The Nicene-Constantinopolitan Creed stands as a cornerstone of Christian doctrinal expression, composed in Greek and later translated into Latin and other languages. This paper details the development and application of our approach to sentence and semantic retrieval applied to the Creed, starting from the selection and benchmark of Transformer-based models and describes the gathering of NLP tools and corpora on the platform DaMSym. Our methodology applies state-of-the-art sentence embedding techniques to open-data textual datasets to capture the semantic richness of ancient texts, opening also to cross-lingual retrieval between Greek and Latin. By embedding the Creed's sentences into a shared vector space, we enable precise comparisons across linguistic boundaries, preserving theological and contextual fidelity. A particular emphasis is placed on the description of how the complex syntax, morphology, and polysemy inherent in Ancient Greek and Late Latin shape this challenge, as well as the preservation of doctrinal subtleties, with a focus dedicated to x-from-x like formulas (e.g. "Light from Light"). The study contributes to the field by implementing and evaluating embedding-based approaches specifically fine-tuned for ancient languages, integrating BERT-like models and domain-specific preprocessing pipelines. A detailed case study of the Nicene-Constantinopolitan Creed demonstrates how this methodology supports tasks such as intertextual comparison, semantic similarity assessment, and theological motif exploration across corpora: a particular interest will be devoted to the denomination of the Church as "apostolic". This research bridges the gap between the study of religious ancient texts and computational methods, offering tools for scholars to explore faith formularies with unprecedented granularity and envisions potential extensions to other multilingual, cross-temporal corpora.