@ariadne I’d like to support multiple “drivers” for this like tensorflow and Google Cloud Vision api, would be up to admins to enable and configure