How the AI works

What the AI read looks like.
And how the score is built.

Every assessment is generated by AI — no human reads your manuscript. Below is a sample verdict on a book you've heard of, plus the rubric the model uses on every submission. No black box.

Sample AI verdict · what the model says about a book you already know

Gone Girl by Gillian Flynn
Adult psychological thriller · Submitted as an unsolicited query, opening pages only
Auction Material

Overall

94/100

Pitch to editors

"A husband becomes the prime suspect in his wife's disappearance — and the diary she leaves behind is the most damning evidence of all."

The voice on page one does the entire job — Nick's casual menace, the wife's diary voice as counterpoint. That's the rare hook that doesn't need a synopsis to land. Comp-wise this is its own thing, which is both a selling point and a positioning problem; I'd be calling editors who buy literary-leaning thrillers (Knopf, Crown) rather than category mystery houses. This is auction territory — I'd want it out to a tight list of editors fast, with a floor.
Craft · 93 (voice 97 · character 92 · story 93 · hook 94 · propulsion 90)
Marketability · 95 (market timing 94)

Illustrative only · for calibration

How we score · the rubric is public

Craft60%

The writing itself, judged on the sample alone. A weighted blend of voice (~30%), character (~20%), story (~20%), hook (~15%), and propulsion (~15%). Voice is the biggest lever — it's how acquiring editors actually read pages.

Marketability40%

How a 2026 acquiring editor would respond to these pages: hook clarity, shelf legibility, commercial viability, and market timing. Positioning (logline, comps, genre, synopsis, query, bio) can lift or drag this — but never the craft score.

Verdict bands

Pass1–54
Reluctant Pass55–69
Requesting Full / Revise & Resubmit70–84
Offering Representation85–91
Auction Material92–95
Preempt Offer96–98
Seven-Figure Debut99–100

Scored like the inbox. Real agents pass on 97–99% of queries — most submissions land in Pass or Reluctant Pass. Requesting Full is meaningfully rare, and anything above is the once-a-quarter or once-a-year submission. The pages carry roughly 60–70% of the verdict; positioning fields adjust marketability within bounded limits and never touch craft. Query letters get a separate effectiveness score.

Everything included · one assessment unlocks the full desk

What you can do once you're in.

Full manuscript assessment

The core read. Verdict (Pass, Revise & Resubmit, Requesting Partial, Requesting Full), overall score out of 100, pitch line, strengths, red flags, and an agent's note — scored against the public rubric above.

First-page read

A free, fast gut check on your opening page. Same agent voice, scoped to the part most queries die on.

Draft a query letter

The agent writes a submission-ready query from your logline, synopsis, and sample pages — formatted the way agents actually want to receive it.

Critique your query

Paste an existing draft. Get a line-by-line critique plus a clean rewrite you can send.

The 5-page test

How your opening reads to the part of an agent's brain that decides in 90 seconds. Where attention drops, where it holds.

Targeted agent list

Real working agents with fit notes and a recommended query order. Always verify on QueryTracker before sending.

Synopsis builder

Industry-format 1-page and 2-page synopses — present tense, names in CAPS, ending included. Grounded in your bullets and pages; no invented plot.

Prose vs. published shelf

How your sentences read next to currently-published novels in your genre. Honest, not flattering.

Ruthless editor

What to tighten, what to cut. Pacing first — substance protected.

Editor vs. Editor

Two veteran editors with opposing instincts argue over your submission until they reach a joint verdict. Watch the case for and against your book get made.

Ask the agent (chat)

Follow-up questions on your read. Push back, ask what to revise, ask what would change the verdict.

Export submission packet

One click opens a print-ready packet — Times New Roman, double-spaced, standard format. Save as PDF from your print dialog and send.

One purchase, one manuscript, everything above. No per-tool fees, no upsell. Use the agent's tools as many times as you need while you revise — the same submission stays in your account so you can iterate query, synopsis, and pages without paying again.

Ready for your read?

Submit a logline and first page, or the full packet.

An AI simulation. Not a substitute for actual representation — but closer to the real thing than most beta readers.