ChatGPT Code Interpreter: Most Promising SEO and Data Analysis Use Cases

This content is generated from the audio voiceover so it may contain errors.

(00:04) [Music] Hello What such How are friends of the world of the artificial Intelligence another week more episode number 10 with Álvaro and with Arnau of iSocialWeb  what such How are very very Hello good morning Am Víctor and at all this week have been preparing things chulas have here episodes in the recámara of powerful people for following week have done glimpsed of which when we have ended up the interviews have said buah what chulo have to put to do things then good will go them publishing inside very little but has
(00:52) one paints incredible then Recommend You that when when they go exiting throw them an eye ask you again subject reviews commentaries be still in us the diverse canals of podcast that helps us a pile to achieve more audience a curious data is that I have seen that are like the sixty and beak in the Canal of technology of ibox that looks me surprising because technology there is a lot of but well this imagine me that spent to the cathedral will be like sixth page therefore neither like this neither are notable no but well when we begin were the 1500 no
(01:29) are how many of 60 and beak there is a progression appear by the pages where are indexando are in some cosillas arnals has prepared this week a curious data on the world of the artificial Intelligence and goes it to us to comment is a curiosity especially for which are not in the day in day out coaching models that is the concept of the sobreajusto of the over fitting and is the interesting is that when you begin to coach a model can spend that the model learn no so much but that learn so much the Dates set that end up as memorising it and to the
(02:17) final really what can not extrapolate the information but you leave it exactly to the frame of what has coached for example Imagine you that is going in images with dogs and the Date set no the the bank of images that use it to coach has a lot of lawn deep down or has for example always there is a door As perhaps he finishes understanding that a dog is somebody that is an element that understands it like a dog and the door that there is behind with which when there is not door or when there is not lawn understands
(02:46) that this is not a dog Then is as very curious the fact that no only is to coach it but afterwards also have to calibrate the on adjust because it Hears Really if I mount all these images of dogs with lawn without add another image of dog with lawn goes me to detect correctly to the dog But can indirectly has not understood well what was a dog then well I is a concept that personally like me am not in this world of the day in day out looks me curious that is not that you can
(03:14) even sin of excess more than to remain you backwards that think that is the evident agree Me when I initiated me these worlds thinking in Ah well are to program cosillas will know also do models of artificial Intelligence when suddenly you @darse that coach a model of pure artificial Intelligence and hard is extremely complex therefore you remain you short spends you the while am not well do not know what a few and throw you an atrocity of time doing the test and all this or was is very complex and the luck that have to to day of today
(03:48) practically not even lacking because have tools like which go to comment in the podcast all the derived of Open Cloud and all this that facilitates us an a lot of atrocity things as as at all we go with the news Álvaro go with yours avenges Go there As I bring you an ad that did official does very poquito and is that the people of antropic has #take clock 2 already know the commercial model possibly that more than more is competing with the models of openay And the truth is that I have been it testando
(04:21) afterwards I will count it in the africada but has things sew interesting go in among what comment that they have improved the new skills that has or the skills improved that has does Reference to the part of coding of generation of code that say that it is notably better afterwards will count my opinion dread of mathematics and subject of reasoning costs or was all the part a bit more technical that count that there is a better there is an interesting thing also And is that the price have maintained it the same that clotress 1.3 voucher And this as
(05:00) when they give you a new model and on do not shoot you the price as it is quite quite interesting is opened right now in United States and United Kingdom Although yes that it is truth that I yesterday went in without without a VPN and am using it do not know if with the pencil that have not arrived to try it still has some type of blockade but went in me from Spain am using it normal and there was not any problem so I do not know if they have removed limitations and other or is by the account that officially register it from another side no here would not know to say you But go
(05:29) in three and nine more cositas subject of tokens that always with clot is an interesting thing already know that we were in ciencas now have 100ks of entrance and ciencas of start or was at the end Would be like 200 K Costs then this is quite well because here the response is used to to be quite elaborated can him dip a lot of information of context as it already spent before but also of start with which is is genial and have dipped a species of approach to the elbow interpreter costs that it is all the part of his life of his life
(06:04) archives and can go up different types of archives to analyse them and process them Or was that every time already these these tools of analysis of documents and things like this every time go remaining a bit further back because already the big models are Are incorporating these functionalities that are relatively basic and that work quite well already in the model in the final model and afterwards another of the things that to me arose me the doubt was if the community already had begun it to testar in front of other models know
(06:39) the of chatbot sand that already have commented it in some anterior podcast to see if it appeared clock 2 in the in the ranking and as it appeared and for the moment at all there has not been there have not been updates So it will be necessary to expect a poquito looks Me that they have spent three four days since it exited or was that Well still are prompt Himself but I think that will #paste an important rise the truth is that we go will be hand manually with gpt 4 or very cerquita because already Cloud was very cerquita and east is notably better so the same have Some surprise
(07:12) here Javier had another another ranking of models that here also has a very big community probably by here yes that they have it a thing that when you have said that first it has come me in command Yesterday afternoon was giving him a lot of cane to lanchin that had some pending things to develop and such and was seeing that lanchin has in the calls that does to the model Has the capacity to dip an intermediate call to condense the content of the main call for example to gpt 4 limited by context can command him x
(07:53) quantity of text then what do first is to command it to 3.5 with 16.000 tokens does you a round-up of that of this Pro that you are commanding And then what #give 3.5 you take and you command it to him to 4 directly with this round-up no and when I saw this functionality that calls condenser or something like this thought automatically in Cloud Or was the capacity that have for example to spend a whole PDF ask him to Cloud that condense this in something that GP of 4 or another model can afterwards do is quite Guay then one of the things that molan
(08:33) of in front it is precisely this no the capacity to connect different models and at the end almost by the day in day out because because gpt 4 is the best with a lot of difference at the end only finish throwing against Open Ai and all this no But and for real Clau begins to put the batteries and all this could begin to have several obviously powerful models as to concatenate them or combine them to achieve things more guays Or was that Well looks me súper chulo that on what have said Álvaro that they maintain the cost of the Api that this
(09:05) it looks me súper súper notable so that we can follow using without that it cost us here a brown Himself besides the truth is that connect different models go technically does not have any complexity I have it mounted in some Script that have a part done with clot and another part done with with gpt 4 Exactly for this because the same needed to dip a lot of information of first context the process #take the information that need and afterwards Need a response of more quality and spent it to him to gt4 Or was that I think that is
(09:36) quite interesting and in fact already have spoken it some time the fact of Hears perhaps there are models that are more specialised in some areas can take advantage of in a part of the calculation a model determined afterwards in another another model and like this go doing a more optimised process was looking what commented of the ranking of models of High in Face and what see is that they are all Open souls are not do not know if there is another but what to me looks me interesting first the Cloud can try or was is curious because already in in pou.com already
(10:15) it can try Cloud 2 that it is the already is as by defect that is to say Cloud 2 of 100ks there is an interesting functionality inside inside to the hour to try it in Pou that is that already also gives you that this think that is a functionality that has not added him that it is already a suichest of follow-up of questions No I have done it a question like this in plan chungui and the response has answered me but afterwards has done me as it Hears I from here would recommend you these four or five questions or was think that it is a functionality that is like the
(10:44) autoplay almost of Netflix to have a conversation of elder or was of elder engagement No if I go you guiding also a bit the conversation remain in if it will be well bad for the society this but good in any case for the use if it looks me interesting the can include this functionality and I with regard to what said Víctor and was agreeing me of the one of the one of mick Journey of the version this of 5.
(11:13) 2 what was will not remember the functionality was was downloading is updating disks and can not it go up I can not see but the functionality is of mick Journey to say to what is giving him weight when when it does the prom because I think that perhaps all the ones of 100ks without removing him merit to the situation think that it is a thing as a bit that over time remain as very obsolete because at the end really what what think that goes to prevail is do a pront no the first prompt serves or was the first model serves to understand that it is important
(11:42) among this Bronx and the following is already execute it to the models that are heavy Or was I think that at the end what are doing we now is a bit churriminguero do not say it like this in the sense say him Looks really dip him all and dip the window of ciencas obviously by the part to dip him big archives but in the time what goes to be Goes to be much more efficient as as it said Víctor with launchen a first agent that put to look for the pieces of useful information that Mount the pront and it refine And
(12:15) afterwards the following that Launch it And to this do not refer me to the prompt engine and Víctor that I was trying it that I am using it by private that commented in the in the anterior message and this is a consumer of tokens brutal but but I think that To where go then this mix of a refinador of proms that go to the prom of the big model already comment it with the football chipití I think that has all sense of the world Expect that by the current time the Cloud 2 already can try in Pou without need to pay at all this think that
(12:48) it is well it was limited to have seen some capture do not have Pau But of some compartición by here that had some type of limitation for example or was although it appears the message of access limited to me yes that it leaves me use it Himself is that I am Ah okay if the limit of Access I think that is because afterwards yes that there is one that am still thinking that I am beginning with which are not it using that is the best option of all because the if you fix you what is limited is because at the end they have a quota of payment then the
(13:23) interesting is that paying 19 euros to the month give you access to all the models Or was think that east could save as that or was say I if I am an user and want to begin before before pay this say that I say for the global fan can not or was by quantity no by quality by quantity pays 19 and have access to gp4 access to Cloud to a pile of agents and really this Or was I am still thinking that it is a project that is of [ __ ] done because it is that anybody goes to consume this consumption of apis they a business of Rope consumes by
(13:59) Api But to you charge you a standard no Then is as anyway it is súper very thought is this idea Then I would try by here I I was giving him gone back to do something alike when when they began to exit all the apis and said by the same anybody goes you to spend to the limit that you pay to the month such But the first that thought that what has come me in command when it was saying this is that although you are consuming the Api gpt 4 to follow having the code assistant all the new things that come exiting and that
(14:30) they are not available in the appy and probably this yes that contributes a main value and at the end finish saying oysters is that yes I am me saving paste with this but am leaving me important functionalities in another place and probably the people always prefers to go there to use them But well that for for users that are beginning and want to try all the models 19 turkeys to the month by this is a bargain okay mine mine news there is a lot of clickbait in the news but looks me interesting costs Leo the headline of the news because there is
(15:08) that do is well the endeavours that there is in doing clickbating one and to China creates the first processor without human intervention is four thousand faster times that it was gpt or was voucher in creating Processors as it was the correct news or was the interesting that to me looks me as or was looks me a case of use now against the news very very interesting because basically What sucede is the clear Chineses envy is an American company voucher And clear they are those that are catering all the subject of the cpus and all the processors that at the end
(15:49) they need for all the subject of artificial Intelligence fish a bit the commentary already speaking to level Macro politician of the that do not have neither contrives Chinese forces United States the competitions of Market ta ta ta clear Chinese does not have access to all this say do not have neither idea then clear a bit the as they do not have access to this what have put is have put a model the kiment 1 to design a processor that was qualified for as be the equivalent to the versions of the of the to 100 or of the H100 of envy Then is as curious that
(16:23) really in five hours removing clickbait and all this know really that it have achieved what they say As it Hears is that it is 4000 more intelligent times that gpt 4 Further of this of if it is it or do not think that what is brutal is that really this case of use of The does not go me to mount the Hardware but I yes am able to coach it giving him all my inventory of of products and that he of intelligent form begin or was I give him the frame of where will go the microchip and all this and I and really begins to do the design of all the electronics this is one
(17:06) thing that looks me as very interesting and is a case of use that think that to day of today yes is much more applicable of what think us Know because at the end a microchip Hears if it is this condenser this can stick with this with this and with this this condenser does not occupy 3 millimetres or good or three centimetres or was and at the end are parameters that if we think it are quite basic know or was are quite basic in what or was it is necessary to coach what that do not remove him merit but looked me very interesting
(17:33) the one who is have seen it in all the subject of detection of X-rays and all this that is something like the human eye is not able to see this type of things or in the system astronomy in the gastronomy is not able to detect some type of planetary movements no as when it discovered Pluto but I no had posed it to me in designing things that also are with pardon of the pressure works of Chineses know Or was that it is basically as design a microchip or directly a whole processor then well that is the idea more
(18:05) there of the clickbait looked me very interesting because when it said at the end inside five years there will be cpus better designed by artificial Intelligence that by humans is that it looks me that evident that goes to be better because it is that it is much more efficient to do this type and all east assay error assay error without me dip me to dip a quantity of applications very strong that this is a bit the thing is that I imagine me The initial idea is like imagining you the uncle with the maquinita mounting it that this also goes to arrive
(18:34) but it is that I do not need to arrive to this to understand that the own design that is what will give me performance this is what is powerful and afterwards already will assemble and already will be another another but is that only the assembling already is another thing Know then good that is the news and looked me quite interesting the truth is an atrocity because on it is that it is applicable to all the chains of process of production of the world anything do clocks do car do do not know what Or was to arrive to points in which this production goes up to level the
(19:05) same example comment it here or was sure that has done of another form but but big shots the same That this vaccine that did no us vacunaron antibiotic that did in base to east in my laboratory have all these these chemical elements that you can join generate an antibiotic for such through randomizar of do not know what no How many and detest in each one of the tests at the end finishes exiting A that those Scientists are able to produce in this moment because have these available resources in the own laboratory
(19:37) as it is that it is an atrocity And the of the processor this is also very notable because it no longer is alone that we are able to produce processors faster but when being able to produce a faster processor will be able to do it of main quality when it normalise this and at the end the computation as we know it the personal computers also will have another big jump of technology precisely by promoted by the artificial Intelligence this week Now will not say names But have been speaking with somebody of the
(20:09) world of the quantum computation and that Expect that sometime it avenges the podcast and sure that goes a lot by the line of this no that at the end it is something so theoretical so complicated to arrive until this point that have to lean all these technicians to be able to be able to do all the calculations for to predict a bit To where go these technologies Very well as I go I with my notification totally to say that thanks to that Álvaro has spent it to me and have news today on my part So thank you very much
(20:42) Álvaro speak of an article of Open it is necessary well Are lately publishing a lot On more on the ethical part that on his own product I think that also is a bit to appoint that at the end this hype that suffer in February March in which now you grieve does a presentation now does it Microsoft now does Google as it goes normalising already are in the point is in which no longer we confront us to so many to so many changes in the different models besides arrives the summer all the world goes on holiday and well
(21:21) lately they are more speaking on on ethical on future and other more than on his own product no the curious of this news is that there was a paragraph that has liked me a lot that think that is important to review speaks on the super alignment that are developing so that the súper intelligence was able to arrive sometime in this article speak on that the superinteligencia will be by much more intelligent that the humans and here yes that begin to speak of Terminator and of similar rolls and then like that
(21:54) they are concerned among comillas by for controlling to this superinteligencia no and what has liked me a lot of this article is that what appoint is that they are devoting 20% of the computation that at present have to research have scheduled it like this During the next four years to resolve the problem on how range this superinteligencia with the interests that have right now to protect the human ecosystem then like that it looks me one a bet a very big investment that invest 20% of something so
(22:32) valuable like the resources that it is right now for Open that remember that Open gave a no official interview and in one in a company in human Loop and afterwards human look remove this article and in that this interview that God amalman said that his main problem right now is scalar the use and that are not using for example the multimodal because they are not able to have capacity of sufficient computation as to launch all these things As now it exits a news saying that they are devoting 20% to investigation therefore also
(23:06) it sees a bit here this philosophy of openide of Well have #turn a bicharraco But on the other hand keep on being a laboratory of artificial Intelligence That wants a bit indagar Research on new things no then good will go seeing as as they pose it in the article speak of interesting things on that right now they are able to unify diverse artificial intelligences but to generate a súper artificial Intelligence would owe a montonazo of these artificial intelligences and generate this
(23:37) that this big structure and this is what right now are not able to do the past week comment that it had Filtered a possible a possible architecture of how worked gpt 4 that it was at the end connect eight models Probably is that these people already is researching in connect 100 200 300 models and try to achieve achieve something like this bigger no so good article a bit bureaucratic but but has liked me a lot That this part to devote 20% of his computation to the investigation neither had a lot
(24:11) alternative Víctor that it did not like you that it was Álvaro as Álvaro gives me options no but that I I this week is being very complicated the audience no but that I think that I think that it is súper important the one of computation and afterwards is this yes if really it splits of his endeavours instead of this 20% a 10 would owe to devote it this think that comment it do not know if in the episode eighth or seventh the subject that Google already had achieved to reduce 40% the consumption of processors by optimisation of resources or was a bit the one is or
(24:59) it was it is this idea to mount an efficient processor but is that clear is that the same logic applies in what or was more powerful excuse but also have to Google according to the news developing or was devoting endeavours to optimise the resources that this is a part or is a leg as very very very correct No if you see it like a company is more income is to sell more but is that the costs is your processing because more it was of the squad is in computation goes to have some brutal costs then looks me as very
(25:31) hit this part that perhaps do not know how much are devoting him the ones of openay And if it interests them looks me intelligent that Google himself was doing it and afterwards for me keeps on being a joy the news that says the past week that GP of 4 was the amalgam of varied of several models because already also are it seeing with all the interviews that are having with people that is in the world already are it seeing that even lang Chain was mounting to specific agents for micro tasks and basically what are saying us
(26:01) the experts was Hears looks is that in the moment that the uncle that put some people do a lot of things loses Then at the end is much more first that looks me much more much more of the best news that I have heard long no the one of that they are an amalgam of several models because it is that this is the best that can spend us that really it was an amalgam of a lot of models no And we adding those that go being useful and those that do not go being useful do not put them But that it was the typical of does not have a trillón of parameters of do not know what
(26:30) hostias that this looks quite difficult to digest know and already see it in all sides in the companies is famous the model of Google of micro squads in the Or was that in all sides look the same dynamics know that it was much better little assembled and agile that no one a cathedral of parameters data and events then Good looks me interesting by those by these edges also the news to business level has quite felt For example thinking in Palm That well have it here a bit averted no but already when they did the presentation
(27:07) the ones of Google Cloud Google already appointed that they went to coach to medpal and to the of ciberseguridad of such And has a lot of sense no because Claro if you leave that users do questions in Google medical and goes you to answer the Google the Palm generic as it goes to make a mistake but if there is something an intermediate Layer some people or a specific model a router that is able to differentiate To where goes go you to ensure that all the medical questions go to medpad that is súper optimised for this that has a source of information
(27:41) very concrete that does not go to exit from here that know with that you have coached it that you have not coached it with any blog that was speaking on medicine among comillas and go you to say something Or was that to business level also covers a lot the backs and do it of this then good form will go seeing I the truth that also me Mola that Concept the curious is that already you comment it the past week in the article said the contrary said taken care what Open here has not reinvented the wheels simply has jointed eight wheels
(28:10) ruedaban better that one that have joined the two news that gave you like this of a form without @darse you practically are here with the ones of Clinical May Or was that yes yes Good simply This looks me that has felt Well as we go with the subject of the week that have not said at all this week there has been Revolution again because Open there is announced during the coming month Go to begin to use and the following day Already had actuated the code interpreter And the truth is that well is internet doing tests and
(28:55) going up videos tweets in linkedin such And here such quantity of cases of use that is that it looks that they have given us another chat gpt and are again confronting us to this is an atrocity how have lived it you that are right now that it is what are doing and in which point of hype are because I already have listened something of weapon or before and some speeds have right now on on this as I in my case am quite hypeado or was was trying it So it is truth that gave me access that in fact spent it to you as I said looks already have
(29:31) access looks Me that it was the past Thursday or something like this had had the access put me to play with this and think that opened it to us to all at the same time and do not know ranged the stars that just among and my first experience was [ __ ] What [ __ ] is this because every time it did a question This is blocked this do not know what do not process the document the document is damaged or was was the first credits melted them to me simply in this but to which left that it went down all all the vorágine of people here testando like crazy and began to
(30:04) try it the next day the truth is that the results have me go are liking me are liking me quite Perhaps before going in in opinions could explain a bit what is and which limitations has and like this a bit in matter if it looks you that I do not know if that you want to explain it any or throw miles As at all basically so that it do not know it the code interpreter Is a complement that they have added in chat gpt that is in beta like the half of the things that are trying and that allows us inside inside gpt 4 do
(30:38) an instance that can execute code of python no only this but besides we can go up archives so that they can process inside this call to this colour costs has a series of limitations eye that said code in python is not code in any programming language is in python Although there are some limitations Here and now will see interesting things limitation number one that does not have access to internet Costs that this there are a lot of people that is complaining but now will see that well this limitation is
(31:09) quite logical afterwards there is an interesting thing that is that this instance of the machine that are giving you is during a period of use and if you do not use it of active form dies and loses all the context that looks a foolishness But there are times that are doing an analysis calls you somebody by the chat do do not know what go back And have remained you without without the half of the information that are processing here or was that this is an interesting point afterwards uploads a maximum of 100 megas by document although already they have done him jailbreak to this and already this have him
(31:39) jumped only executes code of python as it said you and already have jumped it to him also does not allow the installation of external packages but also have jumped it to him Then this is enough quite Guay there is a Polish developer that will leave it in the commentaries that calls scals or something like this afterwards will put it to you here and this uncle has a web go a URL in github where goes counting How has jumped the different stages and how documents it as it has done it and such And is súper chulo But at the end basically almost all the hackeos or the
(32:20) jumps These the jailbreak that are doing is basically that as it leaves you go up clear archives go up anything Costs then when #can go up Anything as you go up a package of another of another programming language install it as it already executes javascript for example or as it leaves you cruise by the structure of folders as they have @darse# cuenta that they could cruise by the structure could see all the tree could execute a model that there was there like gp2 or could change him the permissions to the structure of folders and in the
(32:49) moment that changed him the permissions could install what gave them wins it or could change them the parameters of configuration with which the of the limitations go is acojonante the truth is that the people go us to remove the code interpreter next week are not ready still was súper chulo But well did official limitations would be would be those and now already if you want to arnautas your part of hype Himself or was I I have a lot of hype on a concrete part of the of what can do with interpreter in addition to what have said
(33:24) also it allows that they are things that are quite interesting now tale if if I have achieved it no because it is that that is another but things as they cost me súper powerful as really extract the text of an image costs this according to photos because I already say I this have not achieved it know can extract photos of an image can extract information of a PDF also another thing that I neither have achieved voucher can spend of images to videos this have achieved it but with a Chrome that sincerely no had occurred me
(33:56) never it costs that basically it is I took a photo a long photo say him Hears try to spoil it to me or was try to split it to me tries to do me a Carousel of left to right costs I tried it using my own did not achieve it afterwards achieved found an uncle that had achieved it and is that the prompt was help me to #turn it a video with relation appearance 3/2 with the image filling all the video the video has to be centred in the centre of the image to continuation displaces the video softly without jumps of the following
(34:25) way initiate right centre and left centre centre end uses the bookshop and majeo to help you saves the fotogramas directly in an archive of video in place of a list uses a step of fotograma of 8 pixels if necessary recorta the brinks of the image so that the size of the image was divisible by the step of the fotograma costs Like this achieve to do you one a person already have done it before do not understand how no has occurred you as then really I all these functionalities are súper powerful and I think that the
(34:59) it splits that it is even more powerful is the one to convert csv to code I have not used it because I have not found me with the need or the case of use but think that is a cucumber the truth the thing is that I think that it is a cucumber in beta and and I think that to day of today for my uses the one who more or was the one who more possibilities would have would be obviously can go up PDF and speak with him style kills that to day of today does not allow it or was allows it but do not achieve it I by the motives that was but to me what looks me more interesting is
(35:38) clean clean or was to clean and standardise data I think that there it is this part goes to be for me the most powerful of all all all all because I think that one when those that are analysts and work with data as have data of a lot of Sources is literally a horror or was a horror already only in accounting has in downloading you files of banks already @darse that each one the as it wants to one or was there is not a normalisation Then I this process to normalise Think that think that for me goes to be for me already me
(36:15) it is helping I is here where think that goes to be more powerful do now do a disclaimer eye that Open said do not have do not know if you this information but said that we do not coach the model with the data that come Of Api but If we coach the model with the data that come of the chat Then I do not know if you have seen if cot interpreters expands to this but go is the chat with the function of cot interpreter Then this looks me that it costs very well but to see which data dip him there because to clean and
(36:44) standardise I want to dump you the csv complete do not want to you dump a muestrilla that afterwards do not know if it fulfils all the cases of normalisation and other Then hype by what can allow no hype by by all the flanks that there is behind this no if it coaches the code what say that they do or have seen it and does not allow it Well I am a bit to expect by where improve this the casuística to which have confronted me now that are saying #cleansing of data and has looked me brutal my first feeling when I saw it was a bit of
(37:29) Well I already am programming already have a window of python open program enough in python every day think that do not need that something execute me code in the browser because I am launching it I in venue and already have communication counted and am I programming lame the code put It here such and on the paper said do not think that for me go to be something notable And then put me to try it and @darse of the mistaken that was have brought cases of use that have explored for commenting something by commentaries mine or expect me to that your
(38:13) line and think on this a case súper cathedral of #cleansing of data and of projection and tried to go me as to the most generic as if it did not know as if this had not done it never did not go me to these consol and went down me of these consol the report of of the signpost of performance that allows you extract of one thousand rows in one thousand no Good rows I all this usually I have a database that connects to be consultor every day goes down all the information save it to me a my sickell and I afterwards with power bi model the data
(38:52) and to the database of Google of productive And then mix the information and generated me my give voice from here do things like calculating the the economic impact that would have go up a position or not to go up it inside all the project cathedral and like this is as it defines a bit the strategy of where go Voucher then spent me this case of use precisely to this point then but to a form more less automated less connected to databases something that could do I in a moment in which I need to do something fast and as I have
(39:24) said arnau like this spends long to generate to clean a lot of data said go to try achieve a inside a data of the fastest form possible then go down me to be with Sol the ZIP whole that contains five or six tsv with thousands of registers each one literally think that are one thousand registers each one or 10.