The current batch extraction example prints API errors but could provide clearer context. Suggest adding clarification about error codes and retry behavior, plus best practices for handling rate limits.