The bait, then the rug-pull.
Every AI session charges a hidden fee: re-explaining who you are, what you sell, how you write. Rick Mulready names it the friction tax and then spends fourteen minutes dismantling it with a single portable system you build once and never rebuild.
Where the time goes.
01 · The Hidden Friction Tax
Names the three types of AI friction: duct-taping tools together, re-explaining yourself every session, and subscription roulette. Establishes that better prompts and better tools do not fix the root problem.
02 · Meet the Information Hierarchy
Demystifies the word agent using the eight-word Anthropic definition (models using tools in a loop), then reframes the real differentiator: not the AI itself but what it can read. Introduces the Information Hierarchy as the fix.
03 · Your Business as an AI-Ready Operating Manual
Walks through the two-tier structure: Tier 1 business folder and Tier 2 project folders (Instructions, Voice, References, Examples, Notes). Explains the purpose and content of each file.
04 · Claude Auto-Builds and Organizes Your Hierarchy
Live demo: use Claude Cowork to create the entire folder structure from a single prompt, fill files via voice interview, then have Claude organize pre-existing files into the correct locations automatically.
05 · Same Files, Any AI: The Asset That Never Expires
Proves portability with two live demos: Claude Cowork reads the hierarchy and generates newsletter topics in the right voice; Gemini reads the same attached files and produces equivalent output. Closes on the compounding nature of the Notes folder.
Named ideas worth stealing.
The Information Hierarchy
- My Business folder (about me, about my business, about my voice, my offers)
- Project folders: Instructions, Voice, References, Examples, Notes
Two-tier portable context system: a universal business folder with four identity/voice files, plus per-project folders each with five subfolders. Build once, point any AI at it.
The Friction Tax (3 types)
- Duct-taping AI together
- Re-explaining yourself every session
- Subscription roulette
Framework for diagnosing why AI output feels inconsistent — names the three failure modes before offering the fix.
Lines you could clip.
"You don't build agents. You build the information that they read."
"The AI is the disposable part. The hierarchy is the part that lasts."
"The fix isn't a better prompt or a new tool. It's writing your business down once in a way any AI can read."
Things they pointed at.
Word for word.
One folder beats a thousand better prompts.
Persistent AI quality is not a prompting skill — it is a systems problem solved by portable, structured context you write once and own forever.
- The re-explanation loop is not unavoidable friction — it is a symptom of context living inside tools instead of in files you control.
- Your voice file is the highest-leverage document in any AI context system: specific preferred phrases and explicit banned words outperform vague style descriptions.
- AI learns from examples the same way humans learn from textbooks — organized structure with clear hierarchy beats a wall of continuous text every time.
- Tool-native AI memory creates lock-in: the moment you switch models, platforms, or apps, you lose everything. Files you own travel with you.
- The Notes subfolder turns a static context document into a learning system — logging what worked and what failed lets the AI improve on your actual results over time.
- The fastest SOP you will ever write is a Loom recording of yourself doing the task: let the auto-transcription do the work, then drop it into your References folder.
- Portability is the real test of any AI context system: if the same files produce equivalent output in two different AI tools with the same prompt, the system is working.
- Building the hierarchy does not require manual typing — one conversation with Claude can scaffold the entire structure from a single spoken description.










































































