Before/after precision diagnosis on three real sentences submitted by a SaaS founder.
This is a trimmed sample. The full skill runs six Meta-Model precision drills and gates every rewrite through a Voice Profile lock before delivering the final copy.
Five moves. Four gates. Diagnostic before transformation.
The most common failure of AI rewriting is changing the line before diagnosing what's wrong with it. The skill diagnoses first, then transforms.
Capture the precision task.
What this line has to do, where it appears, who reads it. Locks the criteria for "landed."
Lock the voice.
Builds or loads the brand's voice fingerprint. Sharpened lines must still sound like the brand, not like generic NLP.
Name what's wrong. Then rewrite.
Runs the diagnostic drills first (what's vague, abstract, soft, hedged). Then runs transformation drills to fix what was diagnosed.
Reader test the sharpened line.
Pressure-tests the new version against the original. Confirms the line is sharper without losing meaning.
Final precision with rewrite log.
The sharpened line plus a log of what changed and why. So you can learn the pattern, not just copy the output.
The sentence is the unit. Each drill is a specific surgery.
Six drill drawers. Diagnose drills run first. Transformation drills run after. The skill picks the right combination for the line on the table.
Diagnose.
Drills for naming what's wrong with the existing line. Vague nouns, abstract verbs, hedged claims, missing specificity, unowned voice.
When the line "feels off" but you can't say why.
Audit.
Drills for running an existing page or piece of copy through the precision filter. Identifies every soft sentence on the page.
When the whole landing page needs sharpening, not just the hero.
Recover.
Drills for recovering the lost specificity. Replaces nominalisations with verbs. Replaces abstract adjectives with concrete images.
When the line has been written so many times it's lost its bite.
Sensory.
Drills for adding sensory specificity. Sight, sound, texture, taste. The kind of detail the reader can picture without effort.
When the line is conceptually clear but emotionally flat.
Persuade.
Drills for the persuasive shape. The Pattern Interrupt. The Specific Image. The Named Problem. The Implied Authority. Inserted with surgical care, not slathered on.
When the line is honest but doesn't move the reader.
Transform.
Drills for the rewrite itself. Five sharpened variants. Reasoning behind each. The reader test that confirms the new line lands harder than the old one.
When you're ready to ship and need three or four options to A/B.
Four sequential artefacts. Each one a gate.
Brief. Voice. Rough work with diagnostic and transformation. Final precision with rewrite log.
The Precision Brief
The real task, named.
What the line has to do, where it appears, who reads it, the criteria for success. Halts the work.
The Voice Profile
Locked before rewriting.
Fingerprint of the brand voice. Sharpened lines must still sound like the brand. Non-negotiable.
The Rough Work
Diagnostic plus transformation.
What's wrong with the original. Three to five sharpened variants. Reasoning per change. Halts to redirect.
The Final Precision
Ready to ship.
The final sharpened line plus a rewrite log so you can learn the pattern. Plus the reader-test result.
Diagnostic first. Transformation second.
Built for the people doing the work, not watching it.
- Founders editing their own site hero and tired of "it sounds fine but doesn't move me."
- Copywriters running a precision pass before final delivery.
- Marketing leads briefing agencies who want sharper lines back.
- Product teams writing UI strings, error messages, empty states.
- Anyone with one critical line that has to land and refuses to settle for "good enough."
- ×People who need body copy written from scratch (use the Copy Engine).
- ×People who need openers generated from nothing (use the Hook Engine).
- ×Anyone hoping for one magic sentence with no inputs.
- ×People who want to skip the voice profile.
- ×Brands that don't have a voice yet.