NearshelfFind my comps

Comp titles your query letter can stand on

Recent, mid-list books that readers of your book actually read. Picked from Qloo reader data, checked against the rules agents use, and shown with the evidence.

Find my compsSee how it works

Reader data from Qloo's taste graph

Nearshelf after a real search for books like Western Lane, The Candy House and Memphis, with six comps on the shelf including The Fraud by Zadie Smith

How Nearshelf picks a comp

Three steps, each shown on screen as it happens.

  1. 1

    Tell us three books yours reads like

    Type titles or pick a preset. Each one is looked up in Qloo, and you see straight away whether it was found.

    Three seed books found in Qloo, each shown with its author and year
  2. 2

    Qloo finds the readers

    Nearshelf asks Qloo's taste graph which books the readers of your seeds also love, filtered to recent titles.

    The agent log showing the two tool calls the model chose, with their timings
  3. 3

    Every book is checked against the rules

    Recency, size, and whether it's one of your own books. Every verdict is made in code, not by a model.

    Check my comps results: Where the Crawdads Sing and Fourth Wing rejected with the rules they broke, Untethered Sky passes

What you leave with

Everything you need for the comps line in your query letter.

The Check my comps tab after checking three titles

Check the comps you already have. Paste up to five titles and see which pass, which don't, and a replacement for the ones that fail.

Check my comps

Comps you can defend. Recent, mid-list books with readers like yours, up to six on the shelf, in the order Qloo ranks them.

Evidence for The Fraud by Zadie Smith: Qloo entity ID, published 2023, popularity 0.926

Proof for every book. Its Qloo entity ID, publication year, popularity score and the seed books it came from.

Know why a book didn't make it. Rejected books stay visible with the rule they broke and the value that failed it.

A comp line built from accepted comps, ready to copy

A line ready for your query. "My novel will appeal to readers of A and B", built only from comps that passed. Copy it in one click.

Why not just ask ChatGPT?

Ask a plain LLM for comps and check its answers. That's what the comparison tab does, live, with the same rules.

What a plain LLM did across 8 seed sets

  • Only 7 of 48 suggested titles passed every rule
  • 32 were too old and 14 were too big to cite
  • 11 were your own seed books and 10 had the wrong author

Same seed books, same rules, measured with npm run eval.

What Nearshelf can't do

  • Show a book Qloo doesn't know
  • Return one of your seed books as a comp
  • Let the model pick, add or rename a book
Try the comparison
A plain LLM's suggestions for the same books, each rejected with the rule it broke

Across 8 seed sets, measured with npm run eval

7 of 48

plain-LLM titles passed every rule

11

LLM titles were the author's own seed books

8 of 8

seed sets got three or more Nearshelf comps

Every comp carries its Qloo ID

Shown in the evidence panel for every book on the shelf.

Zero books come from the model

The agent picks tools. Books and verdicts come from Qloo and the rules.

Your search lives in the link

Share the link and it reruns the same search.

Find comps for your book in a minute

Three titles in, recent mid-list comps out, with the proof.

Find my compsCheck my comps