How it works Note types Philosophy Get started
Developmental editing software

Your book,
read the way

you meant it.

The Press is developmental editing software for fiction writers — and narrative non-fiction. It reads your whole manuscript, tracks your characters and structure across every chapter, and builds a model of your book that gets smarter every session.

No credit card required Your manuscript stays yours Never used for training
thepress.app · Book Editor & Writer
+ New Note
The Quiet Hours Studio
Chapter Plot Character Timeline
Chapter 12 — The Return
The station was empty at that hour, as it always…
Character — Mira Solt
Traits, arc, voice profile established…
Chapter 12 — The Return
B
I
H1
H2
Quote
Link
Table
Chapter intent: the reader should feel her decision before she makes it.
Editorial Notes
All Open
Character High
"Mira's decision here contradicts her established aversion to public confrontation from Chapter 3. Is this deliberate?"
Agree Disagree Context
Pacing Mid
"The passage from line 8 sustains low tension for ~400 words. Intentional calm before the confrontation?"
Plot Character Pacing Style Theme
Analyse this note
Manuscript Mode
You have a draft
Upload one or two chapters to start. The Press builds its understanding of the book as you go — moving backward and forward through the manuscript simultaneously.
Chapter by Chapter
You're mid-draft
Upload chapters as you complete them. The Press tracks context progressively — confidence levels adjust to what's available, open questions are carried forward.
Studio Mode
You're starting from zero
A structured intake conversation builds your book's architecture — characters, plot, timeline — before chapter one is written. The Press is present from the first sentence.
Why writers need this

You know something is wrong.
You just can't see what.

"Something isn't working."
The most common thing writers feel at the end of a draft. The structure feels off. A character isn't behaving right. The second act loses momentum. But when you're this close to your own work, it's almost impossible to see exactly why.
0
Existing tools that understand the whole book, learn your intentions, and get smarter over time. Until now.
1

Most tools read one chapter at a time

They have no memory of what came before. They cannot see a promise made in Chapter 3 that goes unresolved by Chapter 14. The Press reads the whole book before it touches any part.

2

Generic feedback treats every voice the same

Most tools flag departures from convention — but your departures might be exactly what makes your writing distinctive. The Press evaluates against your intentions, not a universal standard.

3

Nothing remembers your decisions

Every session starts from zero. You re-explain the book. The feedback contradicts what it said last time. Nothing compounds. The Press builds a model of your book that grows with every session.

The Press is built differently

Whole-book context assembly. Voice protection. A personal book model built from your responses — session by session, decision by decision. It gets smarter the longer you use it.

Early access

Start with the book
you're already writing.

The Press is in early access. Join the list and be among the first writers to bring it into your process — from the first chapter or the last.

How it works

Three ways in.
Seven layers of analysis.

Whether you're uploading a finished draft, working chapter by chapter, or writing from the first sentence inside The Press — the same seven-layer analysis engine runs underneath.

01 / 07

Context Assembly

Before The Press reads a single word of your chapter, it assembles all supporting notes as context — plot, characters, timeline, style. Feedback is never based on the chapter in isolation. Always on the whole book.

Plot note Character notes Timeline Style note
Chapter 12 — The Return Primary
assembled before analysis runs
Plot — act structure, dramatic question Context
Mira Solt — arc, voice profile Context
Timeline — story week 6 Context
Style — 4 voice protections active Context
Session history — 8 sessions Context
Analysis runs with full book context
02 / 07

The feedback loop

Every observation you respond to teaches the system about your book. Agree confirms the pattern. Disagree locks it out permanently. Add context stores your intention — future chapters are evaluated against it.

Agree Disagree + reason Add context Ignore
Character High confidence
"Mira's behaviour contradicts her established aversion to confrontation. Deliberate?"
Agree Disagree Context
Writer: "Yes — this is the moment her arc shifts. It should feel wrong on purpose."
Stored as confirmed intention. Chapter 14 flagged to verify the arc resolves.
Session 9 · Personal book model updated · 34 confirmed intentions
03 / 07

Six focus areas

The analysis never asks an open question. Each focus area has a precise set of structural questions it asks — nothing more. You choose which areas to activate per session. The feedback tone is yours to set: questioning, direct, or harsh.

Plot Character Pacing Style Timeline Theme
Plot — chapter function? Setups without payoffs?
Character — trait contradictions? Voice drift?
Pacing — attention risk? Scenes too long?
Style — voice drift? Sentences doing no work?
Timeline — chronological contradictions?
Theme — present, absent, or contradicted?
04 / 07

Voice protection

List any stylistic choices that are intentional. Sentence fragments. Unconventional punctuation. Tense shifts in flashbacks. The Press locks these out permanently — they will never be flagged as problems, in any session, for any chapter.

Style note Stored permanently Never overridden
Voice protection — active for this book
Sentence fragments in narration are intentional
Present tense shifts in flashbacks are intentional
Mira speaks in run-on sentences — do not flag
Unconventional dialogue punctuation is intentional
4 protected choices · Applied to all sessions and all chapters
05 / 07

Chapter intent

Before analysis, state what this chapter is trying to do. A slow chapter is evaluated against its stated purpose, not flagged as drag. The system evaluates: did this chapter achieve what the writer said it would?

Set per chapter Modifies the analysis
Chapter 7 — Intent
"This chapter is deliberately slow — the calm before the third act. The reader should feel tension building underneath the quiet surface."
Pacing
Low tension evaluated against stated intent. Flagged as deliberate, not drag. Observation confidence adjusted.
Intent confirmed. Chapter passes pacing review.
06 / 07

The personal book model

Built entirely from your sessions on this one book. Never shared. Never used to train anything. Every confirmed intention, every disagreement, every voice protection item accumulates — and is applied to every future session.

Private Never shared Compounds over time
Personal Book Model — The Quiet Hours
34
Confirmed intentions
12
Stored exceptions
4
Voice protections
9
Sessions completed
Scoped to this book only · Deleted if you delete the book · Never leaves your account
07 / 07

Studio Mode

Starting from zero? A structured intake conversation builds your book's architecture — characters, plot, timeline — before chapter one is written. The Press is present from the first sentence. Problems caught at chapter 2, not chapter 20.

Studio Mode Architecture first From sentence one
"Who changes in this story, and what forces that change?"
"Mira. She loses the ability to lie to herself after her brother disappears."
"What must the reader believe for the ending to feel earned?"
"That she always had the courage — she just did not know where to find it."
Character note for Mira built. Plot note drafted. Ready to begin Chapter 1.
The workspace

Eight note types.
One working system.

Each structural element of the book lives in its own note. The Press reads the relevant notes as context before analysing any chapter. Feedback is always informed by the whole, not just the part.

Chapter
The manuscript text itself. The primary subject of every analysis session. Has the chapter intent bar for stating what the note is trying to do.
Feeds all six focus areas
Plot
Story map, act structure, the central dramatic question, setup and payoff tracking. What the book is trying to do structurally.
Read by: Plot, Pacing, Theme focus areas
Character
Traits, arc, voice profile, relationships, chapter appearances. One note per significant character. The editor reads this before evaluating any scene containing that character.
Read by: Character, Style focus areas
Timeline
Chronological sequence in story-time, date anchors, time gaps between chapters, parallel timelines. Catches contradictions your eye skips.
Read by: Timeline focus area
Pacing
Chapter-by-chapter tension map, scene lengths, drag and rush points, reader attention risks. The rhythm of the whole book made visible.
Read by: Pacing focus area
Theme
Central theme, thematic questions the story raises, motifs and symbols, how theme manifests per act.
Read by: Theme focus area
Style
Voice profile, sentence rhythm, protected choices that must never be flagged. Your voice protection items live here.
Read by: Style focus area + every session
Session
Auto-generated after every analysis run. A timestamped record of what was analysed, what was observed, and what you decided. The full history of your editorial process.
Auto-generated · never manually edited
The founding principle

We know what the model is
biased toward. We say so.

The model has read essentially all published anglophone fiction. It has a strong implicit bias toward the conventions that got through agents, editors, and publishers. Cormac McCarthy. Woolf. Saramago. None would pass a generic AI analysis unflagged. We built the anti-bias mechanisms directly into the system.

!
What the model is biased toward

Three-act structure. Psychologically consistent characters with legible motivations. Pacing that matches genre expectations. Prose that doesn't draw attention to itself. That is not good writing — that is commercially viable writing as filtered through the 20th and 21st century publishing industry. The Press acknowledges this bias explicitly. It is named, not hidden.

What The Press analyses instead

Structure, consistency, and stated intention — not quality. Voice protection locks out anything you mark as intentional. The feedback loop actively corrects bias over time. Every observation is a question, not a verdict. The writer remains in complete control at every stage. Quality is your job. The Press makes sure you can see what's actually on the page, not what you think is there.