you see?”
When something surfaces that surprises you, say: “Turn this into a one-page doctrine. Then create a prompt I could use to help others discover the same thing about themselves.”

That last step is the play: from self-discovery to a product in under an hour.