Writing That Sounds Like You (Not Like AI): A Practical Guide

May 28, 2026 • Anthony Guerriero

You can usually tell. There is a particular register that AI-generated writing has — professionally adequate, slightly over-formal, strangely even-toned, and missing the small idiosyncrasies that make a person's writing sound like theirs.

"It is important to note that..." "In today's rapidly evolving environment..." "This comprehensive approach..." "Let us explore..."

If you've submitted AI-written content and had someone ask whether you used AI, this is why. The tells are consistent and learnable, which means they're also fixable.

Using AI to assist your writing is not cheating. Using it badly — outputting generic prose and hitting send — costs you something real: the credibility that comes from writing that sounds like you.

The Core Problem: AI Writes to the Average

AI language models are trained on enormous amounts of human writing. When they produce text, they essentially produce a weighted average of how that type of content usually gets written. The result is technically correct, structurally sound, and completely devoid of your particular way of seeing things.

Your voice is not average. It is the product of your specific experience, your reading history, your industry, your personality, and the particular things you care about. AI cannot replicate that. But you can teach it to work within it.

Voice Calibration: Giving Claude a Starting Point

The simplest method for getting Claude to write more like you: show it your writing before you ask it to produce anything.

Here is a prompt structure that works.

"Here are three samples of my writing: [paste examples]. Study the voice, sentence rhythm, level of formality, and the kinds of examples I use. When I ask you to write something for me, use this as the voice model. Don't copy the content — model the style."

The samples you choose matter. Use writing you're proud of — a strong client email, a well-received article, a presentation section that landed well. Not rough drafts. Not writing you hate. The model will pattern-match to what you show it.

After you do this, ask Claude to write something short, read the output, and adjust. Tell it specifically what is off: "Too formal — I don't write like that." "The sentences are too long." "I would never use the word 'endeavor' — cut that register entirely." The calibration improves with each pass.

What to Change in Every Output

Even with good calibration, AI output requires editing to sound like you. Here is a reliable pass-through to make on anything Claude writes before you use it.

Cut the scene-setting sentences. AI often opens with a sentence that explains what you're about to do: "In this analysis, we will examine..." "This document explores the key considerations for..." Delete these. Start with the actual content.

Remove words that signal AI register. The words that consistently appear in AI output but almost never in authentic professional writing: "utilize" (use "use"), "endeavor," "comprehensive," "leverage" as a verb, "delve into," "it is worth noting that," "importantly," "in conclusion." Scan for these and cut them.

Break the pattern sentences. AI often produces sentences of remarkably similar length and structure, which creates a rhythm that sounds mechanical. Read your edited draft aloud. Anywhere you notice a monotonous cadence, vary it — make some sentences very short. Let some be longer. The rhythm of your actual voice is uneven.

Add something only you would know. This is the step that most transforms AI-assisted writing into something distinctly yours. Add one observation from your actual experience that the AI couldn't have produced. A specific example from a client situation. An honest counterpoint. A detail from your particular market or industry that general writing wouldn't include.

Voice Calibration Prompts That Work

These prompts are worth keeping in a document you can paste from:

For thought leadership posts: "Write this in the voice of a direct, plainspoken senior practitioner who has seen a lot and doesn't need to impress anyone. Short paragraphs. Specific examples. No filler."

For client-facing writing: "Write this as if you are a trusted advisor who is explaining something clearly to a smart client who doesn't have time for jargon. Warm but efficient. Facts before framing."

For internal memos or analysis: "Write this like a senior person briefing a colleague — direct, well-organized, no preamble. State the conclusion first."

Each of these pushes Claude away from the generic professional register toward something more specific. You'll still need to edit, but you're editing toward your voice rather than editing away from something generic.

The Common Tells to Remove

Beyond specific words, certain structural patterns appear consistently in AI writing and consistently signal that the text wasn't written by a human. These are worth recognizing:

The three-part list where each item starts the same way. ("First... Second... Third...") Fine occasionally; a crutch when overused.

The false pivot. "While X is true, it is also important to consider Y." This structure sounds like AI because AI uses it as a template for adding nuance. Real writers just say Y.

The wrap-up paragraph that summarizes what was just said. AI writing often ends with a paragraph that restates the main points. If your readers just read your post, they don't need a summary of it. Cut the recap.

The hedge stack. "It may be worth considering that in some cases it might be possible to..." Real professional writing takes positions.

What You Actually Sound Like

Before you write anything important with AI assistance, it's worth asking: what do I actually sound like?

Pull three or four pieces of your best writing. Read them. Notice the patterns. Do you use very short declarative sentences? Do you often start with a counterintuitive observation? Do you tend toward specific numbers and concrete examples rather than abstractions? Are you dry and precise or warm and discursive?

Those patterns are your voice. The goal of AI assistance is to produce output that fits within those patterns — and then to fill in the gaps with the things only you can add.

The finished product should be unambiguously yours. The AI is doing drafting work, not authoring.

Frequently Asked Questions

How long does it take to calibrate Claude to sound like me?
With a few good writing samples and two or three explicit feedback rounds, you'll see a noticeable improvement in one session. Complete calibration — where the output requires minimal editing — usually takes three to four sessions across different writing types.

What if I don't have good writing samples to share?
Start with what you have, even if it's not your best work. Voice patterns persist even in rough drafts. If you genuinely have no existing writing to share, describe your voice in words: "direct, somewhat dry, specific, low on abstraction, high on examples."

Should I disclose when writing is AI-assisted?
In most professional contexts, no disclosure is required. You are the author; the AI is a drafting tool, like a ghost-writer or a writing assistant. That said, check your specific industry or publication's policies — some have explicit guidelines.

I've given Claude feedback but it keeps reverting to its default style. Why?
Each conversation starts fresh. Your voice calibration doesn't persist between sessions unless you paste it in at the start. Keep a saved prompt block with your voice description and sample text that you paste at the beginning of any writing session.

What types of writing are hardest to make sound like you?
Short-form social media posts and emails are hardest because your voice is most concentrated in them. Long-form articles and reports are easiest because there is more room for the unique content only you would include.

Writing well with AI is a skill, and it develops fastest in a structured context. The Leverage Starter course ($199) covers AI-assisted writing workflows for professionals — including the calibration method, the editing pass, and how to integrate AI into your content production without losing the quality signal that actually matters.

Where this goes next

If you'd rather install this as a system than rely on willpower, See The Leverage Starter — or Turn Experience Into Income with Claude if you want the broader path.

Related reading from The Briefing

Not sure which program fits where you are? take the 2-minute course-fit quiz, or browse the full TLY course catalog.