Skip to main content

About Publication Intelligence

Helping readers find what matters

Publication Intelligence LLC builds software for analyzing books, creating indexes, and reviewing the evidence behind index entries. We help authors, editors, publishers, and professional indexers turn complex manuscripts into useful paths for readers.

From manuscript to a useful index

Document Analysis extracts and organizes PDF text, identifies page and content roles, maps printed page labels, and builds a document outline. You can review the analysis and export it as JSON.

Index Generator produces subject, author, and Source indexes through automated construction, reconciliation, editing, and source grounding. The workspace also provides tools for inspecting and adapting the result.

Index Editor supports saved indexes, and free Index Check finds mechanical issues. Index Audit assesses an index against its Manuscript; Index Score adds Benchmark-based assessment and reader testing. Documents shows current prerequisites and paid availability. Semantic Index Repair is planned separately.

Index generation currently supports English- and Spanish-language books from final paginated PDFs with extractable text. OCR is not supported. Publication teams set the required scope, style, and delivery format.

Explore the products

Quality you can examine

A useful index helps readers find substantive discussion, understand relationships, and navigate the book. Our pipeline performs that intellectual and editorial work automatically, then connects entries to source evidence you can inspect.

Reaching this level took seven versions of the pipeline, extensive experimentation, and many refinements to the current version. Its many computationally expensive, time-consuming stages construct, reconcile, edit, and ground the index across the book. This is an engineered indexing system, not an out-of-the-box language-model result.

In our source-grounded comparison for The Oxford History of the French Revolution, our index scored 98.05 out of 100 as delivered, without subsequent human editing. The professionally created published index scored 88.43. We publish the comparison, indexing standard, and evaluation methodology so you can examine the result and how it was assessed.

Origin

Three wishes became Publication Intelligence

In 2019, John Camden began building Djinndex for a complex 1,104-page reference work. It accelerated author and Scripture indexing, but subject indexing remained out of reach. Modern language models opened that possibility. Seven pipeline versions and extensive experimentation turned it into a system whose unedited output outscored the professionally created published index in our comparison.

Read the full origin story in The Impossible Third Index: How AI Changed Publication Intelligence.

Built for the people behind the book

Publication Intelligence LLC develops and operates the service. Our work grows from John Camden’s experience building indexing tools and a practical goal: help publication teams produce indexes that serve their readers.

Questions about the product or your workflow? Contact us.

See pricing

Start with your book

Explore the products and pricing, then create an account to analyze your document or build an index.

Create an account