their pipeline would be to just prompt it's internal next gen models to create a jev copy given all the data they have as a first pass.