ScriptGrain vs My Writing Twin

ScriptGrain measures your style and then writes with it, checking every draft against the profile. My Writing Twin, the fairest comparison here, measures too, but gives you a profile and a connector rather than a tool that drafts and scores.

The actual difference

Both measure style rather than describe it, which makes this the fairest comparison here. The difference is what you get: My Writing Twin gives you a profile, a prompt and a connector for the models you already use; ScriptGrain also drafts with the profile, scores each draft against your voice, and lets you refine the profile with more samples as your writing changes.

My Writing Twin is a one-off $49-$249 for a profile you take away. ScriptGrain is £12-£299/mo and includes the generation. If you only want the prompt, theirs is cheaper; if you want the writing done and checked, the comparison is not really price against price.

ScriptGrain is not a replacement for the assistant you already use. It measures your voice from writing you have published and hands that profile to ChatGPT, Claude, Muse or any MCP client through its connector, then scores every draft against it. ScriptGrain makes your AI better at being you.

How the connector works

Every AI voice tool compared: prices, scores and API access, checked

ScriptGrainMy Writing Twin
Entry price£12/mo (Writer, 1 voice profile)$49 one-off (Starter)
Free tierYes, See your own voice profile and try one draft. No card.Yes, 1 use case, 2,000 words of samples
Voices on entry plan1 (3 on Operator, £29/mo)One twin per account; 3 style-matrix sections (Starter)
How the voice is built45 measured style attributes, extracted from writing you have already publishedA 2,000-word sample plus a 10-15 minute questionnaire produces a written style profile and a compact "runtime block": a system prompt you paste into any LLM. It also runs as a live MCP connector for AI agents and assistants, with a Chrome extension for writing in the browser.

My Writing Twin figures read from their pricing page on 30 September 2026.

What My Writing Twin is built for

My Writing Twin is built for people who already have a favourite model and want it to sound less generic. That's the job. It doesn't try to replace ChatGPT, Claude or Gemini; it tells them how you write and gets out of the way.

The pitch is stylometry. The company describes its approach as "computational stylometry, the same science used in forensic linguistics", with a 7-dimension framework covering formality, rhythm, transitions, punctuation, directness, vocabulary and context adaptation. It says that framework extracts "50+ style markers", and it points to a whitepaper citing 50+ peer-reviewed sources (all from mywritingtwin.com, read on 2026-09-30). Fair enough. That's a serious foundation, and it's more than most style tools in this space bother to claim.

The buyer it suits is fairly clear. Someone who writes regularly, uses one or two assistants daily, and wants a portable description of their voice they can drop into any of them. A consultant, say, or a founder who drafts in a chat window every morning. They don't need a new writing environment. They need the existing one to stop producing the same smooth, slightly corporate paragraph for everyone.

How My Writing Twin handles voice

The idea is simple, and simple is a virtue here. You supply a writing sample (2,000 words on the free tier) and complete a questionnaire that takes 10 to 15 minutes. Out come two things: a Writing Style Profile and a Runtime Block.

The Writing Style Profile is a readable analysis with examples from your own writing. You can read it, nod, wince, and learn something about your habits. The Runtime Block is the working part: a compact system prompt for ChatGPT, Claude, Gemini or any LLM. You paste it in and the model has instructions about your style. It also runs as a live MCP connector for AI agents and assistants, with a Chrome extension for writing anywhere in the browser.

Two features deserve credit. The questionnaire adds self-report to the mix, so the profile reflects how you think you write as well as what a sample shows. That's a deeper input than purely automated analysis, and for someone with a limited back catalogue it might matter a lot. And the profile is portable. It isn't tied to one vendor's model, which means a model change doesn't mean starting again.

Now the limits, stated plainly. There's a "Writing DNA Snapshot" that compares your profile with literary figures on the same dimensions. That's a nice bit of self-knowledge. What there isn't is a score of each new draft against your own writing. So you get a description of your voice and an instruction set for a model, but no measurement of whether any given draft actually landed on it. You judge that by eye.

How ScriptGrain handles voice

ScriptGrain starts from published writing rather than a questionnaire. You give it work the person or brand has already put out, and it measures a profile of 45 attributes in 8 layers: lexical, syntactic, tone and register, rhetorical, punctuation and format, function words, content patterns, and quirks and cadence. The full method is set out in the voice measurement framework, and the attribute reference lists what gets measured.

Those attributes are concrete. Average sentence length and its variance. Contraction rate. Comma density. Semicolon and dash habits. Formality on a 0 to 10 scale. Confidence versus hedging. Opening style. Argument structure, whether evidence-first, conclusion-first or narrative. Paragraph openers. Pronoun mix. Numbers, not adjectives.

A profile needs at least one sample of 50+ words. The recommendation is three or more pieces and about 3,000 words, because more material steadies the estimates. Extraction usually takes under a minute; the median was 39 seconds when measured on 2026-09-28.

Then comes the part that separates the two products. Every draft is scored from 0 to 1 against the profile. That's voice match. On-voice drafts typically land between 0.85 and 0.95, partial matches between 0.5 and 0.8, and a clearly different voice under 0.45. The score arrives with per-feature deltas, so you see the draft's value next to the profile's value: sentences running longer than yours, fewer contractions than you'd use. Polish revises a draft toward a target score (the default is 0.9), and the app learns from the edits you save.

The profile reaches ChatGPT, Claude, Gemini, Cursor, Meta Muse and any MCP client through the MCP server and the REST API, read live each time, on every plan including Free. It is never a pasted prompt. That's the architecture choice worth understanding: the profile stays in one place and gets corrected in one place.

The same job in each tool

Take a concrete job. A founder needs a 600-word product announcement that sounds like her, and she has a stack of past posts to learn from.

In My Writing Twin, she uploads a sample, answers the questionnaire, and receives her Writing Style Profile and Runtime Block. She reads the profile, recognises herself, pastes the block into her preferred assistant, and asks for the announcement. The draft comes back in something close to her register. She reads it, decides two paragraphs are too polished, rewrites them, and publishes. The judgement call at the end is hers alone, and for plenty of people that's exactly right. She knows her own voice better than any number does.

In ScriptGrain, she adds the same posts and gets a profile in under a minute. She drafts with it, either inside the app or through an assistant connected by MCP, and the draft comes back with a voice-match score and per-feature deltas. Say it lands in the partial band. The deltas show the gap: sentences longer than her variance allows, contractions well below her rate. She runs Polish toward the 0.9 default, reads the result, and saves the edits she makes, which the app then learns from. You can see the exercise at small scale with the style analysis tool.

Same job, different loop. One ends with a well-briefed model and a human eye. The other adds a measured check and a way to correct the profile when her writing shifts. Neither loop is wrong. They suit different appetites for checking.

Which one fits which situation

Choose by how you work, not by which feature list is longer.

Be honest about the first bullet. If your whole requirement is a good system prompt and a browser extension, My Writing Twin does that well and costs less up front. Paying for generation and scoring you won't use is a bad trade. The price gap only looks unfair if you ignore what's included.

Using ScriptGrain alongside My Writing Twin

Running both is less odd than it sounds. Treat them as two lenses on the same writer. The My Writing Twin profile is a readable self-portrait built partly from your own answers; the ScriptGrain profile is a measured one built from what you published. Where they agree, you can trust the picture. Where they disagree, you've learned something: either you describe yourself as more formal than you write, or your published work is shaped by an editor rather than by you.

Practically, keep one job per tool. Use My Writing Twin's Runtime Block or MCP connector wherever you want a quick, lightly styled draft from the assistant you're already in. Use ScriptGrain when a piece matters enough to be checked: run it through the brand voice consistency checker, and screen it with the AI cliché checker before it goes out. If you work through an assistant, both ChatGPT and Claude can read the ScriptGrain profile live, so there's nothing to paste. Developers can go further through the API and MCP server.

Moving from one to the other is straightforward, because neither needs the other's output. Your source material is your writing. Feed the same samples to both and compare what each says. Expect the wording to differ and the broad outline to match.

One housekeeping note. Keep a single source of truth for corrections. If you fix a habit in ScriptGrain's profile, don't assume the pasted block elsewhere knows. Update it by hand, or retire the pasted copy. Prices and plan details change, so check the current terms on the pricing page and on My Writing Twin's own site before committing either way. For a wider view across other tools, the comparison matrix and the comparison index pick up where this page stops.

Where My Writing Twin is stronger

The closest thing to ScriptGrain in the market: it leads with computational stylometry and forensic linguistics, and sells a profile rather than a subscription.

If you already have a model you like and just want it to sound like you, buying a portable profile once and pasting it in is a perfectly sensible thing to do, and cheaper than any subscription.

Questions

Is ScriptGrain better than My Writing Twin?

They solve the problem differently. Both measure style rather than describe it, which makes this the fairest comparison here. The difference is what you get: My Writing Twin gives you a profile, a prompt and a connector for the models you already use; ScriptGrain also drafts with the profile, scores each draft against your voice, and lets you refine the profile with more samples as your writing changes. If you already have a model you like and just want it to sound like you, buying a portable profile once and pasting it in is a perfectly sensible thing to do, and cheaper than any subscription.

How much does My Writing Twin cost compared with ScriptGrain?

My Writing Twin's entry plan is $49 one-off (Starter), with this free tier: Yes, 1 use case, 2,000 words of samples. ScriptGrain starts at £12/month (Writer, 1 voice profile) with a free plan: See your own voice profile and try one draft. No card. My Writing Twin figures read from their pricing page on 30 September 2026.

Does My Writing Twin learn my writing voice?

A 2,000-word sample plus a 10-15 minute questionnaire produces a written style profile and a compact "runtime block": a system prompt you paste into any LLM. It also runs as a live MCP connector for AI agents and assistants, with a Chrome extension for writing in the browser. ScriptGrain instead measures 45 style attributes from writing you have already published, and scores every draft against that measured profile.

Other comparisons