Keep your voice.
Sharpen the copy.
A source-based editing workflow for clearer, more natural copy that you can honestly stand behind.
Make AI-Assisted Copy Sound Like You
The “six-person team” video ends by offering a guide to make AI copy sound human. That offer does not match the gstack demonstration in the clip, so this guide fulfills the actual closing promise. No prompt can make copy “100% human”; the reliable approach is to give the model your real facts and edit the result yourself.
One useful step at a time.
- 01Collect your voice
- 02Draft from facts
- 03Edit for cadence
- 04Read aloud
Create a voice sample
Choose two or three examples of writing you actually like and have permission to use. Note sentence length, vocabulary, humor, transitions and what you avoid. Avoid asking AI to impersonate another living writer.
Analyze these samples of my own writing. Describe recurring sentence length, tone, favorite concrete details and phrases I avoid. Quote short evidence from the samples. Draft a five-rule voice guide and ask me to correct it before using it.
Draft from real material
Give the assistant the audience, purpose, facts, examples and the action the reader should take. Ask it to flag gaps instead of filling them with generic claims.
Draft a [format] for [audience] using only the facts below. Follow my approved voice guide. Lead with the useful point, use concrete nouns and natural sentence variation, and mark any missing fact as [verify]. Facts: [paste].
If the draft sounds flat, add a specific scene, customer question or personal observation that is true. Do not solve blandness by adding fake stories.
Edit until it is yours
First remove filler and repeated points. Next replace abstract claims with a precise example. Then read the text aloud and change sentences you would never say. Keep a record of any source or quote the copy relies on.
Edit this draft for natural rhythm. Preserve every verified fact. Remove stock phrases and unsupported superlatives. Show the revised copy, then list the five most meaningful changes and any claim I need to verify.
Final checklist: Would you say this aloud? Can you prove the claims? Is the opening useful? Does each paragraph move the idea forward? Is the call to action clear?
The video’s gstack reference
The body of the video demonstrates gstack, Garry Tan’s open-source collection of Claude Code workflow skills. Its product-review and engineering-review commands are separate from the writing advice in the closing call to action. Follow the repository’s current README if you want to explore it; do not assume a slash command is built into Claude Code without installing the project.
Sources & notes
Product details were checked on 5 October 2026. Menus, plans, and integrations can change. The video is the source of the lead-magnet promise; the links below support the practical instructions and corrections in this guide.
- OpenAI prompt guidance: Clear context and iterative prompting.
- Claude personalization features: Account instructions, projects, and skills.
- gstack by Garry Tan: Actual open-source project shown in the clip.
This guide follows the video’s closing lead-magnet offer; the gstack topic mismatch is documented in Chapter 4.