Back to Learn
Taste
7 min read

Make AI sound like you

Voice profiles explained. Because "write in my tone" has never worked.

You've tried it. You've typed "write in a professional but friendly tone" or "match my voice" or "make it sound like me." And AI gave you something that sounded like... everyone. A perfectly competent, perfectly forgettable piece of writing that could have come from any person in any industry on any day.

That's a context problem.

When you say "write in my tone," AI has no idea what your tone is. It doesn't know that you start sentences with "And" sometimes. It doesn't know you hate the word "leverage." It doesn't know you're warm but direct, or formal but funny, or blunt but kind.

AI can absolutely write in your voice. But first, you have to tell it what your voice IS. In a way it can actually use. That's what a voice profile does.

What a voice profile is

A voice profile is a document (a .md file) that captures the DNA of how you write. Not what you write about. How you write. The rhythms, the word choices, the things you always do and never do, the emotional register, the level of formality, the way you handle humor, the way you open a piece, the way you close one.

It's specific. Not "I write in a conversational tone" (useless) but "I use contractions. I start sentences with 'But' and 'And.' I never use semicolons. My paragraphs alternate between 1 sentence and 4-5 sentences. When I'm excited, my sentences get shorter. I swear occasionally but never gratuitously. I never use the word 'crucial.'"

That level of specificity is what transforms AI output from "generic professional writing" to "this actually sounds like something I'd publish."

Why "write in my tone" doesn't work

1. "Tone" is too vague

Tone is one dimension. Voice is sentence structure, vocabulary, rhythm, and formatting. Saying "match my tone" gives AI one slider out of twenty.

2. AI defaults to average

Without specific instructions, AI falls back to the statistical average. It's the writing equivalent of a hotel room: functional but belonging to no one.

3. You can't describe it

Articulating WHY your writing sounds like yours is hard. You need to analyze patterns across dozens of pieces to see what makes you, you.

A voice profile helps AI work ALONGSIDE you.

What goes inside a voice profile

A good voice profile covers these areas in detail.

Not your bio. Your writer identity. How you relate to your reader. Are you a peer? A mentor? A guide? A friend? Are you someone who shares the messy process or someone who presents polished results? Do you write as someone who's figured it out or someone who's figuring it out alongside the reader?

This matters more than people realize. An AI writing as "an expert sharing wisdom" produces fundamentally different text than an AI writing as "a peer who's a few steps ahead." Same content, different relationship, different voice.

How to build one (two paths)

Path 1: Build it yourself
Gather 5-10 pieces of your best writing and analyze them.

Go through each area above and write specific observations. Not "my tone is friendly" but "I address the reader as 'you' directly. I use humor about once every 3-4 paragraphs, usually a dry observation."

This works, but it's time-consuming and hard to do about yourself. Most people stay too general because they can't see their own patterns clearly.

Path 2: Let AI build it
Use the Voice Profile Builder skill to automate the analysis.

A structured system that analyzes your samples, interviews you about your preferences, and produces a comprehensive .md file in about 20 minutes.

Learn how to build working context

How to use a voice profile

1

Option 1: Upload it each conversation

Start a new Claude conversation. Upload your voice profile. Then give your task. Claude will reference the profile and write accordingly.

2

Option 2: Add it to a Claude Project

Create a Project in Claude and add your voice profile as a project file. Now every conversation in that project automatically has access to your voice.

3

Option 3: Pair it with skills

When you run a writing skill from RobotsOS and your voice profile is in the same project, the skill follows its methodology IN YOUR VOICE. This is the compounding effect.

What changes when you have one

Before voice profile

You spend 30% of your time on cleanup. Rewriting generic parts. Removing words you'd never use. Adjusting rhythm. Every piece needs a heavy edit.

After voice profile

The first draft is already 70-80% in your voice. You're editing for content, not for tone. Cleanup time drops dramatically. Output is consistent.

When AI already knows your voice, you can focus on ideas, structure, and substance.

The honest limitation

A voice profile doesn't make AI write exactly like you. Nothing does. Your voice is alive and evolving and responds to context in ways no document can fully capture.

What a voice profile does is get AI into the right neighborhood. Close enough that the editing is about refinement, not overhaul. Close enough that a reader wouldn't notice the difference unless they were looking for it.

Where to go from here

One more thing

The fear that AI will erase what makes your writing yours is real. It's also solvable. Not by avoiding AI. By teaching AI who you are before you ask it to write for you.

Your voice isn't generic. It has patterns, preferences, habits, quirks, hard lines. A voice profile codifies those into something AI can follow.

The free newsletter is where I teach this stuff every week. And RobotsOS is where the Voice Profile Builder and the rest of the skill library live.