Banks we are working with are using Co-pilot-like solutions to usually enhance the style / language of documentations or proof read them. In terms of more value-add documentation support, we are working with a number of banks to do what is being suggested here. Using the code-base, governance documents, regulations, model development templates/guides as input, and giving a best example model document in the training set, we are aiming to improve the effectiveness of AI. If interested, we would be happy to demonstrate how this would work to you and colleagues

I’d say best practice is to have a single taxonomy, which covers both Compliance and Operational Risk. They are often separate L1 risks, and the sub-risks within Compliance are often not aligned with the ORX structure, but rather speak to compliance-related risks (such as market conduct, customer / client protection, conduct, privacy, prudential & bank administration).