AI Editorial Analysis

The Red Pen

An AI-powered manuscript editor that shows you where your writing is working and where it isn't — without touching a single word of it.

Claude
·
ChatGPT
·
Gemini
Open the Editor → See it in action ↓
A different kind of AI tool

AI that edits.
Not one that writes.

Most AI tools do the writing for you. The Red Pen does something different — it reads what you've already written and tells you, honestly and precisely, where your draft is working and where it isn't. Your words stay exactly as you wrote them. The insight is ours.

The writing is entirely your own.
The AI is just the sharpest editor you've ever had.

✎
Your words, unchanged
The AI never rewrites your sentences, swaps your vocabulary, or inserts its own voice. It reads your manuscript the way a human editor would — then marks what it finds, directly on the text, leaving the revision entirely to you.
◈
Feedback, not substitution
Every annotation is a suggestion, not an alteration. Where to cut, what to fix, what's already working. You decide what to act on — and when you do, the revision is yours to write in your own voice.
↗
Clearer eyes, same voice
There's a real difference between AI that produces text and AI that helps you improve yours. The Red Pen exists to show you the weaknesses in your writing so that you can fix them — not to replace your voice with something generated.
Interactive preview

Click any highlight to read the note

Real editorial output on a sample passage — strengths, cuts, fixes, and audience notes placed directly on the text

the-red-pen.app / results
the_pier.txt
7 annotations · Claude Sonnet 4.6
Strength
Cut
Fix
Audience
Weakness
Strengths
Churned endlessly
Alarm she couldn't silence
Cuts
Looking out at nothing
Very difficult situation
Fixes
Wherever it wanted to go
Audience
Cold, relentless hands
Weaknesses
Marina lights

Sarah stood at the edge of the weathered pier, looking out at nothing in particularCUT. Below, the ocean churned endlesslySTR. She thought about what Marcus had said to her three days ago — the words still ringing in her ears like an alarm she couldn't silenceSTR. It was a very difficult situationCUT. The wind pressed against her with cold, relentless handsAUD, and for a moment she thought she might let it push her wherever it wanted to goFIX.

The lights of the marina glittered across the waterWKS. Somewhere behind her, a car door slammed. She didn't turn around. She had been standing here for — how long? Twenty minutes? An hour? Time had gone slack in a way it only did when she was this tired, this spent, this close to simply not caring what happened next.

← Click a highlighted passage to read the editorial note
● Strength
Verb choice carries emotional weight
"Churned" is exactly right — it conveys both physical motion and emotional turbulence simultaneously. The word works on two levels without announcing itself, which is the mark of controlled prose.
"the ocean churned endlessly"
Recommendation
Keep as-is. Consider whether "endlessly" earns its place, or whether the verb alone is already doing enough work.
● Strength
Extended simile earns its length
Most extended similes overstay their welcome, but this one doesn't. The specificity of "alarm she couldn't silence" feels lived-in rather than literary. It's a strong emotional anchor for the whole scene.
"the words still ringing in her ears like an alarm she couldn't silence"
Recommendation
Leave it. This is the emotional core of the paragraph — don't tighten it.
✂ Cut
Redundant qualifier undercuts the opening
"At the edge of the weathered pier" already implies she's staring out. Adding "looking out at nothing in particular" dilutes the physical specificity of the image. The vagueness undermines what the pier detail establishes.
"looking out at nothing in particular"
Recommendation
Cut the entire phrase. End the sentence after "pier" — the period does more work than the qualifier.
✂ Cut
Telling in the middle of showing
This sentence stops the scene cold. Everything around it — the ocean, the memory, the wind — is doing the emotional heavy lifting. Naming the feeling the reader is already experiencing deflates what the surrounding prose has built.
"It was a very difficult situation."
Recommendation
Delete entirely. The paragraph is stronger without it — and the rhythm improves when you move directly from the memory to the wind.
◈ Fix
Ambiguous — clarify the character's intent
This line reads as potential suicidal ideation. That may be intentional — but if it is, the scene hasn't built toward it, and the casual phrasing ("wherever it wanted to go") undersells the gravity. If unintentional, it needs revision before readers run with it.
"she thought she might let it push her wherever it wanted to go"
Recommendation
Clarify Sarah's mental state. Either lean into this moment deliberately, or reframe it as surrender / release — something that can't be misread.
◉ Audience
Personification activates the setting
Readers will feel the shift here — the wind stops being weather and becomes an agent. "Cold, relentless hands" is the moment the environment becomes complicit in Sarah's emotional state, which is what the whole paragraph has been building toward.
"cold, relentless hands"
Recommendation
This lands well. Make sure what follows matches the emotional intensity it establishes — don't let the next beat be flat.
◌ Weakness
Setting detail doesn't earn its place
This image is visually pleasant but emotionally inert. The marina lights don't connect to Sarah's state of mind — they appear and disappear without contributing to the scene's tension or atmosphere. A strong setting detail would do more than describe; it would reflect.
"The lights of the marina glittered across the water."
Recommendation
Connect the lights to Sarah's emotional state — make the glittering feel cold, indifferent, or beautiful-in-a-way-that-hurts. Right now they're decorative. They should be doing work.
Process

Four steps from draft to annotated manuscript

Upload a file, pick your editors, and receive a fully marked-up manuscript in under a minute.

1
Upload
Drop a .txt, .md, or .pdf file. The manuscript is read locally before any analysis begins.
2
Configure
Choose which AI editors to consult — Claude, ChatGPT, Gemini, or any combination in parallel.
3
Analyze
Each selected editor reads the full manuscript simultaneously and returns structured feedback.
4
Review
Browse inline highlights, click any annotation for the full note, and export when finished.
What's included

Every tool a developmental editor uses

Built around how human editors actually work — not how AI chatbots respond.

✎
Inline annotations
Highlights sit directly on the text. Every note is anchored to the exact passage it refers to — no more "see comment 14 in the appendix."
⊕
Three editors in parallel
Run Claude, ChatGPT, and Gemini simultaneously. Each model brings different editorial instincts — compare them side by side on the same manuscript.
↻
Second pass analysis
Request a follow-up read that skips what was already flagged and specifically hunts for what the first pass missed.
◈
Five annotation types
Strengths, cuts, fixes, audience notes, and weaknesses — each colour-coded throughout the manuscript so you see the shape of feedback at a glance.
↗
Scores & export
Voice, tension, pacing, prose craft, and characterization — scored 0–100 per editor. Export the fully annotated manuscript as a print-ready document.
◷
Analysis history
Every analysis is saved locally. Return to any past manuscript without re-uploading or re-running the analysis — including second pass results.
Voice
82
Tension
68
Prose craft
88
Pacing
71
Characterization
90
Overall
B−
Computed from the
average of all five
dimension scores
Editorial scoring

Five dimensions, one clear picture

After reading your manuscript, each AI editor scores five craft dimensions independently. Scores are concrete enough to track across drafts — a rising pacing score across three revisions tells you something a comment alone can't.

Scores above reflect the sample passage. A full manuscript produces scores across all five dimensions for each AI editor that ran.

AI editors

Three readers, three perspectives

Each model approaches a manuscript differently. Running them in parallel surfaces blind spots any single model would miss.

Claude
Sonnet 4.6 · Haiku 4.5
Strongest on nuance — identifies structural issues, character voice inconsistencies, and passages that work against the author's apparent intent. Most reliable with long manuscripts and complex POV work.
ChatGPT
GPT-4o
Thorough line-level reader — catches grammar and consistency issues that other models overlook. Particularly good at flagging prose that reads clearly to the writer but confusingly to a fresh reader.
Gemini
2.0 Flash · 2.5 Pro
Fast and dense — produces the most annotations per pass. Useful for broad first-pass coverage, then using Claude or GPT-4o to dig deeper into the issues it surfaces.
Ready to begin?

Bring your own API key.
Start editing in seconds.

Works with Claude, ChatGPT, or Gemini — your choice. Your manuscript never leaves your browser.

Open the Editor →