Paste your writing into ChatGPT and ask it to "match your style." It'll produce something. It might even be good. But read it back a few times and something feels off — slightly too polished, slightly too generic, like a version of you that went to too many workshops and forgot to have opinions.
The problem isn't the AI. The problem is that it doesn't know you. It knows the prompt you gave it, which is a description of your voice — and descriptions of your voice are almost always wrong. Not because you're a bad observer, but because the gap between what you think you sound like and what you actually sound like on the page is wider than most people expect.
A Riterly writing profile is built to close that gap. Here's what it actually contains.
The Voice Workshop: built from your writing, not your answers
Most tools ask you questions. "How would you describe your tone?" "What's your target audience?" "Formal or casual?"
Riterly skips those questions. The Voice Workshop — the mechanism that builds your profile — starts with your actual writing instead.
You upload samples of your writing: past posts, articles, newsletters, whatever you've published. They don't need to be your best work. They need to be your real work, the kind of writing that actually came from you on a regular day. Riterly analyzes those samples the way an editor would: sentence rhythm, recurring phrases, how you open a piece, how you close one. Not broad impressions — patterns that show up consistently across everything.
Before you've uploaded anything, there's already a foundation in place. When you select your niche at the start of the Voice Workshop, Riterly pre-loads a template built for that content type. So you're not building from a blank slate. You're correcting and shaping something that already has the rough contours of what you need — starting ahead of zero.
After the analysis runs, the Voice Workshop populates the profile fields with what it found. You review them, adjust anything that's off, and fill in what the samples couldn't tell it. The wizard gets most of the way there. You shape the rest.
The six dimensions, and what each one changes
The profile isn't a free-text description or a long prompt you paste in at the start of every session. It's a structured document organized into six sections, and each one does specific work.
Who you are
This section grounds the AI in context — your name, your professional background, the perspective you're writing from. Without it, the output defaults to a generic expert voice. With it, the drafts reflect that you're a solo consultant who's been in the industry for twelve years, or a developer writing for other developers, or a founder who's still doing all the work themselves.
Audience
Who are you actually writing for, and what do they already know? This field shapes vocabulary, assumed knowledge, and how much explaining to do. Get it wrong and the output either talks down to your readers or leaves them behind. A post written for senior engineers needs to move faster and assume more than one written for team leads who are just getting into technical work.
Voice and tone
This is where the feel of the writing lives — the personality, the register, the emotional temperature. Dry and precise. Warm but direct. Skeptical with a sense of humor. These aren't just descriptors. They calibrate how the model approaches sentences: whether it hedges or states, whether it explains or shows, whether it softens or leans in.
Language patterns
This section has two parts, and both matter.
Phrases you use — the vocabulary, expressions, and rhythms that are distinctly yours. If you always say "the actual problem is" or "here's what I mean by that," this is where that goes. The output starts reaching for the same phrases naturally.
Phrases you avoid — this is the field that does more quiet work than any other. This is where AI-isms die. "Leverage," "unlock," "game-changer," "delve into," "it's worth noting" — any phrase that makes you cringe when you read it back goes here. The model actively steers away from what you've flagged, which is the difference between a draft you'd clean up and a draft you'd delete.
Post structure
How do you typically build a piece? Do you open with a problem or a scene? Do you use headers or let the prose run? Do your posts end with a takeaway or stay open-ended? This section means the model isn't starting fresh every time — it builds to the shape your writing naturally takes.
Topics
What are you actually here to write about, and what do you never touch? Topic scope keeps the output relevant and focused. It also prevents the model from wandering into adjacent territory that might be technically accurate but isn't what you do.
Beyond these six sections, the structured editor and advanced markdown mode give you access to additional fields: CTAs, guardrails, brand messaging, code examples. But the wizard covers what matters most to get the profile working.
The Sandbox: where confidence gets built
Once the Voice Workshop is done, your profile doesn't immediately start guiding real drafts. It moves to the Sandbox first.
The Sandbox is a proof step. You generate test drafts, read them back, and ask a simple question: does this sound like me? Not "Is this good writing?" — but "would I have written this?" If the answer is no, you go back into the profile, adjust what's off, and run another test.
Sandbox runs can't be published. They exist for one purpose: to build confidence in the profile before it guides anything real. This is a deliberate separation. The Sandbox is for getting the profile right. The pipeline is for creating content. Conflating the two means you're never sure whether the output reflects a calibrated profile or a half-built one.
Most people find that three or four sandbox runs is enough. You read the first draft, catch something that sounds slightly off — maybe the tone is more formal than you'd actually write, or the openings aren't landing the way yours do — you fix it, run another, and it tightens. By the time the profile goes live, you've already seen what it produces.
A profile is never finished
The first draft a calibrated profile produces is a starting point, not a destination. That framing matters, because it changes how you respond to output that isn't quite right.
When a draft needs shaping, that's not failure — it's information. The gap between what came back and what you'd actually publish tells you something specific about the profile. Maybe the voice and tone section needs a sharper descriptor. Maybe there's a phrase in your writing you haven't flagged yet, and it keeps showing up in the drafts. You update the profile, and the next draft is closer.
Multiple profiles work this way independently — each one has its own niche selection, its own Voice Workshop output, its own language patterns and structural preferences. A profile built for your technical newsletter doesn't bleed into the one built for your company blog. They're separate, and they sharpen separately.
Over time, the output gets more accurate — not because the AI gets smarter, but because the profile gets more precise. Each refinement closes the gap a little more.
If you want to see how the Voice Workshop works in practice — what to bring to it and what to expect when it runs — take a closer look at how it's built.
COMMENTS
Login or Register to post comments.