Digital SourcesAI-assisted archival digitisation project
Digital Sources

Archives closer to the reader.

An independent project by Łukasza Rzepińskiego.

The author began by using artificial intelligence for his own genealogical research. The good results inspired him to develop a broader pro publico bono project — for the public good.

Fonds being processed

We are currently processing two fonds of the Central State Historical Archives of Ukraine in Lviv (TsDIAL):

From scan to reading

The site brings together sources from many archives, Polish and foreign. Every document carries its full archival address: archive, fonds, inventory, file and reference code. We combine the register of files, transcriptions, Polish translations and historical commentary. The first archive being processed is TsDIAL in Lviv; further Polish and foreign archives will join as the work progresses.

How to read the editions

The materials are prepared with artificial intelligence. These are working editions and may contain reading, identification and translation errors. Each text states its status, scope and uncertainties. An available translation does not imply a complete critical edition or a completed bibliographical search.

Brackets and ellipses mark gaps and uncertain readings, among other things. The original document, later annotations and catalogue information should be considered separately. Scholarly use requires checking the scan and citing the archival reference.

Project logo

Jan III Sobieski with a digital artefact is a modern AI-assisted interpretation based on a historical portrait. The logo is not a reproduction of the original painting.

Acknowledgements

We thank TsDIAL in Lviv for digitising and sharing the materials. The archive’s work enables further reading and study of these sources. This is an independent initiative, not the archive’s official website.

Visit the archive website ↗

Collaboration

We welcome help checking readings, locating published editions and expanding indexes of people and places. Comments that provide the archival reference, scan page and reasons for a proposed correction are especially valuable.

Illustrations

Privacy and statistics

Our usage statistics use neither analytics cookies nor external tracking scripts. We record the date and time, event type, recognised page path without search parameters, and the unit number where relevant. This database does not store search terms, messages, email addresses, referring pages or browser identifiers.

The IP address is used to calculate a pseudonymous identifier that changes daily; the raw IP is not stored in the statistics database. Unique visit counts are daily estimates, not counts of people. Event data are retained for 30 days. Separate technical server logs may exist.

Forms use a session cookie to prevent unauthorised submissions. Your name, optional email address and message are used to handle correspondence and stored separately from statistics.

Contact