It reads, not writes
No blank-page takeover. No prompt theater. No trying to co-author the manuscript. The product begins after a draft exists and helps the writer see it more clearly.
Reading-first revision environment
AI that learns how to read your book.
Kaizen R/W is built for serious drafts. It surfaces the few passages worth your attention, learns what is intentional, and helps you revise without turning your manuscript into a dashboard.
- Fewer, better notes
- Author intent shapes every future pass
- Private by design, calm by default
Chapter 7
3 findings worth attention
Mara kept the lantern low as she crossed the ruined nave. The glass underfoot gave back almost no sound. At the altar she stopped, not because of the body, but because of the impossible neatness of it. Someone had arranged the dead boy's hands.
The first time she saw Daniel after the fire, he smiled in exactly the same way. It made her distrust the kindness before she trusted the grief.
Outside, bells were already starting. Not an alarm. A welcome.
Why this exists
Most AI writing products want to join the act of writing. KRW is built for the act of revision.
The market is full of drafting assistants, broad writing utilities, grammar layers, and story systems. Kaizen R/W takes a narrower stance: it reads like an attentive editor, not a co-author, and it only interrupts when a passage is worth attention.
Author intent wins
Teach the system why something is deliberate and that decision becomes part of how future passes interpret the book.
Flow matters as much as accuracy
A manuscript should feel like a book, not software. KRW keeps chrome light, opens one panel at a time, and avoids layout shifts that break reading rhythm.
Interactive demo
The core loop is simple: notice, decide, teach, continue.
This is the part most products miss. KRW is not just scoring text. It is learning how this manuscript should be read as the author dismisses, explains, and defers.
System mark
1 of 3 surfaced in this chapter
“She smiled the way people do when they have already forgiven themselves.”
KRW flags the passage because it sharpens character judgment while potentially flattening ambiguity. It is worth attention, but not necessarily wrong.
Future behavior
Teach action recorded
This becomes part of the manuscript's reading memory.
The author explains that this narrator often makes fast moral judgments that are later revised. KRW stores that preference and feeds it into later prompts so similar phrasing is less likely to be treated as accidental overreach.
- Dictionary entry added to prompt context
- Later scans become less noisy around intentional sharpness
- The writer keeps moving instead of litigating every sentence
What you actually get
A product designed for serious revision, not AI spectacle.
Everything here is chosen to support reading flow, author agency, and a calm path through complex drafts.
Focused chapter scans
AI chapter scanning is designed to surface a handful of findings instead of flooding the page.
Teachable feedback loop
Teach and dismiss actions feed future prompts, so the system adapts to the manuscript over time.
Keyboard-native revision
Tab, Enter, D, T, and F keep the workflow in motion for writers who want to stay in the text.
One panel at a time
Inline review expands in place without turning the manuscript into a maze of sidebars and drawers.
Full-manuscript pass
Cross-chapter scanning surfaces higher-level observations after the local reading pass is complete.
Density controls
Hide categories, control noise, and shape the editorial pass to fit the kind of revision you are doing.
Mobile-aware navigation
Touch TOC, swipe chapter navigation, and responsive panels extend the reading model to smaller screens.
Export and backup
Save your state, export reports, and keep control of the editorial conversation around a draft.
Who it is for
Best for fiction writers revising real drafts.
KRW is strongest in the stretch between draft completion and outside editorial feedback. It is built for the author who wants developmental-quality signals before pages leave their hands, and who cares about preserving voice as much as improving clarity.
- Draft 2 to 4 novels
- Indie authors doing self-edit passes
- Writers who want fewer, higher-signal interventions
- Authors who do not want AI to flatten the prose
Not another AI writing app
- Not a blank-page drafting copilot
- Not a grammar layer chasing every sentence
- Not a sprawling story bible platform
- Not a co-author trying to take over the page
KRW's lane is narrower and more defensible: a better first reader for revision.
Trust surfaces
Private by design. Legible by design. Built to earn confidence.
Writers do not just need capable AI. They need a product that feels safe, predictable, and easy to understand.
Local manuscript state
Core state lives in browser storage instead of depending on server-side persistence for basic use.
Bring your own model key
Writers can use their own provider key and keep control over model access inside the app settings.
Legible product behavior
The difference between noticing, teaching, dismissing, and deferring is clear in both UI and product logic.
Static-site simplicity
No build step for the frontend, no app-shell bloat, and a deployment model that stays understandable.
Under the hood
Simple enough to trust. Serious enough to scale the experience.
The product architecture reinforces the product philosophy: static frontend, AI proxy worker, browser-first persistence, and streaming feedback that keeps revision moving.
Frontend
Static HTML, CSS, vanilla ES modules
AI proxy
Cloudflare Worker with streaming endpoints
Models
Gemini 2.5 Flash Lite and Claude Sonnet
State
localStorage, no server-side manuscript persistence
Deploy
Cloudflare Pages + Worker, GitHub-friendly static hosting
# Static site
npx wrangler pages deploy public/ --project-name=krw
# Worker
cd worker
npx wrangler deployThe result is a product that stays close to the metal: easy to host, easy to reason about, and intentionally constrained.
Product principles
What KRW optimizes for
FAQ
Questions serious writers actually ask
Is this trying to replace editors or beta readers?
No. KRW is designed to give immediate, teachable revision feedback before pages go to other humans. It is a first reader, not the final authority.
Does KRW write for me?
That is not the core job. The product is designed to read attentively, mark passages worth attention, and preserve author agency during revision.
How is this different from broad writing assistants?
Broad assistants are designed to be everywhere and do everything. KRW stays narrower: it focuses on revision, reading flow, teachable feedback, and a calmer manuscript experience.
Will it keep flagging choices I already explained?
The teach loop is built specifically to reduce that problem. Explain an intentional choice once and later prompts can incorporate that context.
Can I host this kind of site easily?
Yes. This marketing site is static HTML, CSS, and JavaScript, so it works well on GitHub Pages, Cloudflare Pages, Netlify, or similar static hosts.
Early access
Your draft does not need another AI trying to write it. It needs a better reader.
Kaizen R/W is built for authors who care about judgment, not just capability. Enter your email below and we'll send you an access link.
No account needed. No password. We send a link and your browser remembers you.