This paper briefly introduces the recently started DESCRYPT project (Riskbankens Jubileumsfond 2025 – 2032), which aims to develop a scalable and interdisciplinary framework for the automated transcription and decipherment of historical manuscripts, written in rare or unknown scripts. The project integrates expertise from various domains, such as, for example, computational linguistics and cryptanalysis. The present work focuses on aspects related to computer vision, in particular handwritten text recognition and document layout analysis, outlining challenges and advancements in low resource settings.