Patrivox

Patrivox turns collections of PDF archives into searchable material. Its French product page describes OCR, automatic identification of people and places, and links between related documents. It is presented for municipal archives, history societies, heritage libraries, and church archives that want researchers or the public to explore a collection.

Import and index an archive

You can upload PDFs individually or in batches. Patrivox says Mistral AI reads the pages and extracts names, locations, and dates, enriching document metadata. The integrated viewer supports zooming, navigation, and copying OCR text.

The useful first check is a representative sample from the archive. Inspect the extracted text and entities against the original pages, particularly if scans contain unusual typography or names. The captured page promotes rapid processing but supplies no accuracy benchmark for your collection's documents.

Search, ask questions, and follow connections

Patrivox offers full-text search with tolerance for spelling errors, semantic search, and natural-language questions with source citations. Filters include dates, document types, and tags. You can use the cited pages to inspect the context of an answer instead of relying only on the generated response.

Its knowledge graph links people, places, and organizations mentioned in the documents. Browsing connected entities gives you another route into the collection when you do not know which file contains a relevant reference. Shared access supports researchers, a team, or public readers. All displayed plans include unlimited readers, with separate limits on administrators.

Budget pages, queries, and storage separately

The free Découverte plan includes 100 lifetime pages, 200MB of storage, 50 AI queries per month, and one administrator. Starter is listed at €29 monthly before tax with 1,000 pages per month, 2GB of storage, and 150 AI queries. Larger plans and add-on packs increase those allowances.

The feature comparison explains that paid page quotas reset on the billing date and unused pages do not carry over. Deleting a document frees storage but does not restore consumed OCR pages. That matters when estimating a large import or repeatedly processing the same material.

The publisher states European hosting and advertises GDPR-related features. These are the publisher's statements; the captured pages do not provide an independent compliance audit. Confirm access settings, retention needs, and the relevant plan before opening an archive containing restricted material.