Turn Any Book Into an AI Skill You’ll Actually Use

Main ideas / lessons

• Reading ≠ learning. Even if you remember what’s in a book, real learning only happens when you change behavior and then evaluate results.

• Problem with typical “book → AI” workflows:

  • Uploading a book or using AI summaries often creates an over-abstraction chain: original experience → book abstraction → summary abstraction → “summary turned into a skill” abstraction

  • Each compression removes nuance and examples, producing skills that are vague and not reliably actionable.

  • What a useful AI “skill” must do (3 core requirements): 1. Know when to use the method: identify the specific behavior/action to change and the exact situation where it applies. 2. Know how to make the change: turn the method into concrete steps tied to behavior change. 3. Record and review outcomes: measure what happened after applying the method and determine whether results improved, worsened, or changed direction.
  • Feedback loops are the engine of learning, and AI can help build and maintain them by tracking, remembering, and surfacing evidence over time.
  • Skill mastery framework: Shu-Ha-Ri (Japanese martial arts / tea ceremony)
  • Shu (obey / follow): replicate the teacher/expert’s method exactly at first.
  • Ha (break / adapt): once you understand, explore edges and customize.
  • Ri (move away / innovate): after sufficient experience and evidence, develop your own framework/style.
  • Don’t customize too early. A key failure mode is trying to implement all stages (follow, adapt, innovate) simultaneously—leading to confusion and no measurable progress.
  • Use AI for “emotionless support.” Learning is emotional and uncertain; AI can guide the process without emotional friction, helping you stay on track through steps and analysis.

---

Method: how to turn a book into an AI skill you can actually use (detailed steps)

A) Build a behavior-change skill from a book/media source

• Choose a source (book/podcast/YouTube/video) that contains a method you want to apply.
• Identify the exact behavior you want to improve/change (what gap you’re trying to close in real life).

B) Avoid the “abstraction of abstraction” trap

• Don’t rely only on summaries or “make a skill from the condensed ideas.”
• Instead, give the AI enough context so it can produce an actionable system that preserves important nuance.

C) Design the AI skill around the 3 required functions

• 1) When to use the method
  • Define the situation/trigger.
  • Define what action you are trying to change (behavior-level intent, not “read this for fun”).
• 2) How to apply it
  • Translate the method into decision points and checks tied to the behavior.
• 3) Record & review results
  • After use, capture evidence of:
    • whether you followed the method,
    • what results changed,
    • and what improved/worsened/didn’t change.
  • Use the evidence to decide whether to pivot, stay, or refine.

D) Create an evidence-driven feedback loop (continuous improvement)

• Track outcomes over time so learning becomes:
  • “What worked?”
  • “What didn’t?”
  • “Should I pivot or improve the strategy?”
• Use AI to maintain the loop (AI can track patterns humans may miss and do it faster/consistently).

E) Apply Shu-Ha-Ri properly

• Shu: do the method exactly as specified first (build base competence).
• Ha: then experiment with boundary-pushing and limited customization.
• Ri: only later, after evidence accumulates, develop your own approach.

F) How AI is used during the workflow

• Articulate the situation where you will apply the method.
• Take action (e.g., run the marketing launch, update pages, execute the process).
• Let the AI agent record results and compare work against the method.
• Review output and update the next iteration based on diagnostics.

---

Practical example shown (One-Page Marketing Plan → marketing skill)

Setup goal

• Use the book “The One-Page Marketing Plan” (Alan Dib) as a “skill” to evaluate marketing assets (e.g., a sales page) against his structure.
• The skill is intended to:
  • ensure correct targeting and messaging (“nine grid”),
  • check if the creator followed the method,
  • and measure outcomes.

How the agent’s skill is structured (as described)

• The agent creates phases/decision points corresponding to the book’s “nine grid” journey.
• It also creates a set of specific checks (minimum evidence) for each quadrant, including things like:
  • core message
  • offer mechanism
  • proof
  • objections
  • CTA (call to action)
  • claims/boundaries
  • and other evidence checks aligned to the grid

Demonstration/testing

• A sales page URL is provided to the agent without extra explanation.
• The agent:
  • analyzes it as a test,
  • follows the book’s method,
  • reports which parts align and which are broken/unsupported,
  • flags missing or unverified items (e.g., UTM/source attribution, evidence, lead nurture paths).
• The presenter then plans updates based on the agent’s diagnostics (e.g., add missing tracking parameters, create an “interested non-buyers” path).

Outcome tracking / iteration

• After a course launch, the presenter plans to return to the agent with what happened so the feedback loop can record:
  • what was fixed,
  • and whether results improved next time.

---

Note-taking enhancement (Obsidian / Zettelkasten context)

Optional improvement: add personalized context via notes

• If the user has notes (e.g., in Obsidian under Zettelkasten), the agent can:
  • retrieve relevant notes related to the book,
  • identify knowledge gaps (based on patterns like repeated highlights/themes),
  • tailor the skill to the user’s specific weaknesses.

What the skill becomes better at doing

• Instead of only applying the generic “nine grid,” it can:
  • identify the largest overall gap (including the bridge from learning → application),
  • propose which grid cell or component is the bottleneck,
  • prioritize diagnostics (e.g., find the “4% causes 64% impact” type constraint),
  • turn evaluation into a focused, step-by-step diagnostic plan (current constraints, customer transition, audience/offer/channel/CTA, and what to not solve yet).

---

Speakers / sources featured (identified in the subtitles)

People (speakers mentioned)

• The video narrator / presenter (unnamed in subtitles; provides instruction and examples)
• Christo (mentioned as an “awesome YouTuber”)
• Alan Dib (author of The One-Page Marketing Plan)
• Vicki (mentioned as someone who recommended the books)

Works / sources

• The One-Page Marketing Plan — Alan Dib
• Obsidian (note-taking system mentioned)
• Zettelkasten / Zettocasten (note system/folder mentioned)
• Japanese martial arts / tea ceremony mastery concept: Shu-Ha-Ri (concept referenced)
• Mentioned AI tools/platforms:
  • Codeex
  • Claude (and “claude code/claude work” variants mentioned)
  • ChatGPT (and “chat version” vs agents)
  • Agents / “AI agents” (general concept)