WMy students and I are developing a project to see whether we can refine and customize a "DigitAI" (named for our "Digit" program at Penn State Behrend).e want our DigitAI to serve as a guide to TEI XML, and we are preparing a RAG knowledge graph for it built from the p5.xml single source of "ground truth" of the TEI Guidelines. The challenge of our project is to introduce the meaningful hierarchy of XML and TEI to an AI language model trained on sequenced tokens and word vector embeddings. If successful, the DigitAI will be minimal in size and adaptable by others, but most importantly, we hope to learn from the experiment.