Indie authors buy attention through review swaps, advance copy services and reader giveaways, and all of them can be gamed: people skim, copy a blurb, and post five stars. Paid to Read pays only for proven reading. After the book, a short AI conversation asks about specific plot points and the reader's own opinions; a pass releases the payment, and the review that follows is optional and never has to be positive. The author gets something no store gives them: where readers actually stopped, and what they really thought, in their words.
The data it runs on
The book: the author uploads the manuscript (EPUB or DOCX). It is chunked by chapter; a model drafts a question bank with expected facts per chapter, which the author approves.
Reader conversations: answers, time spent per chapter in the reading app, and optional reviews. Stored per book, shown to the author only in aggregate or with consent.
Limits and gaps: a determined cheater could read a detailed summary; very short books leave few questions; opinions cannot be "wrong", only vague.
Fallback: questions are drawn at random from a bank of 30 or more, mixing early and late chapters; answers are compared against public summaries and other readers' answers to spot copying; reading time that is impossibly fast skips payment and asks for a re-check later.
1. The mission
Give every independent author real readers and honest feedback, and give readers a fair reward for their time. Proof of reading replaces the fake review economy.
2. Why people can't resist it
Getting paid for a hobby: readers earn for something they already love, and the check feels like a fun book club chat, not a test.
Being heard: authors see readers argue about their characters. Readers see the author respond.
Collecting: reader levels, verified-reader badges and an early-copy queue for top readers.
3. How it works for real
Ingest: manuscript parsed to chapters; a frontier LLM (Claude via OpenRouter) writes a question bank: fact questions with expected answers and acceptable variants, plus opinion prompts tied to scenes. The author edits and approves.
Read: a simple reader with chapter timing. The book is watermarked per reader to discourage leaks.
Talk: a fast cheap model (Gemini Flash or DeepSeek) runs the conversation, checks fact answers semantically, asks a follow-up when an opinion is vague, and flags copied text. Borderline cases go to a frontier model.
Pay: the author's reader budget sits as USDC on Base in an escrow; the author tops up by card. Readers log in with email and get a smart wallet (Coinbase Smart Wallet, Privy or Dynamic); gas is sponsored by a paymaster, so crypto stays invisible. A pass triggers the payout.
Insights: an LLM clusters all conversations into themes, drop-off points and quotes, refreshed nightly.
4. Cost per use
About $0.05 per reader check. A 4 to 6 turn conversation on a fast model is under $0.01; a frontier second opinion on maybe 1 in 5 checks adds about $0.02; question bank generation is about $0.50 once per book. Transfers on Base cost a fraction of a cent, sponsored.
Free for readers.Authors: pay readers directly from their budget plus a 12% platform fee on each verified read. Free tier: up to 10 verified reads to try it. Insights dashboard included.
5. Growth loop
Readers share "Paid to read The Salt Clock" cards that make friends ask how. Authors share their honest reader insights, which brings more authors, and each new book brings its own readers. Top readers become a trusted early-copy panel that publishers want. Every Sunday the Paid Shelf drops three paying books matched to past book chat scores, and each proven read adds to a reading-proof record that raises the reader's pay tier, so the shelf is worth more every week a reader comes back.
6. Risks and guardrails
Store rules: paying for reviews breaks many store policies. Pay is for reading only, released before any review is asked, and reviews carry a disclosure. No payment ever depends on rating.
Cheating: random question draws, timing checks, copy detection, one account per verified phone.
Book leaks: per-reader watermarks, in-app reading only, no downloads.
Privacy: authors see aggregate insights and only quotes readers agreed to share.
7. MVP build order
Author uploads a book and approves an AI-drafted question bank.
Reader signup with email smart wallet and an in-app reader with chapter timing.
The book conversation with fact checks and vague-opinion follow-ups.
USDC escrow on Base with sponsored payouts, then the author dashboard.