Epoch Elements  ·  v10  ·  Windows Desktop

The writing tool
that tells the truth
about AI detection.

We don't help writers evade detection. We help them recover their own voice.

Most AI detectors flag stylistic conformity — not machine authorship. Coaching Stylist helps you understand the difference and write like yourself again.

Based on 8,500+ writing samples across 15 categories  ·  April 2026  ·  Epoch Elements

85%

Stephen Hawking · A Brief History of Time · 1988

"If a detector flags Stephen Hawking as 'likely AI,' what is it actually detecting?"

Hawking's masterwork scores 85% on AI-style alignment in our system — placing him in the top 5% of all 8,500+ samples analyzed, above most AI-generated text.

The detector found expository excellence. Clarity. Structure. Skill.

These are not flaws to be punished. They are skills to be taught.

Also in the study

Author / Text
Score
Distribution
Stephen Hawking
85%
Albert Einstein
82%
Charles Darwin
78%
The Bible (NT)
22%
The Quran
18%
Shakespeare · Hamlet
14%
Rumi · Selected Verses
10%
8,500+
Writing samples
analyzed in the study
15
Categories studied
across genres & languages
−60%
Maximum score reduction
via coaching interventions
+15%
Average score increase
from Grammarly polishing

What Coaching Stylist does

Not a detector.
A coach.

Three integrated systems. One platform. The only tool that analyzes the writer — not just the document.

📊

Genre-Aware Scoring

Scores are interpreted against genre expectations across 6 calibration profiles. A business brief and a personal essay aren't held to the same standard — because they shouldn't be.

🔬

Author Fingerprint System

Build a stable author fingerprint from verified samples. Compare new writing against that baseline to measure voice drift, z-scores, outlier markers, stable markers, and verification confidence.

📈

Style Evolution Tracking

Track how an author's stylistic register changes over time. Every sample is stored as a time-stamped snapshot. See normativity trends, dimension drift, and evolution insights on a live chart.

🎯

Personalized Coaching Plan

After every analysis, receive a ranked list of specific interventions — genre migration, lexical restraint, friction-first opening — each with projected score reductions.

📄

Stylistic Verification Report

Export a professionally branded PDF with voice alignment score, confidence level, outlier markers, stable markers, and a radar chart overlay comparing baseline to new sample.

🧬

Voice DNA

Paste 2–5 writing samples and extract a complete style profile — tone, sentence rhythm, vocabulary, punctuation habits, and a ready-to-copy Replication Prompt you can paste into any AI to make it write like you.

📦

Batch Analysis & Export

Analyze a full class set at once. Export individual reports, a summary CSV, or a complete ZIP bundle — PDF and DOCX for every student in one click.

Interactive Walkthrough

See every system
in action.

Click through all five analytical systems below — real output from real writing samples.

Single Analysis

Paste any writing sample and get an instant stylistic breakdown — AI-style percentage, confidence level, genre detection, and a full Stylistic Fingerprint radar chart across eight dimensions of your prose.

Single Analysis tab showing a writing sample with 45% AI-style score, radar chart, and full analysis report
🎯

AI-Style Score

A percentage estimate of how closely writing matches standardized, AI-associated stylistic patterns — not a verdict on authorship.

📡

Radar Chart

Eight dimensions visualized: Structure Control, Transition Density, Abstraction Level, Personal Presence, Reader Awareness, Narrative Anchoring, Rhythm Variation, Template Pressure.

🔍

Primary Indicators

The specific stylistic signals driving the score — what the writing is doing that reads as distinctively human or stylistically normed.

📄

Export Options

Export as TXT, JSON, DOCX, or PDF. Add any sample to a Baseline Profile with one click.

Research Finding

Stephen Hawking scores 85% AI-style. The same writer can score 3% in fiction and 79% in business writing. This tool measures stylistic normativity — not authorship.

1 of 5

Coaching Plan

Every analysis automatically generates a personalized coaching plan — ranked interventions with projected score impacts, your strongest stylistic asset, and an overall assessment. No extra steps required.

Coaching Plan tab showing personalized interventions, score zone, and strongest stylistic asset
📊

Score Zone

Your score placed in context — Low, Mixed, or High — with plain-English guidance on what that zone means for your writing goals.

Ranked Interventions

Specific, actionable changes ranked by projected impact — so you work on what matters most first.

💪

Strongest Asset

The one stylistic quality doing the most work for you — what to protect and lean into as you revise.

📑

Export

Export the full coaching plan as PDF or DOCX to share with students, clients, or keep as a revision reference.

Why This Matters

Research shows stylistic coaching reduces over-reliance on AI-normed patterns by up to 60% within three revision cycles — without changing the writer's core argument or ideas.

2 of 5

Baseline Profiles

Build a stable author fingerprint from 3–10 verified writing samples. Compare any new sample against that baseline to measure voice drift, signature consistency, and outlier markers. Exports a branded Stylistic Verification Report as PDF.

Baseline Profiles tab showing locked author profile, average fingerprint metrics, and verification result with 80.8% baseline similarity
🔒

Locked Profiles

Once enough samples are added, the profile locks — creating a stable fingerprint for comparing against any new writing.

📐

Average Fingerprint

Eight dimensions averaged across all baseline samples with standard deviation — showing what's stable versus variable in the author's voice.

🧬

Verification Result

Baseline Similarity %, Voice Drift by Dimension, Z-Scores, Confidence Band — full forensic-style output in plain English.

📋

Verification Report

Export a branded PDF with radar chart, drift analysis, and confidence level — ready to present to any audience.

Key Output

The Baseline Profile system produces a Confidence Band (Low / Moderate / High) alongside similarity scores — giving context that raw percentages alone cannot provide.

3 of 5

Style Evolution Tracking

Every sample added to a Baseline Profile is stored as a time-stamped snapshot. See a live two-panel chart of stylistic change over time — normativity trends and dimension drift — with a full Evolution Summary alongside.

Style Evolution tab showing two-panel chart tracking AI-style percentage and key dimensions over time with evolution summary
📈

Normativity Trend

The top chart tracks AI-style % across every sample — showing whether writing is becoming more or less stylistically normed over time.

🌊

Dimension Drift

The bottom chart shows how Structure Control, Personal Presence, Narrative Anchoring, and Rhythm Variation shift across samples.

Stable Markers

Identifies which dimensions remain consistent across all samples — the writer's true stylistic constants.

📉

Notable Drift

Flags dimensions that have shifted most significantly — positive or negative — since the first sample was recorded.

Research Finding

Style is not fixed — it is a trajectory. Style Evolution is the only tool that makes that trajectory visible, measurable, and exportable as a dated PDF report.

4 of 5

Voice DNA

Paste 2–5 writing samples and extract a complete style profile — Executive Summary, Detailed Style Rules, Signature Vocabulary, Punctuation Habits, and a ready-to-copy Replication Prompt you can paste into any AI to make it write like you.

Voice DNA Style Profile showing Executive Summary, Tone and Voice, Sentence Structure metrics, and signature vocabulary
Voice DNA Replication Prompt ready to copy into any AI to replicate the writer's exact voice and style
🧬

Executive Summary

A one-paragraph description of the writer's overall stylistic register — tone, rhythm, sentence mix, and signature vocabulary in plain English.

📏

Detailed Style Rules

Tone & Voice, Sentence Structure, Paragraph Density, Vocabulary — each broken down with specific measurable data from the samples.

🔑

Signature Vocabulary

The words that appear most frequently and distinctively in the writer's prose — the vocabulary fingerprint that makes their voice recognizable.

Replication Prompt

One click copies a complete prompt capturing voice, rhythm, vocabulary, punctuation, and sentence variety. Paste into any AI and it writes like you — not like generic AI.

The Core Use Case

Voice DNA answers the question every writer asks when using AI: "How do I make it sound like me?" The Replication Prompt is the answer — built from your actual writing, not a generic style description.

5 of 6

🧪 Stress Test My Writing

Paste any writing sample and generate rewrite instructions for all three gradient positions — More Normative, Hybrid, and More Voice-Driven — side by side. This demonstrates live that detection scores respond to stylistic construction, not authorship.

📐

More Normative

Instructions to restructure toward higher formality, clearer transitions, and impersonal register. Expected score: 60–90%.

Hybrid Version

Instructions to balance structure with voice — targeting the 35–55% hybrid zone where stylistic signals conflict.

🎙

More Voice-Driven

Instructions to introduce first-person presence, fragments, observation, and personal framing. Expected score: 5–30%.

📋

One-Click Copy

Each column has a Copy Instructions button — paste directly into ChatGPT, Claude, or any AI tool to generate the rewrite, then verify the score shift in Single Analysis.

The NGM Proof

The Normativity Gradient Model predicts: the same content will score approximately 60–90% normatively, 35–55% as a hybrid, and 5–30% with strong personal voice — regardless of who wrote it. The Stress Test lets you verify this live, on demand, with any text.

6 of 6

Five key findings

What the research
actually shows.

01

Genre is the strongest predictor — not origin.

The same writer produces scores from 0% to 79% depending on genre and intent. Personal narrative scores 0–20%. Academic writing scores 40–85%. A score tells you about register, not authorship.

02

The Polishing Paradox: editing raises your score.

Standard grammar correction raises AI-style scores by +3%. Full professional polish via Grammarly raises them by +15%. Students are being penalized for improving their writing.

03

Sacred texts score as "distinctively human."

The Bible, Quran, Torah, and Book of Mormon all score 18–24% — because they use narrative, poetry, and personal voice. Ancient texts pass modern AI detectors. Hawking does not.

04

The pattern holds across languages.

Literary works in Russian, Spanish, Chinese, Japanese, and Arabic score 8–20%. Scientific writing in French, German, and English scores 60–80%. Detectors measure register — not language, not origin.

05

Coaching reduces scores by up to 60%.

Targeted interventions — moving friction to the opening, lexical restraint, philosophical framing — consistently reduced scores without sacrificing clarity or purpose. Writers can learn to control their register.

06

Style is not fixed — it is a trajectory.

Every writing sample tells you where a writer is. A series of samples tells you where they are going. Style Evolution Tracking makes that trajectory visible — normativity trends, dimension drift, and stylistic shifts over time. No other tool does this.

07

Coaching drops scores — without changing authorship.

In repeated testing, AI-style scores dropped dramatically after coaching interventions that increased autobiographical specificity and narrative unpredictability. One reflective essay dropped from 68% to 25% after revisions that preserved the original ideas while introducing personal memory. A doctrinal essay dropped from 82% to 55% after replacing formulaic summary language with lived experience. Same writer. Same ideas. Different voice.

08

Every writer has a Voice DNA — and it can be extracted.

Paste 2–5 writing samples and Coaching Stylist produces a complete linguistic portrait: tone, sentence rhythm, vocabulary signatures, punctuation habits, and a Replication Prompt — a ready-to-copy instruction block that makes any AI write in your specific voice. Your style is not a mystery. It is a pattern. And patterns can be named.

Pricing

One purchase.
Yours forever.

No subscription. No cloud. Your data stays on your machine. The white paper is included at every tier.

Student

$19

one-time · .edu email required

  • Full Coaching Stylist app
  • All analysis modes
  • PDF & DOCX export
  • White paper included
  • Single device

Verified via .edu email at checkout

Buy Student — $19

Classroom

$99

one-time · 1 educator license

  • Full Coaching Stylist app
  • All analysis modes
  • PDF & DOCX export
  • White paper included
  • Home + school install

Single teacher or professor

Buy Classroom — $99

Department

$299

one-time · up to 10 seats

  • Full Coaching Stylist app
  • All analysis modes
  • PDF & DOCX export
  • White paper included
  • 10 seat license

Academic departments & teams

Buy Department — $299

Enterprise

$799

one-time · unlimited seats

  • Full Coaching Stylist app
  • All analysis modes
  • PDF & DOCX export
  • White paper included
  • Unlimited seats

Districts & large institutions

Buy Enterprise — $799

All purchases include the full research white paper — Stylistic Normativity, Not Machine Origin.
Questions? inquiries@coachingstylist.com

Included with every purchase

The white paper that
started all of this.

Stylistic Normativity, Not Machine Origin — a research paper synthesizing findings from 8,500+ writing samples across 15 categories. It explains what AI detectors actually measure, why the polishing paradox exists, and why coaching recovers voice rather than evades detection. New revision sensitivity findings show scores dropping from 68% to 25% — same writer, same ideas, recovered voice. Included free with every purchase.

Get the App + White Paper