Editing and Review Prompts¶

Writing prompts create content. The 14 remaining prompts make that content publishable. They're organized in three layers -- editing (refine the prose), review (gate the quality), and fix (target specific problems). Each layer catches things the previous one misses.

Three-Layer Quality System¶

Layer	Prompts	What It Does
Editing (5)	Voice consistency, audience calibration, de-AI patterns, add balance, simplify	Refines existing content into the target voice
Review (4)	Chapter complete, authenticity, audience fit, final quality	Quality gates that produce reports, not rewrites
Fix (5)	Generic language, add examples, add action items, two-audience, tighten prose	Targeted improvements for specific failure modes

The distinction matters. Editing prompts change the text. Review prompts evaluate the text and tell you what to change. Fix prompts handle narrow problems you've identified.

The 5 Editing Prompts¶

1. Voice Consistency¶

Input: Draft section + voice reference files. Output: Revised content with change notes explaining what shifted.

This prompt aligns content with the author's authenticity markers -- signature phrases, sentence rhythm, balanced skepticism, practitioner perspective. It checks whether the draft sounds like the author or like a capable but generic AI.

The prompt references the full phrase tables from the master system prompt: opening gambits, transitions, balance markers, action orientation. It's not just looking for bad patterns -- it's actively inserting good ones where the prose feels flat.

2. Audience Calibration¶

Input: Draft section + audience profiles (Sarah and Marcus). Output: Revised content with added audience callouts where advice diverges.

This is the prompt that prevents the book from accidentally becoming an enterprise-only or startup-only guide. It reads through the draft and flags anywhere the advice would land differently for a skeptical COO with 2,000 employees versus a startup founder with 18 months of runway.

Where the difference matters, it adds callout boxes:

For Established Organizations: Your existing workflows are constraints, not obstacles. Design for integration.

For Startups: You have fewer constraints -- use that. But don't mistake "fewer constraints" for "no constraints."

3. De-AI Patterns (The Most Critical Editing Prompt)¶

Input: Draft content, especially anything initially generated by AI. Output: Cleaned content that passes four authenticity tests -- personality, specificity, controversial opinion, and surprise.

This prompt earned its place more than any other in the system. It detects and removes 9 categories of AI tells:

Filler phrases -- "It's important to note that," "It goes without saying"
Corporate buzzwords -- delve, leverage, utilize, harness, synergize
Generic openers -- "In today's rapidly evolving landscape"
Weak hedging -- "It could be argued that," "One might consider"
Over-structured lists -- "Firstly, Secondly, Thirdly" with suspicious parallelism
Artificial enthusiasm -- "Exciting possibilities," "Tremendous potential"
False balance signals -- "While some may argue," "It's a double-edged sword"
Conclusion tells -- "In conclusion," "To summarize," "As we have seen"
Verbosity patterns -- same point three ways, decorative adjectives, throat-clearing

The output includes a severity assessment (critical AI tells vs. moderate corporate speak vs. minor style improvements), a change log with original text and replacement, and the full revised content.

The before/after examples tell the story:

Before: "In today's rapidly evolving AI landscape, it's important to note that organizations must leverage emerging technologies to harness their full potential."

After: "AI is changing how companies compete. Not theoretically -- practically. The businesses figuring this out now will have advantages that compound over years."

Same meaning. Half the words. Zero AI smell.

4. Add Balance¶

Input: Draft content. Output: Revised content with risks added to opportunities and opportunities added to risks.

Balanced skepticism is a voice non-negotiable. Every exciting claim needs its caveat. Every warning needs its upside. This prompt scans for one-sided takes and inserts the missing perspective. "This approach works well, but here's what can go wrong..." is the signature pattern.

5. Simplify¶

Input: Draft content with complex passages. Output: Revised content with improved clarity without losing depth.

This prompt targets unnecessarily complex sentences, jargon that isn't earning its place, and abstractions that could be concrete. It's the clarity pass -- reducing reading friction without dumbing down the content.

The 4 Review Prompts¶

Review prompts produce graded reports with specific locations and recommendations. They don't rewrite -- they evaluate.

1. Chapter Complete¶

The most comprehensive review. It checks structural elements (opening hook, body flow, closing synthesis), content elements (case studies, frameworks, action items, audience callouts), voice elements (signature style, anti-AI patterns, density, complexity acknowledgment), and technical elements (frontmatter, links, formatting).

Output: A-D letter grade with quick stats (word count, section count, case study count, framework count), critical issues that must be fixed, improvement opportunities, and minor polish items. Each issue includes a specific location and actionable recommendation.

2. Authenticity¶

Voice fingerprint verification. This prompt scores the draft against the author's authenticity markers -- not just avoiding bad patterns, but actively exhibiting good ones. Does it use numbered frameworks? Is there balanced skepticism? A pop culture bridge? Pattern interrupts? Direct "you" address?

The output is a scorecard. If the chapter hits all markers, it passes. If it's missing practitioner perspective or honest complexity, the report flags exactly where.

3. Audience Fit¶

Are both audiences served? This review checks whether Sarah (enterprise) and Marcus (startup) would both finish the chapter, both find an "I never thought of it that way" moment, and both know what to do next. It flags sections that accidentally exclude one audience -- enterprise transformation timelines that make Marcus check out, or startup-only examples that make Sarah dismiss the advice.

4. Final Quality¶

The pre-publish gate. Go or no-go with a list of remaining issues. This runs last, after all editing and fixes are complete. It's the final checkpoint before a chapter moves from "editing" to "done."

The output is binary at the top level -- publish or don't -- with a prioritized list of anything still wrong. At this point, issues should be minor. If they're not, the chapter goes back through editing.

The 5 Fix Prompts¶

Fix prompts target specific problems identified during review. They're surgical -- one problem, one prompt.

1. Fix Generic Language¶

The transformation engine. "The initiative faced resistance" becomes "Store managers were quietly turning it off and reverting to spreadsheets." This prompt carries a reference table mapping consultant-speak to visceral language:

Generic	Specific
"Various stakeholders"	"Your CEO, your customers, and your engineers -- who all want different things"
"Organizational readiness"	"Can your people actually use this thing?"
"Change management"	"Getting humans to actually change their behavior"

2. Fix Add Examples¶

Theory-heavy sections get company narratives and concrete scenarios injected. The prompt identifies passages where the writing explains without showing and inserts case studies or real-world examples that teach the point through demonstration.

3. Fix Add Action Items¶

Sections that tell the reader what to think but not what to do get "What This Means For You" callouts. The prompt scans for insight-heavy passages with no practical application and adds specific next steps.

4. Fix Two Audience¶

When a section speaks exclusively to one audience, this prompt adds guidance for the other. It inserts the callout box format for enterprise and startup where the advice differs meaningfully.

5. Fix Tighten Prose¶

The 20% rule. This prompt takes completed content and cuts until every word earns its place. It removes redundancy, decorative adjectives, throat-clearing, stacked examples proving the same point, and paragraphs that restate the previous paragraph. The target: same meaning, fewer words, more punch.

This prompt runs last in almost every workflow. First drafts are reliably 20-30% too long. The tighten pass is what gets them to publication density.

When to Use What¶

Problem You're Seeing	Prompt to Use
Sounds like AI wrote it	De-AI Patterns
Doesn't sound like the author	Voice Consistency
Only works for one audience	Two Audience + Audience Calibration
Theory-heavy, no examples	Add Examples
No actionable takeaways	Add Action Items
Too long, too padded	Tighten Prose
Vague, corporate-sounding	Fix Generic Language
All upside, no caveats	Add Balance
Ready for quality check	Chapter Complete
Ready to publish	Final Quality