Surfio
AI is the new bookshelf

Readers ask AI what to read next. Does it ever name your book?

“Best cosy crime like Richard Osman?” “A good business book on habits?” — more readers now ask ChatGPT, Claude, Perplexity, Gemini and Amazon’s Rufus for their next read. Surfio asks all five the questions readers actually ask, and shows you whether AI recommends your book — or sends the reader to someone else’s.

Are you found across 5 AIs?The exact sources to get cited onPaste-ready Amazon / KDP listing rewriteStolen-by-AI check

5 AI engines · incl. an Amazon Rufus simulation · ~90 seconds · no login

See a real sample report →

The only book check that models Amazon’s Rufus shopping AI (a simulation) — where most book-buying decisions are heading.

5 engines

ChatGPT · Claude · Perplexity · Gemini · Amazon Rufus — tested in parallel

Reader prompts

The real questions readers ask AI for their next book — scored on whether you show up

2 scores

Does AI know your exact book — and does it own your genre?

Why this matters now

Book discovery is moving from search to recommendation.

Readers used to browse and search. Now they ask an assistant — “what should I read after this?” — and get three confident titles back. If your book isn’t one of them, the reader never discovers it. Surfio shows you exactly where you stand across the five engines readers use, and how to get recommended.

How it works

01

Tell us the book

Title, author, genre, and (optional) a couple of comparable titles. 30 seconds.

02

We ask the AIs

Reader questions across ChatGPT, Claude, Perplexity, Gemini & Amazon Rufus — direct, “best [genre]”, “books like…”, gift and use-case prompts.

03

You get the scorecard + fixes

Where your book ranks, which titles AI recommends instead, why — and the paste-ready changes to get recommended.

Title score

71

AI knows your book

Genre score

38

but loses the “best [genre]” lists

Recall82/100
Recommendation fit64/100
Genre authority35/100
List position28/100

What the audit checks

Five ways a reader reaches for AI.

  • Direct recall“Tell me about [your book].” Do the engines even know it exists?
  • “Best [genre]” listsThe #1 discovery surface — scored on whether your book makes the list.
  • “Books like…”When a reader asks for read-alikes to a comp title, does your book come up?
  • Reader intentGifts, beach reads, book-club picks, “something like X but lighter” — do you fit the brief?
  • Agentic buy“Buy me a good [genre] book” — your placement on the Rufus / shopping-agent path.

The part that actually moves the needle

AI recommends what it sees cited. We tell you exactly where to get cited.

Every “best [genre]” answer is stitched together from a handful of sources — Goodreads, Reddit threads, the “best of” lists AI trusts. Surfio finds the exactsources for your genre, shows which already mention you, and ranks the rest by how likely each is to feature you. It’s an outreach shortlist — not a vanity score. Get on these and AI starts recommending you.

PitchableAspirationalOut of reach
See the sources AI cites, by genre →

Most-cited sources · 215 live audits

01youtube.comOthercited 1856×
02goodreads.comGoodreadscited 1752×
03en.wikipedia.orgWikipediacited 2312×
04penguinrandomhouse.comPublishercited 722×
05fivebooks.comListiclecited 748×
06barnesandnoble.comRetailercited 398×

Pulled live from Surfio audits — and growing.

Also in your report

Did AI train on your book — without paying you?

Anthropic just settled for $1.5 billion; NVIDIA was ordered to pay $322 million— for training on pirated books. Surfio cross-references your title against the datasets the models admit using (LibGen, Books3, The Pile, Common Crawl…). If you’re in there, your report hands you a defense pack: a copyright-assertion letter, a DMCA template, and an Authors Guild class-action eligibility check.

See the stolen-book check on a real report →

What lands in your report

Not a number. A fix-it plan.

Dual Title + Genre score

Whether AI knows your book vs owns your genre — so you know which problem to fix.

4-axis breakdown

Recall, recommendation-fit, genre authority, list-position — your exact weak points.

“A reader asks AI…” replay

The real answers each engine gave, verbatim — see yourself recommended (or not).

Your outreach shortlist

The exact sources AI cites for your genre — Goodreads, Reddit, the “best [genre]” lists — ranked by how likely each is to feature you. Get on these and AI starts recommending you.

Who gets recommended instead

The competing titles AI names, and why — so you know who you're up against.

Paste-ready listing rewrite

Blurb, bullets, A+/About-the-Author, keywords — rewritten for how AI reads a book page.

Stolen-by-AI check + defense pack

Whether your book is in the AI training datasets (LibGen, Books3…), with copyright-assertion + DMCA letters if it is.

Ranked action plan + re-check

What to do first for the biggest lift, then re-run to prove the score moved.

What it’s worth

One missed reader costs more than this.

Doing this by hand isn’t cheap. Surfio does it across 5 engines in ~90 seconds — and the check is free.

Manual 5-engine check, incl. Rufus (analyst)£300–600
AI-visibility SaaS seat (Profound, Peec…)£99–499/mo
Verbatim AI answers + citation forensics£150–300
Comp-title / competing-author analysis£150–300
Paste-ready blurb + metadata rewrite£150–300
Prioritised action plan + re-check£200–400
Comparable total£950–2,400+
Surfio full report£29.99

Start free

See if AI recommends your book.

Free quick check across 2 engines — your score, weak spots and the titles beating you, in ~90 seconds. The full 5-engine report (incl. Amazon Rufus) + paste-ready fixes is the £29.99 report.

Quick start

Or have us do it for you.

Surfio is powered by publishing.co.uk— the UK team behind the KDP formatter and book-discovery tools used by thousands of authors. If the report shows AI is skipping your book, we’ll fix it for you: blurb, metadata, structured data, the lot.

Talk to publishing.co.uk →

Questions

Is this just asking ChatGPT once?

No. It's repeated reader prompts across five engines and several question types, aggregated into a deterministic score — not a one-shot vibe check.

What do I need to run it?

A book title, author, and genre. Optionally an ASIN/ISBN and a couple of comparable titles to sharpen the ‘books like…’ tests. No account.

How is the free check different from the £29.99 report?

Free gives you the scores, breakdown and which titles beat you. The £29.99 report adds the verbatim AI answers, citation sources, and the paste-ready blurb/metadata rewrite, across all five engines incl. Amazon Rufus.

Does it work for fiction and non-fiction — and magazines?

Yes — any book or print title readers might ask an AI about, including magazines and bookazines.