Put difficult records to work.
We turn poor OCR, handwritten ledgers, phone photos, PDFs, and database exports into clean records people can search, connect, and cite.
Each extracted fact keeps a link to the page or record it came from. Your institution keeps custody of the originals.
Digitized does not always mean usable.
A folder of TIFFs is not a database. OCR is often incomplete. Names, places, dates, and events remain buried in page images and disconnected files. We do the work between scanning a record and making its contents useful.
Existing digital collections
Scanned pages, TIFF collections, PDFs, weak OCR, XML, CSV, and legacy database exports.
Records that still need work
Phone photos, handwriting, ledgers, directories, newspapers, and mixed-format collections.
What happens to a collection.
Records ready to use
Processed text, document sections, normalized fields, people, events, claims, relationships, citations, and data for search or export.
The evidence
A researcher can follow an extracted claim back to the passage, page, document, and collection it came from.
You keep the originals.
We can work in an institution-controlled environment, a hosted environment, or an air-gapped system. The collection agreement sets the access, storage, and reuse rules.
Add verification when the records are ready.
Stable IDs, signed manifests, and processing logs let another system check a record against an earlier state. Corrections are recorded as new events, so the earlier state remains available.
Give staff and researchers better access to records while keeping local control.
Move from a search result to the document and page behind it.
Support property, government, genealogy, and other research under the collection agreement.
Once the public verifier is released, institutions will be able to check records without relying on a private Kettle database.
Have a collection that is hard to use?
Send a representative sample and tell us what you need from it. We will review the formats, condition, access rules, and likely processing steps.