Naturable.
Chrome extension

Free · open-source rules · runs in your browser

Find the parts of your draft that sound like AI.

Naturable flags assistant chatter, template openers, vague sources, stacked hedges and words like delve and leverage. Each flag comes with an explanation, so you decide what to change. Nothing you paste leaves your browser.

Check a draft

Runs in your browser · nothing is uploaded

Great question! In today's digital landscape, experts believe this robust technology is not just a tool, it's a pivotal shift. In order to leverage it, teams could potentially rethink their work. The future looks bright. I hope this helps!

Open the full editor
64 / 100 Some AI patterns

Pattern score. It doesn't tell you whether a person or an AI wrote the text.

8 strong · 3 suggestions

  1. StrongVague source
    “experts believe”The attribution cannot be checked as written.Name the source, date, or remove the attribution. Learn more
  2. StrongGeneric conclusion
    “The future looks bright”The ending makes a broad claim without evidence or action.End with a specific result, limit, or next step. Learn more
  3. StrongNot-just / but frame
    “is not just a tool, it's”The 'not just X, it is Y' frame often sounds rehearsed.State the main claim directly without the fake contrast. Learn more
  4. SuggestionFormulaic signpost
    “In today's digital landscape”The opening announces content instead of delivering it.Begin with the concrete point. Learn more
+7 more in the full editor →
  • Your text stays on your device
  • Every flag explains itself
  • 39 open rules, English + Chinese beta
  • No account, no word limit

Learn the patterns

All guides →
Word list AI words to avoid Delve, tapestry, leverage, robust, and the stock openers that come with them. Each one gets a plainer word you can use instead. delveleveragein order to Guide Signs of AI writing The tells readers notice first, how to fix each one, and why none of them proves who wrote a text. Read the guide → Honest review Do AI humanizers work? What one-click rewriters actually change, what they break, and a better way to edit an AI-assisted draft. Read the review →
Em dash checkerCount dashes per 1,000 words and swap them one by one Burstiness calculatorSee your sentence-length rhythm at a glance

Check where you write

The Chrome extension underlines the same patterns in Gmail, LinkedIn, Notion, Reddit and most text boxes. It runs locally, like the site.

Chrome extension: coming soon

Open rules you can check

Every rule is public JSON with examples and tests. Use it in your own tools, or as an Agent Skill for Claude and Codex.

npx naturable-rules draft.md View on GitHub →

How it works

  1. 1 · Paste or typeEmails, essays, posts, product copy. English, with Chinese in beta.
  2. 2 · Read whyEach flag names the pattern, explains why it reads as AI-written, and suggests a direction.
  3. 3 · Fix what you agree withApply safe fixes in one click, ignore the rest, and recheck. Numbers and names are never changed for you.

Questions

Can Naturable tell whether a text was written by AI?

No. It finds writing patterns that are common in chatbot output, and people write them too. Don't use the score to judge someone else's work.

Will this help my text pass an AI detector?

That isn't what it's for. We don't test against detectors or promise any result. The goal is clearer writing that sounds like you.

Is my text uploaded?

No. The rules run in your browser. Drafts are saved only on your device.

Is it free?

Yes. No account, no word limit, no watermark.

Where do the rules come from?

They're open: every rule is public JSON with a reason, a suggested fix and before/after examples that are tested on every change. You can read them, use them in your own tools, or suggest new ones.