Search non-fiction by claim, not keyword, and compare how authors argue the same idea.
Findsight is a claim-based search engine for non-fiction books and articles. It lets researchers, writers, and students compare how different authors treat the same idea, navigate between related claims, and jump to original sources, using both basic keyword filters and AI-powered claim/question filters.
Findsight is a search engine built around claims rather than keywords. Instead of returning pages that merely mention a search term, it indexes the specific points authors make in non-fiction books and articles, then lets users compare how different sources treat the same idea. The site frames this approach as "syntopical reading" — moving between related claims across many works to build a broader picture of a topic than any single source provides. The tool is aimed at people who read and research non-fiction seriously: students, writers, researchers, and readers who want to cross-reference arguments rather than just find a definition or a citation. Anyone drafting an essay, preparing a lecture, or compiling notes who wants to see how multiple authors have approached the same claim is a reasonable fit. It is less useful for casual browsing or for content outside non-fiction, since that is the only material the site describes as indexed. In practice, Findsight combines basic and AI-powered filters. The "mention" filter is a literal text search, similar to a standard web search engine, and is intentionally broad. The "reference" and "discuss" filters narrow results using named entities and topic tags — some entities link out to Wikipedia, others are Findsight-specific categorized concepts — and support wildcard and OR syntax for more precise queries. Two AI-powered filters extend this further: "state" takes a custom claim and surfaces related claims elsewhere, useful when hunting for citations, while "answer" takes a question and returns claims that address it along with supporting context. A "Random" button and a "→" explore arrow let users jump into an unfamiliar claim and follow a thread of related ideas, with every result linking back to its original book or article. Findsight sits apart from general search engines and book-cataloguing tools by working at the level of individual claims rather than whole documents, and apart from generic AI chat tools by grounding answers in traceable sources rather than generated summaries. It is best understood as a research and syntopical-reading aid rather than a full-text library, note-taking app, or citation manager, and it is built and maintained by Summarity.
The homepage does not list pricing plans or paid tiers. Core search appears free to use, with a stated cap of 50 AI-powered (state/answer) searches per day; users can exceed this by removing AI filters and relying on the basic mention, reference, and discuss filters instead.
Learn moreThe homepage points users to an in-product "search guide" for help using filters and syntax; no other support channels, such as email or chat, are described.
Findsight links certain referenced entities directly to their Wikipedia entries for added context, but the homepage does not describe any third-party app integrations, plugins, or API access.
Claim-level search across non-fiction works, basic mention/reference/discuss filters, AI-powered state (related claims) and answer (question-answering) filters, entity and Wikipedia-linked suggestions, wildcard/OR query syntax, random claim discovery, and links to original source books and articles.
