2025-12-12_Mastermind

2025-12-12 AI Mastermind

Table of Contents

• Insight - The Psychographic FAQ as Authority Infrastructure

Session Overview

The December 12 session was a hands-on workshop centered on Lou’s GEO FAQ application — the tool he had previewed in the November 27 session and has now completed to the point of live group testing. Lou joined from Thailand where he’d been enjoying warm weather, hair still damp — and despite bandwidth challenges with Zoom, the session delivered substantial practical value.

Lou walked the group step by step through the full GEO workflow: generating an Ideal Client Handbook (ICH) using his prompt in a leading LLM (with web search enabled for verbatim customer quote extraction), logging into the GEO app, populating the canonical entity profile (social profiles, biographical information), pasting the ICH output, and running the generation pipeline to produce a structured JSON-LD file ready to embed in an HTML FAQ page. The group worked along in real time, with several members hitting familiar friction points — the intake form needing to be pre-populated, ChatGPT 5.2’s behavior of pulling from memory rather than asking intake questions, the importance of using plain text editors rather than Word to avoid special character contamination.

Kasimir’s technique for the intake form — running it through his virtual advisory board to pre-populate the answers from data the board already had — generated a useful discussion about using AI memory and context as productive shortcuts in workflow design. The session also surfaced the important distinction between Lou’s approach (using live web search to pull verbatim customer quotes into the ICH) versus the synthetic voice-of-customer generation in other ICH frameworks — a meaningful quality upgrade that produces semantically richer FAQ questions.

The session concluded with a forward-looking discussion about connecting the GEO FAQ infrastructure to a broader knowledge graph (Qdrant, with permanent memory and document ingestion), setting up the December 19 session’s deeper exploration of the LinkedIn content layer.

High-Signal Moments

• Live GEO app walkthrough: the full pipeline from ICH input to JSON-LD schema output — the most concrete demonstration of GEO infrastructure the group has seen
• The verbatim quote distinction: Lou’s ICH prompt uses live web search to pull real customer language from Reddit, LinkedIn, and Twitter — producing semantically dense FAQ questions vs. synthetically generated ones
• Kasimir’s advisory board intake trick: using the virtual board to pre-populate the intake form from memory — a model for how to use configured AI contexts as workflow accelerators
• “You should know this much about your market, for crying out loud” — Lou on the ICH intake form, a coaching anchor point about client avatar clarity as foundational, not optional
• The special character warning: always use plain text editors (TextEdit, Notepad) not Word when working with LLM-generated markdown; invisible characters break schema validation
• LLM context mode discussion: ChatGPT 5.2’s tendency to pull from memory rather than ask intake questions — the importance of understanding how memory mode affects prompt execution
• The Qdrant knowledge graph vision: Lou previewing the next evolution — permanent memory that updates as new documents are added, creating a continuously enriching sovereign knowledge base
• FAQ page deployment variations: WordPress, GoHighLevel, VPS — and the recommendation to use Fiverr to implement if you lack the technical background

Open Questions

1. How do we measure whether the GEO FAQ schema is actually being picked up by AI engines — are there reliable citation monitoring tools for LLM mentions?
2. What are the best public community sources (beyond Reddit and LinkedIn) for specific coaching niches — how do we customize the ICH web search to find the highest-signal verbatim quotes?
3. How long does it typically take from FAQ page deployment + indexing submission to first measurable AI engine citation?
4. Can the Qdrant knowledge graph be queried across multiple practitioners in the mastermind — creating a shared authority network?
5. What’s the minimal viable deployment for someone without technical resources — the simplest path from JSON-LD output to a live indexed page?

Suggested Follow-Through

1. All mastermind members: Complete the ICH prompt in your preferred LLM (with web search enabled), populate the GEO app, and publish at least one FAQ page by January. Share results in Telegram.
2. Lou: Integrate the ICH generation directly into the GEO app so it becomes a single-entry-point pipeline; document WordPress/GoHighLevel/VPS deployment paths.
3. Lou: Build out the Qdrant knowledge graph prototype and share a demo — the “sovereign mind” permanent memory system with document ingestion.
4. All: Begin monitoring AI engine responses to queries in your domain — note when your name or content appears; track changes after FAQ page deployment.
5. Kasimir: Share his virtual advisory board intake pre-population technique with the group — document as a reusable prompt pattern.

Additional Resources

Links & Tools Shared in Chat

• PsyGen App — shared by Lou; the GEO FAQ generation application being demonstrated in this session

Books & Articles Mentioned

• None.

Ideas from Chat

• Noota.io as AI meeting note-taker: Kasimir’s noota.io assistant joined the call — its auto-introduction identified it as a GDPR-compliant, automated note-taking bot. Noota.io is a European AI meeting assistant that records, transcribes, and summarizes meetings. A tool worth knowing for mastermind members wanting automated session notes without manual effort.
• ICH + GEO app workflow traction: Several chat messages indicated members were successfully running the workflow in real time — Elizabeth Stief noted Claude and Perplexity both pre-populated her intake from memory; Donald Kihenja had his data picked up from AI memory. Confirms AI memory as a meaningful workflow accelerator for returning users.
• Plain text editors for LLM output: Chat reinforced the session’s core technical tip — Ri Ca was still working on replacing square-bracket text in Microsoft Word, and the session note about Word’s invisible characters breaking schema validation was directly relevant to her problem.