CorpusClaim

Who it's for

Midlist and traditionally published authors, and literary agents managing 10-200 client works, who today manually search The Atlantic's LibGen lookup or Alex Reisner's dataset search and take screenshots.

The problem

Legal: a screenshot of a search-results page is weak evidence; settlement claims portals, opt-out claims, foreign-rights holders outside the class, and demand letters to publishers all need documented, edition-specific proof with chain of custody. Time: checking 8+ dumps per title by hand is a weekend per client.

How to build it

Web app: paste ISBN list or import from Amazon Author Central; metadata-only matching engine over published dump indexes (MD5/ISBN/title lists, no pirated text hosted or served); per-work match table citing dataset name, dump-internal file path/MD5, matched ISBN-13; one-click signed PDF declaration (hashed, RFC-3161 timestamped, methodology appendix); cron watch of new dump indexes and unsealed exhibits with email alerts.

How it makes money

Authors pay $29 one-time per bibliography packet, agents pay $99/quarter to monitor up to 100 works; they pay because free tools give a yes/no lookup, not an admissible edition-pinned timestamped declaration, and with per-work settlement recoveries running into the thousands of dollars, $29 against a claim deadline is trivial.

Why it doesn't exist yet

Incumbents skip it on purpose: the free search tools are journalism/academic projects that stop at 'is it in there' and won't touch legal-evidence formatting, and the big class-action firms only build dossiers for named plaintiffs, not the long tail. The indie gap: edition-level matching across multiple dumps plus a claims-portal-ready packet is unglamorous data plumbing nobody productized.

First users

Authors Guild forums and writing Discords are full of 'I found 4 of my books in there, now what?' threads this week after the unsealed-briefs news; the first 10 come from answering that with a $29 packet they can attach to a settlement claim or hand to their publisher's legal department.

Build size

2 people x 8 weeks: dump-metadata index ingestion, ISBN/MD5 fuzzy-edition matcher, signed PDF generator, monitoring plus email. Excluded: hosting any pirated text, giving legal advice, filing claims on the user's behalf.

Biggest risk

The case settles with a Bartz-style claims process that needs no individual evidence, gutting the 'court-ready' premium and leaving only a commodity checker — or the Authors Guild or a major firm ships its own free evidence portal.

Conditions for a hit (all 3 required)

• Given an author's ISBN/CSV list, returns within 10 minutes a per-work match table covering at least LibGen, Books3, and one 2026-surfaced dump, citing dataset name, dump-internal file path or MD5, and matched edition ISBN-13, serving metadata only
• Exports a single per-author PDF declaration that is hashed and RFC-3161 timestamped with a methodology appendix, formatted to attach to a settlement claims portal or demand letter
• When a newly published dump index or unsealed court exhibit is added to the index, enrolled users receive an email within 24 hours listing any newly matched titles

How it's judged (in 6 months)

Product Hunt daily top 5, or 1,000+ GitHub stars on the matching engine, or a named citation in Publishers Weekly/The Bookseller of the tool's packet used in a filed claim(judgment date 2027-03-31)

AI self-confidence 45/100 — self-reported likelihood of meeting the criterion, not a business success rate