# Repurposing is where voice goes to die
One good idea, five platforms, and by the third rewrite it sounds like the AI wrote it. Not you.
Here's the thing
You start with a solid idea. Genuinely good, the kind you'd say out loud to a friend without flinching. Then you run it through a rewrite tool for LinkedIn. Fine. Then a newsletter version. A bit stiffer. Then a thread. Then an email. By the time you get to the fifth format, the personality has been sanded off entirely. Everyone reading your content across platforms is, technically, reading five different people. None of them you.
That's the actual cost of repurposing. Not the time. The voice.
The fix
One measured voice profile. It sits underneath every format you produce and constrains the rewrite, so the shape changes but the voice doesn't.
ScriptGrain works from 45 measured reference points pulled from your own writing: rhythm, sentence length, how you open a paragraph, how you argue a point, the words you reach for and the ones you don't. That profile gets applied every time, whatever the format.
Sixteen formats, one voice
Articles, newsletters, LinkedIn posts, threads, emails, social captions, press releases, ad copy, product descriptions, landing copy and more: sixteen content types in all. Each format has its own structure spec, because a thread isn't built like an email and shouldn't be. And all sixteen are held to the same 45 reference points, so the newsletter and the thread sound like siblings, not strangers.
Structure changes. Voice doesn't. That's the whole idea.
You get a score, not a hope
Every draft comes back with a voice-match score, measured against your profile. You're not guessing whether the rewrite still sounds like you; you're checking.
To be fair, most tools ask you to trust the output. This one shows its working. If a draft drifts, you'll see it drift, and the polish loop can revise it back toward your voice before it goes out under your name.
It works from whatever you've already got
Notes scribbled before a call. A bullet list. A transcript from a podcast you recorded eight months ago and never touched again. A finished article you want reshaped into a script.
ScriptGrain doesn't need a polished brief. It needs your voice profile and whatever raw material you already have lying around. That's it.
Three steps
- Measure your voice once. Feed it writing you already have; get a profile back in about 45 seconds.
- Brief each format from your source material. Notes, transcript, finished piece, doesn't matter.
- Check the score on every draft. See that the voice survived the format change instead of hoping it did.
Measure once. Reuse everywhere. That's the whole system.
Why the old way fails
Generic AI rewriting tools optimise for "sounding fine" in isolation. Fine for a LinkedIn post. Fine for an email. But run the same idea through five separate "fine" rewrites and you get five separate voices, and none of them is yours; they're all trending toward the model's defaults.
That's not a one-off glitch. That's what happens every single time you rewrite without a constraint holding the line. The model doesn't know your voice. It knows patterns. Without a measured profile forcing it back to your shape, it drifts toward its own.
What this actually protects
Your readers notice when your newsletter sounds nothing like your LinkedIn posts. They notice when the "you" in an email reads like a helpful stranger. Consistency isn't a nice-to-have here; it's the reason people trust a byline enough to keep opening your stuff.
See your own profile, free
Before you commit to anything, see what a measured breakdown of your own writing looks like. The free tier includes one voice profile and one full analysis, no card required, and generating across the sixteen formats starts at £12/month.
Repurposing shouldn't cost you your personality. One idea, sixteen formats, one voice held constant throughout, with a score on every draft so you're not just hoping it worked.
Questions
Can I turn a transcript into a post in my voice?
Yes. Paste the transcript (or notes, or bullet points, or a finished piece) as your source material, pick the format, and the draft generates under your measured profile's constraints. The voice-match score on the result shows how closely it landed.
Which formats are supported?
Sixteen content types: long-form articles, newsletters, blog intros, short stories, LinkedIn posts, social captions, threads, Instagram captions, emails, press releases, product descriptions, ad copy, and the website formats (landing, homepage, product and about pages). Each has its own structure spec, all share your voice profile.
How do I know the voice survived the format change?
Every draft returns a voice-match score measured against your profile, with the per-feature deltas behind it. If a draft drifts, the polish loop revises it toward the profile and rescores, so you ship on evidence rather than hope.
How is this different from asking an AI to rewrite for LinkedIn?
An unconstrained rewrite optimises for sounding fine, and every pass drifts a little further toward the model's own average. A measured profile holds all sixteen formats to the same 45 reference points, and the score tells you it held, which is the part no generic rewrite can offer.