How to Make Claude Sound Like You, Not Like Claude (2026)

To make Claude sound like you, give it samples of your own writing and ask it to describe your voice before it drafts anything. Then list the habits you want avoided, set a length, and edit the result yourself. Claude can match rhythm and tone from a sample, but it can't supply your opinions, your examples or your small odd details. Those are what make a draft read as yours.

The short version

1. Pick two or three things you wrote yourself, 150 to 300 words each. Emails and posts you liked are better than formal essays.
2. Ask Claude to describe your voice in five bullet points, and correct anything wrong.
3. Give it the job, the reader and a length limit.
4. Add a banned list of habits (below).
5. Edit the draft by hand: add one real detail, cut one polished line, rewrite the opening.

Why do Claude drafts sound the same?

With no sample, a model writes the safest version of the task. That means even sentence lengths, a tidy list of three, a neat closing line that repeats the point, and mild confidence about everything. None of it is wrong. It's just what everyone's prompt produces. The 10 signs of AI writing lists these habits in detail if you want a longer checklist.

The fix is not a magic phrase like "write like a human". Vague instructions give vague results. Concrete input does the work: your samples, a specific reader, and facts only you know.

A prompt that works

Use two messages. First:

Here are three things I wrote. Describe my writing voice in five bullets: sentence length, how formal I am, how I open and close, words I overuse, and anything unusual. Don't write anything yet.

[paste samples]

Read the bullets. If one is off ("you say you're formal, but I'm not"), say so. Then:

Using that voice, write [a 150-word update to my team about the launch delay]. The reader is [my manager, who dislikes fluff]. Use these facts only: [list]. Don't add facts. Avoid: opening with a greeting about the week, "I hope this finds you well", lists of three, hype words, and a closing sentence that restates the message. Plain words. Under 150 words.

Two things carry most of the weight: "use these facts only" stops invented detail, and the banned list stops the stock habits.

Anthropic also documents custom styles in Claude, where you upload or paste a writing sample and Claude builds a reusable style from it (see Anthropic's help article on styles). It saves repeating step one, and the same advice applies: use a sample that really sounds like you.

A worked before and after

The text below was written for this post. It is an illustration, not output from any tool, and no detector was run on it.

A typical default draft:

I hope this message finds you well. I wanted to reach out regarding the launch timeline. Unfortunately, we've encountered some unexpected challenges that will push the release from November 4 to November 18. We are committed to delivering a high-quality product, and our team is working diligently to ensure success. Please don't hesitate to reach out with any questions.

After a voice sample, a banned list and a hand edit:

Heads up: launch moves from November 4 to November 18. The payment integration failed testing twice and I'd rather fix it than ship it. QA gets the build on November 11. If that date is a problem for the client call, tell me today and we'll sort something out.

What changed: the dates stayed, the vague "challenges" became the real cause, the filler closer became a concrete ask, and the sentences got shorter. The cause line is the kind of detail only you can add.

Edit pass: what to change by hand

1. Replace one vague phrase with a specific. "Unexpected challenges" becomes the actual problem.
2. Cut the closer. If the last sentence only restates the message, delete it.
3. Break the rhythm. Make one sentence very short and merge two others.
4. Check every fact. Names, dates, numbers and quotes should match your source. Models can state wrong things confidently, so see the fact-check checklist and the real hallucination cases.
5. Read it aloud. If you'd never say a line, change it.

Doing the edit pass in Unflag

If the draft is close but still reads flat, Humanize AI rewrites it so it reads more naturally, and you can set a tone. The tone menu has 18 options, including Casual, Direct, Friendly, Confident and Professional. It keeps the language of your text, and before rewriting it picks out prices, percentages and other figures and restores any the rewrite dropped. Check the numbers anyway, since no rewrite keeps every figure with certainty.

1. Open Humanize AI on the web or in the iPhone or Android app and paste the Claude draft. It needs at least 50 words.
2. Pick the tone closest to your voice. Direct or Casual often suits work updates.
3. Run it, then do the edit pass above on the result.
4. Optionally run the AI Detector as a self-check on how the draft reads. Its score is a rough signal about your own text. It doesn't predict what any other detector will say.

The free plan takes up to 600 characters per input, a few uses a day, and the web needs sign-in. Unflag Plus raises the limit to 10,000 characters. As of October 2026, Plus is $8.99 a month or $45.99 a year on the web. Results are saved to your History.

For emails specifically, the AI tone rewriter guide covers matching tone to the reader. If you're unsure which kind of tool you need, read paraphraser vs AI humanizer.

Does it matter which Claude model I use?

Newer models, including Claude Opus 5.5 and Sonnet 5.5, write well, but the pattern holds: a model without your samples and facts will write a generic version. We haven't compared models for voice, so don't pick one on that basis. Spend the effort on the sample and the edit.

Is it okay to use Claude for my writing?

It depends on where the writing goes. Many workplaces and publishers allow AI drafting, and some ask you to disclose it. Schools often have strict rules. Read the policy first and follow it. Using these steps to make a draft read well is fine. Using them to hide AI use where it is prohibited is not something this guide supports.

Honest limits

A voice sample teaches style, not substance. Claude may still invent details, flatten humor, or slip back into old habits halfway down a long reply. Unflag can't guarantee any output will pass a third-party detector, and no amount of editing replaces having something real to say. For long pieces, work a section at a time.

How we put this together

We checked Anthropic's published help page on styles and Unflag's Humanize AI, tone and detector behavior against its own tools (October 2026). The example drafts were written for this post and carry no detector scores. This is general writing advice, not a test of any model.

Frequently asked questions

How do I make Claude sound like me?

Paste two or three things you actually wrote, tell Claude to describe your voice first, then ask for the draft in that voice. Add a short list of habits to avoid and a length limit. Finish with a hand edit: change the opening line and one sentence that only you would write.

Why does Claude's writing sound like AI?

Without a sample, any model drifts toward safe, even prose: balanced sentences, tidy lists of three, summary closers and stock phrases. Specific details are what make writing sound personal, and Claude only has the ones you give it, so the fix is better input and a final edit.

Can Claude copy my writing style?

Yes, to a degree. Anthropic's help center describes custom styles that Claude builds from a writing sample you upload or paste. It captures patterns like sentence length and tone, but it does not know your opinions or memories, so you still need to add real details.

How do I stop Claude from using words like delve and tapestry?

Put a short banned list in your prompt, for example stock openers, hype words and closing summaries, and say to write plainly. Then scan the draft yourself, because models slip back into habits over a long reply. Cutting the habits in an edit pass works more reliably than asking once.

Can Unflag edit a Claude draft?

Yes. Paste the draft into Humanize AI on the web or in the iPhone and Android apps and pick a tone. It rewrites in the same language and restores figures the rewrite dropped, so check numbers yourself. The free plan takes up to 600 characters per input, a few uses a day, with sign-in on the web.

Do I need to tell people I used Claude?

Follow the rules of whoever you are writing for. Schools, employers and publishers set their own AI policies, and some ask for disclosure. Where AI help is allowed, you are still responsible for every claim in the text, so read and fact-check it before it goes out.