AI-Tell Cleaner for Romance Authors
Some habits make a page read as though a machine wrote it, and readers are getting quicker at noticing. The AI-Tell Cleaner reads your whole manuscript and marks what it finds as comments in your own Word file: the passage, why it reads that way, and what you might do about it. Nothing is changed behind your back.
Run AI-Tell Cleaner →What AI-Tell Cleaner does
It shows you each tell, explains it, and lets you decide. Every flag is a margin comment you accept or reject in Word — the tool does the finding and the explaining, and the call on each line stays yours.
It reads by rate, not by single instances. An em-dash, a short fragment, a filter word — those are ordinary craft, and often the most human thing on the page. What it measures is how often the patterns pile up across the whole book, and the report says so plainly.
It won't catch everything, and it doesn't rewrite your book for you — the Deep Edit drafts suggestions, it doesn't apply them. Your own read-through still decides what stays. What you get is the map, so the decisions are yours to make quickly.
Two passes — you pick which one you're running
One tool, two run modes, chosen each time you send a book in. The Light Pass is the deterministic read. The Deep Edit is the same read, taken further.
Light Pass
- Flags every pattern it finds, in context, as a comment in your own Word file
- Says why each one reads machine-made, so you can judge it yourself
- Applies only the mechanical fixes that can't touch your voice — with a change log
- Density report: per-chapter heatmap, category breakdown, and the primer
Deep Edit
- Everything in the Light Pass
- Overused punctuation eased by model judgment, line by line, not by rule
- Drafted rewrite suggestions in your margins — each one screened so it can't introduce a tell
- A fuller report that says what it thinks is wrong, not just where
Light Pass vs. Deep Edit — which should I run?
The difference is how deep the read goes. Same detector, same pack, same manuscript back in your hands — the Deep Edit adds the model-judged layer on top of everything the Light Pass already does.
Light Pass
What it does: reads the whole manuscript by rate, not by single instances, and flags every tracked pattern as a comment in your own Word margins with the reason beside it. It applies only the mechanical fixes that can't touch your voice — every one of them logged — and ships the full pack: density report with a per-chapter heatmap, clean copy, change log and the primer.
Run it when you want the full map of what reads machine-made, and you'll make the calls line by line yourself.
Deep Edit
What it does: everything in the Light Pass, plus a model judges the ambiguous punctuation the rules won't touch — eased only where the line reads identically. It drafts rewrite suggestions into your margins, each one screened by our own detector so a suggestion can't introduce a tell, reads every chapter for the structural patterns surface-editing can't reach, and writes the blunter report.
Run it when you want drafted alternatives waiting in the margins and the structural read, not just the map.
Both passes end the same way. Every change and every suggestion is yours to accept or reject in Word — neither pass rewrites your book behind your back.
What actually gets changed — and what waits for you
Applied to your clean copy (both passes): the mechanical corrections that can't touch your voice — quote and ellipsis consistency, spacing, and em-dash easing where the sentence reads word-for-word identically without it. The Deep Edit goes further here: a model judges every ambiguous dash the fixed rules decline, easing only the ones that pass the same word-for-word test. On our own test book that came to 128 applied corrections on the Deep Edit against 25 on the Light Pass — every one of them in the change log, with the before and the after.
Drafted into your margins — never applied: anything that would change your words. Clichés, negation-flips and echoed phrases arrive as Word comments with the rewrite already drafted (“Suggested rewrite: Something short of a smile.”) — accepting one is a single click in Word, but the click is yours. Structural observations, like a theme said outright or an arc that resolves too neatly, are flagged and explained, never touched.
The hard rule both passes share: nothing that changes a word of your prose happens without your hand on it.
How the density reading works
The score in your report is our own descriptive measure of how densely these patterns run through your manuscript. It is not a probability that anything was machine-written, and it is not a reading taken from anyone else's detector — no third-party scores appear anywhere in the pack.
Because it reads by rate across the whole book, a lone em-dash or a single sentence fragment never counts against you. The heatmap shows you where the patterns cluster, chapter by chapter, and the category breakdown shows you which habits are doing the clustering.
Some findings are marked safe to fix mechanically and are already applied in your clean copy. The rest are marked as your judgment call — flagged, explained, and left for you.
What you get
- A commented manuscript — every tell marked in your own .docx, with the reason beside it, to accept or reject in Word
- A density report (PDF) — your overall density reading, a chapter-by-chapter heatmap and a category breakdown
- A "Spot Your Own Tells" primer — one page on catching these yourself next time
- A clean copy with the mechanical fixes already applied, no markup
- A change log of exactly what was changed, so nothing happens off the page
Pricing
FAQ
Does it rewrite my book?
No. Every tell is marked as a comment in your own Word file with the reason beside it, and you accept or reject each one. The Deep Edit drafts rewrite suggestions in the margin — it doesn't apply them.
How long a manuscript can it take?
Up to 150,000 words, both passes. The engine scales its reading to the size of the book, so a long manuscript gets the same coverage as a short one.
What is the density score?
It's our own descriptive measure of how densely these patterns run through your manuscript. It isn't a probability, and it isn't a reading from any third-party detector — no outside scores appear in the pack.
Will it strip out all my em-dashes?
No. Em-dashes, short fragments and filter words are ordinary craft, and often the most human thing on the page. The tool reads by rate across the whole book, so single instances aren't treated as problems.
Does it catch everything?
No, and the report says so. It gives you a map of what it found and why it flagged it, so your own read-through can spend its time where it matters most.
What's the difference between the two passes?
The Light Pass is the deterministic read: it flags, explains, and applies only the mechanical fixes that can't touch your voice. The Deep Edit adds model-judged easing of overused punctuation, drafted rewrite suggestions in your margins, and a fuller report.
Do I need Microsoft Word?
The commented manuscript is a .docx, so anything that reads Word comments will open it. The density report and the primer come as PDFs, and the clean copy is a plain .docx with no markup.