ScriptGrain vs Anyword

ScriptGrain measures your voice as 45 attributes from your actual writing and scores every draft against the measured profile. Anyword pairs a detected tone of voice with brand rules and approved phrasing, and is built to predict which copy converts.

The actual difference

Anyword detects a tone of voice from sample text or a page and pairs it with brand rules and approved phrasing. ScriptGrain measures 45 style attributes from your writing, shows every number, and scores each draft against the measured profile feature by feature.

Anyword's entry plan is $49/mo for one brand voice. ScriptGrain's Writer plan is £12/mo for one profile, and the free tier lets you see the profile before paying anything.

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

	ScriptGrain	Anyword
Entry price	£12/mo (Writer, 1 voice profile)	$49/mo
Free tier	Yes, See your own voice profile and try one draft. No card.	None
Voices on entry plan	1 (3 on Operator, £29/mo)	1 brand voice
How the voice is built	45 measured style attributes, extracted from writing you have already published	Brand rules, a copy bank of approved descriptions, tones of voice and target audiences. A tone of voice can be set by hand or detected from sample text or a web page.

Anyword figures read from their pricing page on 3 August 2026.

What Anyword is built for

Anyword is a performance tool first. Its own positioning calls it "Gen-AI Performance Writing" for marketers and agencies, and the job it is built around is predicting how copy will convert. Voice is part of the product, but it sits inside that larger purpose. You write the ad, the email, the landing page line; Anyword tells you which variant it expects to win.

That shapes everything else. The buyer Anyword has in mind is running campaigns, testing variants and answering to a conversion number. Authorship is secondary. If a line reads a little off-brand but lifts click-through, the tool is on the side of the lift.

Anyword says its prediction score is 82% accurate. Treat that as the vendor's claim, which it is, but notice what kind of claim it is. It is a statement about marketing outcomes, trained on marketing data. ScriptGrain makes no equivalent statement because it attempts no equivalent thing. So if you are choosing between the two, the first question is what you are actually trying to improve: how well the copy performs, or how much it sounds like a particular writer or brand.

How Anyword handles voice

Anyword's brand voice setup has four working parts. Brand Rules hold banned and approved terms. A Copy Bank stores approved descriptions, the phrasing you want reused. Custom Formulas control tone and format. Target Audiences record who the copy is for. A tone of voice can be set by hand, or detected from sample text or a web page.

That last option is the quick route, and it works the way you would hope. Paste in a few paragraphs or point it at a page, and it proposes a tone. From there, the rules and the copy bank do the enforcing: banned terms stay out, approved phrasing stays in.

Checking happens in a "Tone of Voice Insights" panel. It compares generated copy against predefined brand tone traits, flags deviations and suggests fixes. That is a real feedback loop, and it is useful for a team that needs a second pair of eyes on every draft.

The trait model behind that panel, though, is not published. You can see that a draft was flagged; you cannot see the list of traits or the numbers underneath. For a marketing team that mostly wants guardrails, that is fine. For someone who wants to know precisely why a draft drifts, it is a black box. And to be fair, Anyword never claims otherwise. It sells brand control, and on that it delivers.

The brand rules deserve credit. Banned-term control at fine grain is something many teams genuinely need, particularly in regulated sectors or where legal has opinions about a word. ScriptGrain does not attempt that, and the section on where Anyword is stronger says so plainly.

How ScriptGrain handles voice

ScriptGrain starts from a different place. It measures a voice from writing you have already published, and it produces 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 attribute reference lists them all.

What does an attribute look like? Average sentence length and how much it varies. Contraction rate. Comma density. Semicolon and dash habits. Formality on a scale from 0 to 10. Confidence against hedging. Opening style, argument structure (evidence-first, conclusion-first or narrative), paragraph openers, the pronoun mix. Each one is a number or a named setting, and you can see all of them.

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

Then comes the enforcement. Every draft is scored from 0 to 1 against the profile; that is the voice match score. On-voice drafts typically land between 0.85 and 0.95. Partial matches sit between 0.5 and 0.8. A clearly different voice comes in under 0.45. The score arrives with per-feature deltas, so you see the draft's value next to the profile's value for each feature. The method is published in the measurement framework, which means you can argue with it.

Polish then revises a draft toward a target score, 0.9 by default. Drafts can also be written with the profile as a constraint from the start, and the app learns from the edits you save.

One more point, because it changes how you would use it. The profile is not a pasted prompt. It reaches ChatGPT, Claude, Gemini, Cursor, Meta Muse and any MCP client through the MCP server and the REST API, and it is read live each time. Both are on every plan, including Free.

The same job in each tool

Take one job: a founder wants a product announcement that sounds like the founder, not like a press office. Short sentences, dry humour, a conclusion in the first line.

In Anyword, you would point the tool at the founder's existing blog or paste a few posts, and let it detect a tone of voice. You would add brand rules: two banned buzzwords, a couple of approved product descriptions in the copy bank. You would generate variants. The Tone of Voice Insights panel would flag any that drift from the detected traits and suggest fixes. Then the performance score would tell you which variant it expects to convert best. The output is a set of on-brand, ranked options, and the ranking is the bit you cannot get elsewhere.

In ScriptGrain, you would feed in the founder's published writing and get the 45-attribute profile in under a minute. You would see that the founder's sentences average short with high variance, that contractions run high, that the argument structure is conclusion-first. You would draft the announcement, and the score would come back. Say it lands in the partial band, somewhere between 0.5 and 0.8. The deltas would show why: perhaps the draft's sentences run too evenly, or the openers are too formal. You would run Polish toward 0.9 and check again.

Same job, different questions answered. Anyword tells you which version should do best with an audience. ScriptGrain tells you which version sounds most like the person, and which specific habits are off. Neither answer substitutes for the other.

Which one fits which situation

The short version: pick by what you are optimising.

Paid ads and conversion testing. Anyword. Predictive scoring trained on marketing data is its core, and ScriptGrain has no equivalent.
Strict terminology control. Anyword. Banned and approved terms, plus a copy bank, give you rules that hold.
A named author's voice, such as a founder, columnist or ghostwriting client. ScriptGrain. You get measured attributes and a score per draft, not a detected tone.
Needing to know why a draft sounds wrong. ScriptGrain. Per-feature deltas show the exact gap.
Using your voice inside ChatGPT, Claude or Cursor. ScriptGrain. The profile is read live through MCP and the API on every plan.
A team that wants one tool and mostly needs guardrails. Anyword is the simpler fit.

Then there is the middle case. Plenty of teams need both: a voice that is measurably theirs, and copy that has been tested against what converts. That is where running both makes sense.

On cost, the entry plans differ in shape. Anyword's Starter plan was $49 a month on 2026-09-20, for one seat and one brand voice; Data-Driven was $99 for three seats and still one brand voice, and Business, at five brand voices, is custom. ScriptGrain's Free tier lets you see the profile before paying, and Writer is £12 a month for one profile. The pricing page has the rest.

Using ScriptGrain alongside Anyword

The combined workflow is straightforward. Use ScriptGrain to establish and score the voice, then use Anyword to choose between variants that already sound right.

In practice that looks like this. Build the profile from published writing. Draft in your usual tool with the profile feeding in live, so the first pass already leans on-voice. Score it, use Polish if it falls short of your target, and only then take the surviving drafts into Anyword to rank them by predicted performance. You keep authorship and you keep the conversion check.

A few practical notes. Anyword plans carry a limited number of brand voices, one on Starter and Data-Driven, and five on Business. If you write for several clients, that matters; ScriptGrain's Studio plan holds 10 profiles across 3 seats, and Agency holds unlimited profiles across 10 seats. Both tools can hold your rules, so keep the banned-term list in Anyword, where it is enforced, and the measured voice in ScriptGrain.

On integration: Anyword lists "full API access" on Enterprise, and we found no MCP server. ScriptGrain's API and MCP server are on every plan, including Free; the API and MCP page covers the setup.

To see whether a finished draft has wandered, the brand voice consistency checker gives a quick read, and voice drift explains why it happens over a long run of pieces. Moving from one tool to the other is low-risk either way, since the profile comes from your writing, not from either product's settings.

Where Anyword is stronger

Performance-led: predictive scoring of how copy will perform, trained on marketing data.

Predictive performance scoring, which nothing in ScriptGrain attempts
Strong fit for paid-ads copy where conversion data matters more than authorship
Fine-grained brand rules and banned-term control

If your priority is predicting which variant converts, Anyword is built for that and ScriptGrain has no equivalent.

Questions

Is ScriptGrain better than Anyword?

They solve the problem differently. Anyword detects a tone of voice from sample text or a page and pairs it with brand rules and approved phrasing. ScriptGrain measures 45 style attributes from your writing, shows every number, and scores each draft against the measured profile feature by feature. If your priority is predicting which variant converts, Anyword is built for that and ScriptGrain has no equivalent.

How much does Anyword cost compared with ScriptGrain?

Anyword's entry plan is $49/mo, with no free tier. ScriptGrain starts at £12/month (Writer, 1 voice profile) with a free plan: See your own voice profile and try one draft. No card. Anyword figures read from their pricing page on 3 August 2026.

Does Anyword learn my writing voice?

Brand rules, a copy bank of approved descriptions, tones of voice and target audiences. A tone of voice can be set by hand or detected from sample text or a web page. ScriptGrain instead measures 45 style attributes from writing you have already published, and scores every draft against that measured profile.