Some people are talking about OFX, and yes, there are tools for a lot of banks (for example) that convert their various export formats to OFX. However, I prefer

Some people are talking about OFX, and yes, there are tools for a lot of banks (for example) that convert their various export formats to OFX. However, I prefer it a different way: CSV. When loading a CSV, GnuCash lets you map each column to the GnuCash-equivalent field. Column mapping configurations can be saved for later reuse. One nice addition is that you can select columns that contain the _accounts_ related to a transaction. It's pretty easy to have ChatGPT write a script that takes a CSV and categorizes transactions into accounts using some matchint logic. Splits are a bit trickier as you have to use multiple rows, but doable if you enter a sample transaction first and export it to GnuCash CSV format to see how it should look.