This site publishes machine working drafts over public-domain Armenian print: a digitized Classical Armenian dictionary, complete bilingual service books, and a shelf of the period's popular and curious printing. Every transcription is produced by two independent vision models and reconciled against the page image; every English line is a machine draft that carries its page reference. Every Armenian word on the site can be tapped for its dictionary entry, resolved through a morphological layer built for classical inflection.
The rule this project runs on: never publish a page a scholar was hoping to write. Literary monuments with living translators appear here, if at all, as Armenian transcriptions with full apparatus and a deliberately paused translation layer. What this site publishes freely is the layer every reader uses and no career is built on: dictionaries, rubrics, almanacs, imprints, home medicine, games.
What a draft is: a way to find and read, not a thing to quote. Anyone citing these texts should read the page image, which every passage links.
Sources: the British Library Endangered Archives Programme (EAP180, originals at the Fundamental Scientific Library, Yerevan) and archive.org public-domain scans. Compute runs on research credits.
A Steps Ventures project. Contact: mike@stepsventures.com.