The 99 Things That Make Your Writing Unmistakably Yours
Your colleagues can pick out your emails without reading the sender line. They recognise your reports by the second paragraph. In a meeting, your phrasing is distinctive enough that someone could quote you and the room would know exactly who said it.
Your writing voice is not vague. It is specific, built from hundreds of small decisions you make without thinking and repeat across everything you have ever written.
The real question is whether any tool has ever taken that seriously. Most have not.
What 99 Markers Actually Captures
Kretell's Voice Profile maps 99 distinct markers of how a specific person communicates, organised across 11 dimensions.
The visible markers are structural: sentence length, vocabulary complexity, paragraph architecture, use of emphasis, formatting habits. These are measurable and consistent, and they form the skeleton of a recognisable voice.
The markers that do the real work are quieter. Does this person lead with data or with an observation? Do their examples support an argument or replace it? When they establish authority, do they reach for range or for precision? Do they credit others first, or themselves? How does their formality shift between audiences? And what is their relationship to humour, dry and occasional, or absent altogether?
None of these patterns are accidents. They formed over years of working inside a specific professional context, in a specific market, inside a specific culture, and they carry real information about who this person is and where their expertise was built.
The Voice Profile maps them, and the output generates from that map rather than from the internet's average of how someone in your position is supposed to sound.
The Three Markers That Were Reserved
Out of the full architecture, three markers were left out of data collection on purpose. All three have to do with credential signalling, how a person represents their experience and qualifications.
The system could have collected them, and doing so would have produced more authoritative-sounding output. It would also have introduced hallucination risk. A tool that learns how you signal expertise will eventually produce expertise you never claimed: inflated credentials, invented statistics, implied experience that is not real. One fabricated credential in a professional post is a credibility event that no amount of polish elsewhere can repair.
So the call was architectural. Better to hold the output to what the user actually provided than to manufacture a more impressive version of a person who does not exist. Zero hallucination is not a setting you switch on. It was built into the architecture from the start.
How the Profile Deepens Over Time
The Voice Profile builds from the first samples you share, and early accuracy arrives fast, with the primary structural and tonal markers established in that first analysis.
After that it deepens through use. Every piece you generate, every edit you make, every refinement you apply teaches the system more about you. The patterns that hold across subjects and audiences come into sharper focus. The edge cases get mapped too, like how your voice shifts when the stakes rise, when the audience is unfamiliar, when the topic is sensitive.
Six months in, your Voice Profile is more precise than it was on day one. A year in, more precise still. The output grows more distinctly yours with use, not less.
Priya in Pune, Amara in Nairobi, and James in Lagos all have different Voice Profiles. Partly because their markets differ, and Kretell calibrates to each one natively. But mostly because they are different people, with different histories, different communication patterns, and different voices that deserve to survive the trip from thinking to page.
The Mirror Moment Is an Architectural Outcome
When a professional reads their Kretell output and stops mid-sentence, not because anything is wrong but because it sounds exactly like them, that is not luck. It is the result of 99 markers mapped accurately, cultural register calibrated correctly, and zero fabrication introduced.
Generic tools are encyclopedias. Kretell is a mirror.
Encyclopedias are built to describe everyone. A mirror shows one specific person. The architecture is different, so the result is different.
Your voice has patterns. They are specific to you, and they are mappable. The only question is whether the tool you use was built to find them.
Try it. You already know something better had to exist.
Frequently Asked Questions
What is a Voice Profile?
A Voice Profile is a 99-marker map of how a specific person communicates. It captures sentence architecture, vocabulary patterns, attribution style, expertise signalling, and cultural register. It is built from analysis of your existing writing rather than from preference questionnaires, and it is what keeps every Kretell output sounding like you specifically rather than like professional writing in general.
How does Kretell build a Voice Profile?
Kretell analyses the writing samples you provide across 99 distinct markers in 11 dimensions. The initial analysis runs on the first samples you share, and the profile deepens with every generation you produce and every edit you make. Edits matter most, since each correction teaches the system exactly how your patterns differ from what it generated.
How many writing samples does Kretell need to build a Voice Profile?
Seven to ten samples produce the most accurate initial profile. Two or three will build a working one, but accuracy improves a lot with variety. The most useful samples are not necessarily your best writing, they are your most authentic. Emails to respected peers and internal documents often beat polished LinkedIn posts as training material.
Why does AI-generated writing always sound like AI?
Most AI tools approximate tone. They match formality and general register well enough. The gap opens at the level of the small patterns that make one professional's writing distinct from another's: sentence architecture, attribution sequences, expertise signalling, cultural register. General-purpose AI was never built to hold individual specificity at that depth, and every session starts with no memory of you. The output sounds professional. It does not sound like you.
Why does Kretell use 99 markers rather than more?
Three markers were deliberately excluded from the architecture. All three relate to credential signalling, how a user represents their experience and qualifications. Collecting them would have introduced hallucination risk, because a tool that learns how you signal expertise will eventually produce expertise signals you never gave it. The decision was architectural. Accuracy at 99 markers with zero hallucination is the right standard.
How long does it take for the Voice Profile to become accurate?
Initial accuracy comes quickly, with the primary structural and tonal patterns emerging from the first analysis. The profile reaches the Mirror Moment threshold, where output is consistently recognisable as the specific user, at different points for different people. With strong initial samples and deliberate edits, most users get there within ten to thirty generations.
