Collaborate on Invoice PDF for Support with Ease Using airSlate SignNow

Invoice pdf for Support

what's going on everybody Nick here one of the most common requests for automator is hey I have this PDF or this file and I want to parse it and turn it into some s sort of structured data how can we do it in this video I'm going to show you exactly how using a PDF parsing library make.com and artificial intelligence in my specific case we're going to be doing this on invoices but you can apply the exact same approach that I'm going to show you to literally any type of data out there whether it's maybe a resume white paper an ebook an article PDF literally any sort of file type can be converted into structured data using the system that I'm about to show you so if you guys have wanted to level up your Parson game learn a little bit more about Ai and maybe a couple of the cooler features that make.com has that you might not be familiar with this is the video for you stay tuned let's get into it all right so what I have here is I have a Google sheet and inside of this Google sheet I have an invoice URL column where we are going to store URLs of our invoices you can get the URL of your invoice in a variety of ways sometimes you'll also just get the raw file say if it's emailed to you or maybe if it's in a big Google drive or something that the client uploads um but I'm just showing you how to do it in this way and this is probably the more complicated way so if you know how to do it the way that I'm about to show you'll know how to do it in every other way as well then we have um some fields that we're going to just parse the invoice for so we have an amount field which we're going to use um to just store whatever the amount is that that is owed we're also just going to take all of the line items here and then just pay P them in this line items column and then we're also going to um do some fancy text parsing and we're going to extract the uh the from address as well uh we're probably going to use AI for this cuz it's a little more um creative of an exercise in terms of what the invoices themselves actually look like I have them open over here an example might be something like this so this is a PDF from I don't know some service called sliced invoices which has a uh which has a default template available for us here and uh PDFs are kind of tricky to parse you know there are several different types of PDFs and I've explored all sorts of them over my career like you have PDFs with clickable text so for instance this is a PDF with clickable text I can actually double click and I can select the data and actually most of this is like already formatted for us so the parsing is going to be pretty easy um I want to say that the majority of PDFs nowadays are probably like clickable text if they're being exported from some invoice software all the stuff is already done for you but there are also a lot of PDFs out there that do not have clickable text that are essentially just images they're sort of sort of like PNG images or jpgs or whatever and when you Mouse over them um it doesn't actually do anything uh and you can't select it and so because of this there are a variety of different ways to parse PDFs you can parse them through like um if you guys are familiar with optical character recognition basically there are a bunch of services out there that allow you to feed an image and then it will try and make out whatever the letters are so you know we could we could like have Let's Pretend This is a PNG or something or like a JPG right those are those are both image files we could take these image files and we could feed them to some optical character recognition service and then we would say convert to text and they would try to kind of figure out that this is an s that this is an L that that's an i you can imagine because there's like five billion different font choices you know like the the the endpoints um for this optical character recognition and a lot of them are like older Services as well from 5 10 years ago before we started using Advanced AI they're pretty crappy at that and the results from the ocrs are are not very good so that's sort of like one way to do it but another way to do it is um you could just feed in um like the whole PDF into chat GPT or something um GPT 4 seems to do a pretty reasonable job especially if the text on the invoice is selectable like this um so whatever your PDF style is you know will have sort of an effect on on the bottom line but yeah that seems to work okay and then the last thing is you can just use like a like a dedicated PDF parsing service and that's personally what I do nowadays I just use like pet dedicated PDF parsing Services cuz they're super easy and super simple so I'm not yet affiliated with any U dedicated PDF parsing Services I've had a bunch of them reach out to me and try and you know sell me on sponsoring a video and that sort of deal um but I have poked around and I've tried a bunch of different apis and the best one that I found is a probably the simplest one to remember just called pdf. uh pdf. uh basically if I just go to their little website here um Brands themselves is like just the the PDF and all be all that just turns PDFs and apis basically um and it's just very very straightforward they also offer you like 10,000 credits I think for free at the time of this writing uh so you can imagine how like low cost it is on their back and how streamlined and efficient everything is and you can use those 10,000 credits um to do everything that you need to do in you know testing purposes and then once you figured out whether or not this works for you and you want to you know implement it into your Enterprise workflow then you can obviously sign up for a plan and subscribe and that sort of deal but basically um the the purpose why of using these dedicated PDF par platforms is because it just gets the the data into a format that is much better um that you can use for basically anything else later on like if you think about it text is the universal interface right especially nowadays with large language models so if you can just like convert a PDF just into text format doesn't even matter if you lose the formatting it doesn't matter if like a human being would have to like squint and stare at it for 3 hours before making any sense of it you can just feed that into a large language model and it can like process and for reformat the text and extract everything for you and that's sort of like the Bas upon which now U most of the the the PDF parsing flows that I've built out rest upon and that's what I'm going to show you how to do in this video I'm going to show you how to feed in one of these invoices into a PDF parsing service we're going to extract the text we're then going to feed it into AI then we're going to use AI to extract various portions of this text that we are then going to map into columns like amount line items from address Etc and I don't want you guys just to think that this is only doable with invoices you can do this with uh basically any type of PDF off format another common request is doing this with resumés so you know maybe you're like trying to hire somebody or something or maybe you run like some resumé service we can actually just parse the PDF and then feed it into AI extract I don't know like their job history you could extract how long they've been working for what their hobbies are you could obviously like extract keywords if you want to like automatically not allow somebody in or you know automatically Greenlight somebody so so there's a lot a lot that you could do obviously and um yeah that's one of the reasons why I love AI so much and I love these applications so uh what I'll do in this video is I'm going to show you guys just how to take these and then just parse out these three columns but just know that you can add however many columns you want you just need to like add the logic on the AI parsing side and you don't have to use invoices you could use uh more or less anything and then just to reiterate uh you also don't have to do it in this format like you might be getting them as emails um but this is just the hardest way to do it so I figured I'd start with a Google sheet method okay got a scenario open here called Parson invoice add to sheet first thing I'm going to do is I'll just go to sheets and then because we're going to be searching so I'm going to be testing this flow out a bunch I'm going to be using the search rows module so I have my um email address set up over here and then what I'm going to do is I'll just go to my invoice parsing example sheet and then I have to select the sheet name and map all the stuff so bear with me I'm going to select a maximum number of return rows and I'm just going to set that to one and that way if I right click on this and press run this module only I will always get the exact um same invoice and then I can just use this to test so you see the invoice URL I got here was this right and this is the file essentially which is pretty cool okay um there are a couple other things that we need to do but I'll just run I'll run you through the PDF stuff um just go to add and then type pdf. Co and then there are a bunch of endpoints here that you could use right there's convert from PDF convert from PDF to images convert into PDF convert HTML to PDF you can do a lot right they have like a million end points that are PDF specific the one that most people use and the reason why it's first is because uh we're going to convert from a PDF to one of these structured data types and I'll show you like three or four of them just that you could um see what they look like for your use case now when you click on this you're going to have to connect it if you've never connected it before so go to add and then what you're going to have to do is sign up with the pdf. API key the way that you get that is go to app. pdf.com sign up there'll be an API key over here that you can just click copy then once you've copied you can go back to your make scenario just paste it in and then after you've pasted it in um you know you're going to be be left with a connection like this I just called my invoice parsing example okay so there are two options to put in the PDF you can upload a file or you can import a file from a URL now obviously our files are already in url format so it's going to be very easy for us just to paste the URL in um let me just refresh this and then go invoice URL here and then uh we can also add an output file name so I'm just going to do like example for now all right now there are a variety of different ways that you can do this there's a simple endpoint PDF to text simple no layout fast PDF to text which is where they try to preserve the layout of the PDF for you so uh what I what I mean by this I mean like well from is over here on the left right all this stuff is over here on the right they actually are going to try and convert this just into plain text like a plain text file and then just add a bunch of spaces between the two to so that it's still in the same layout um there's no real reason for us to do this because we're just going to feed this in AI so it's more or less meaningless uh you can do PDF to CSV PDF to Json 2 uh Json 2 is just their like new endpoint which converts into Json I haven't actually experimented with any of the Json or CSV ones so we'll do that together uh you can convert into HTML you could also just like turn it into a JPEG or a PNG so one of those like um text list images that I mentioned before and then you can also do like XML Tiff and that sort of deal we're going to do the simplest one to start PDF to text um and then Pages allows you to par specific Pages if you just leave this empty it'll do every single page if the PDF has a password so some invoices may be password protected some Rising Ms may be password protected you put those the password in here if you're trying to do this like on mass or in bulk you can imagine how you might need like a password column to like dynamically map the password to the specific PDF um and then uh you know we want a small task so I'm just going to go to sync but if you're running large batch tasks so these PDFs have like 15 pages or something then select async essentially what's going to happen if you go async is you'll send the request to the server it's not going to Output the the actual like result in this module um you're going to have to go and then grab the result after a certain amount of time um but we're we're not going to do that we're going to select sync which is very simple and very easy and then there are um some other they have like a big list of documentation um they have a bunch of documentation that shows you um how to adjust uh the way that the PDF is parsed using what are called profiles we're not going to do any of that so we're just going to feed that in here and then output links expiration they actually set an an expiration date on the output Link in case you I don't know are doing this at like batch services or in case some of the data is um secure and you don't want to like reveal it you could set this theoretically like one minute the thing that you'd have to do here is you'd have to do something with the data within a minute essentially you'd have to like request that URL that it gives you uh within 60 seconds so in our case I'm just going to leave this to 60 minutes because these are just silly little invoices that are examples that don't really matter then the export type is download a file or Json output I'm going to show you how to do both um but essentially you can either turn this into a file that you download and then parse or you can turn this into Json and uh why don't we actually use Json to start that'll probably be a little simpler okay we should have everything we need I fed the invoice URL in and then I've done the rest of the settings and then we're going to pull that from here so I'm going to click run once let's see what happens all right so our Google Sheets search row module ran and selected the correct URL and then we converted the PDF um and it sent us back a bunch of um a bunch of codes I guess there's a page count so it counted up the number of pages in The PDF um there's an error status and this is false so if this was true maybe you could send up some error handling to do something with it um there's a status there's a file name there's a credits remaining credits how many credits you got left the duration I think the dur duration maybe the number of characters or maybe like the number of milliseconds I don't know but you can see we Ed four credits there so we technically have 9,996 I don't know exactly how they're doing that processing but whatever okay now this is the text that we get we get a bunch of links we get um the title which is invoice we get the payment is due within 30 days from the date of the invoice late payment is subject to fees of 5% per month thanks for choosing demo sliced invoices then we have a from as you can see here we have the inv number we have the due date we have the total due date right the two we have um some line items here is pretty cool then we have the total and the total is what we're going to be extracting um along with let's see here so we are going to be extracting the line items we're going to be extracting the total amount maybe we'll just call this total that'll probably be easier to see and then we'll also try and get the the from address as well and I want I actually want to get the full address I don't just want to get the email address so we may have to do some prompt finagling there but uh yeah you know this is just in a format that we can basically take and now dump into Ai and then AI can like actually parse and extract out the text for us the benefit to doing it this way is you don't need the invoices to always be in the exact same format like previously procedurally if you wanted to parse this you would have to like parse the term invoice number space and then you know however many spaces there are between these two terms you'd have to like match it every time um using you know Reax or something like that but now you don't you don't have to do that you just feed this in Ai and it's a lot simpler so I'm just going to show you how to do this now you the create a completion endpoint just while we're on a roll and then I'll go back through and then I'll run through some of the other data types here which may be a little bit simpler maybe even a little bit harder um there are a variety of models so I'm going to choose the create a completion endpoint uh we could do gbt 40 we could do a very um handsome individuals driving a very cool sports car outside gp2 I don't know why I always leave my freaking balcony door open when I record these videos it's so funny I got to stop doing that uh gbt 40 gbt 40220 4513 gp4 turbo gp4 I'm going to just choose GPT 40 uh this is like sort of a I don't want to say it's a dumber model but it is kind of a dumber model so we'll see if it works you're a helpful intelligent um let's say data extraction assistant that's what I'll do then we're going to add some messages I'll say you extract excuse me I'm about to sneeze no I'm not you extract data from PDF invoices you'll from PDF um invoices you'll receive a a text we'll just say like a text scrape of of the invoice data and your job is to turn that invoice data into three structured Fields we'll say um total and then line items and then what was the last one here from address um return the results in Json so that looks pretty good to me uh let me just think if I need to do anything else for from address I'm going to say not the email but the actual address of the sender then light items can just be a plain text dump then I'll say using this format and then I'll say total total goes here we'll say uh line items line item string goes here and then we'll do from address and I'll say from address goes here respond only in Jason your first character the first character of your response should be a opening bracket okay great the reason why I'm feeding at all of this information here is because I don't want to have to go through the pain in my ass of like teaching it basically training it with a few examples um you could and it would probably be a better choice if you wanted to train it uh you could feed it in a couple of example PDFs and then see what the output is and then add a user prompt for your instructions which would be something like this and then an assistant prompt sorry another user prompt with an example then an assistant prompt with like what you want it to output and it even if you provided one example it would do a really good job of um of adhering to that okay now what I'm going to do is I will feed in the let's see the body okay so it's just body so now I'm just going to feed in the body which is I believe just going to be everything I hope it's going to be everything I don't know I guess we'll find out um max s I I said 496 here um but that's just because that's like the maximum it'll ever output I'm going to change the temperature to 7 I just find it a lot faster and easier to do it that way and yeah I'm just going to give this a run and we're going to see what it outputs the reason we're doing this in Jason by the way yeah so it already screwed up seed add of these three little back Texs I hate when it does that um but the reason why uh we're doing in Json just so we can parse it later which is really nice so the total was 9350 which is good the line items is this the from address is demo sliced invoices sweet a okay great I got the from address and stuff like that as well which is perfect the only thing we need to do is we need to change it that it does not output do not output backtick Json or any other sort of formatting okay I'm just writing a very explicit instructions here then what I'm going to do next is I'm going to add a parse Json module so go to parse Json drag that puppy here um you know you may not even need to do this anymore I mean gp4 keeps on changing um you they keep on changing how they do this I believe you can now potentially even use Jason object but I'm just going to use Jason here cuz I love wasting credits no I'm just more comfortable using it this way and if there's one thing that I know it's that um make.com is a platform where just learn how to do it once like you should obviously be constantly improving your skills but if you know how to do it then do what you know how to do um a lot of this AB a lot of this complexity is going to be abstracted away from the customer or the client or whoever you're building this for they're not going to see any of it okay great now I have the output here which is nice we have a total line items and from address and now what I can do with this is I can update a row which means I can go back into my Google sheet and then I can add that data so I'm going to go to invoice parsing example here and then I just have to map the rest of this like I did in the first module what we'll do for row number is we'll go here select the row number from the very first module which actually outputs the row number and then not going to update the invoice URL cuz we already have it what I'm going to update is the total the line items and the from address okay so let's give this one final run and let's see how it performs if I go back to my Google sheet now you'll see that I've now extracted all of the three pieces of information that I wanted to the Total Line on in the from address and that looks pretty good to me honestly um I mean we could change the format of this maybe we don't want you know one dot z00 Web Design This is a sample description but my whole idea here was I was just going to pump in the line in them so that you know if anybody wanted to like double check this later that they could easily you know go back and and do the math um we could also like extract taxes as I mentioned ear you can extract whatever the hell you want you could do taxes you could do the due date you could do the invoice date you could you know grab I don't know like if does this include Bank information if so add their Bank details here um what is the payment due policy right what is the late payment subject fee right you can extract however many columns you want here I'm just doing it um I'm doing fewer because I just want to show you guys how to do it and then all of you guys have your own use cases so that's that okay so we've done it on one example one example is never enough to like verify that this works so what I want to do is I want to run this through another example one that's much more complicated and see if still works uh and then if it doesn't work then see what sort of changes we have to make to allow this system to scale as needed and essentially handle like multiple different forms of of invoices and data now keep in mind you may not have to do this for your use case you may be receiving all of your invoices in the exact same format and all that stuff so I don't feel compelled to I guess is what I'm saying now what I'm seeing here is this is a really really long invoice this is like five pages or something so realistically the invoice is really just on the first page and then everything else is like a bunch of silly stuff that's probably like required by the government because this is um excuse me Canada Post it's sort of like uh I don't know like Federal like postage thing probably in the US uh if you're more familiar with that so what I'm going to do is I'm actually only going to grab the first page of this um we could even have a rule that says like if the URL contains Canada Post or something only grab the first page uh you can imagine how you know you're probably not going to receive invoices from like everything and everybody under the sun um so that that seems pretty pretty reasonable to me so why don't we do that so here's a comma separated list of page indices or ranges to process first page is zero so I'll say if and then I'm going to say contains URL and then it'll say Canada Post then uh let's output zero otherwise let's output um empty string or null I don't know I'm just going to do empty string so logic here um is it's going to check this invoice URL which is this over here Canada post- whatever then if it contains the term Canada Post then it is going to um output zero which is going to say hey we only want you to look at the first page it's one of those Canada Post invoices which sucks otherwise it's going to put an empty string and the idea is an empty string should just be nothing um if that's not nothing then we may have to use a different character here sometimes different apis treat these things differently uh but that should be okay uh for our purposes and then I'm still going to do the Json output export and we're going to give all this a try and then um we're going to feed into open AI we're going to parse it and then we're going to update a row what I'm going to do first is I'm just going to test this on these two I just want to make sure that the parsing is good before I send everything over to open AI it's sort of a good habit to get in or sorry it's a good um thing to get in the habit of doing and you see here that this pars two um two rows sorry one row uh so I just got my old sample one so I'm just going to do two two here and I'll see how a batch job looks like when we sent two doesn't look like there was any more time that it took which is pretty cool okay so looks like this um parsed out all the text basically instantly um I thought maybe there would be some issue with this but doesn't look like there was any issue at all and just because I'm testing this I'm I'm just going to cut this down here and then I'm going to put this up again just so we can just get one result on this next test I don't actually want to run this twice I'm very um careful with the credit usage clearly and then I'm going to feed this into open Ai and let's just Parts the Json before we update the row let's just do this incrementally at every step okay all right so the total is here line items are here from address is here um client one client two sample business name address one Barry Ontario l4m uh 3c2 let me just double check that this is correct what I would like is I would like this however probably because this is client one client two sample business name it's probably just pulling that I wish we could have like an actual address that would be so much simpler yeah so the reason why it's pulling this is because it's attempting to um it just doesn't realize that this is like an example that this system put in uh on a real invoices you presumably would have an actual client name here right hence why it's pulling that uh that in is the from address as opposed to this this should kind of be the the from address but say Levy that's all right we're just using an example any invoice that you guys use on your own end is not going to include that I would imagine okay there is another issue here the issue is this text is like formatted quite perfectly so we actually have a dollar sign here um we have a dollar sign here too that I'm thinking so there are a couple things we could do you could go on the Google sheet side of things and then you could change the formatting of the text so that it's like dollar and then um this won't be a big deal and then you can still perform like text operations on this um I think like multiplication operations and stuff I wonder if we do five if that's going to work no it's not going to work so actually one more thing we have to do is we have to strip out this dollar sign um and then we we'll just do the formatting on this end the reason why I want to strip out the dollar sign is because right now this is sort of a string I don't really actually want a string um we can output it from the service fine if it's a string uh but I I just want to like turn this into a um I want to turn this just into a number so instead of like dollar sign 17 comma 813 57 it's like 17813 .57 instead that's going to be pretty important because if you ever want to perform like text operator sorry numerical operations on this data later if you want to pull it back into make.com you're going to have to do this text formatting step anyway so why don't we just solve that problem right now I'm going to say set multiple variables I'll drag that here and then what I'm going to do is I'll say strip dollar let's say strip total okay what we have to do is we have to feed in the output that we get from the Json module and then we have to um basically remove the dollar sign and there are I believe a couple of ways to do this I don't actually remember all of them I think what we could probably do is format number maybe parse number so we might actually just be able to use a parse number module feed in the total and then get the strip total I'm going to test this and I'm going to use this exact string on my test just to verify that this works so let's say we got this the number is perfect that's exactly what I wanted fantastic so what we can do then is on the last module instead of feeding the total from here we'll feed in the total from here which will include that strip total which would be much easier for us all right great so that looks good uh why don't we go and run this from end to end now on the Canada Post data let's go back here and you see that we now have the data formatted but you see how it's on the right the reason why it's on the right is because it's actually like formatted it's just a number and then we the Google Sheets if we go to form format and then go number and then um US dollar Google Sheets is doing the formatting for us so now we could theoretically multiply this by five or 10 or 100 we can't do that here because this is still technically you know a piece of uh piece of text it's a string and now what I'll do sort of as my final Act is I'll go to maximum number of return rows we'll go two and then we're going to run it on both of these puppies that was number one and then looks like gp4 is going to be the vast majority of the the time usage here but that's okay and then you'll see now that um the text is is formatted or sorry the total is now formatted as a number which is nice and then we have all the line items from the invoices that we received previously now whether or not this is particularly interpretable I don't know but as I mentioned to you previously you can uh you can train the model pretty easily to deliver you the exact format that you want you just need to provide a couple of examples of of before and after and if you're not how to do that you can watch any one of my videos on AI text processing I probably have like 15 of them up um but yeah that's more or less it what I'll do now is I'm just going to play around a little bit with the various formats of um invoice to PDF just so you guys can see what's going on how this works um remember how previously there was PDF to text simple no layout fast what I want to do now is I want to show you guys PDF to text which will include the layout which is going to be really interesting so what I'm going to do is I'm going to unlink this and then we'll go one returned row I'm going to run this puppy this took a lot longer and it used more credits but you can see now that this text is actually laid out just like a real invoice is and you know now that I think about it this might actually help AI determine what's what so maybe this is what you feed into AI if you're comfortable spending a little bit more time and credits um making something happen presumably the formatting of this text like the words and the the the sentences and the words that are next to each other are probably more relevant to each other they're sort of conceptually categorized as a group so maybe the model would do better um with that than with the previous flow so that was PDF to text we could do PDF to Json 2 let's give that a try this is probably going to use some internal model to format stuff as well I don't know so we go documents um and now everything's fed um it's sort of like encapsulated within this Json so this page count page index whatever and then objects value we've essentially broken all of this stuff down into um what looks like uh tabular data where we have the customer number invoice number um customer reference number customer reference 2 right I'm just going to open up one of these in the middle here that's probably more recognizable let me see here okay we have some prices here which is nice so if I go before you'll see this Json is now formatted so basically what what it did is is it took this whole thing and then tried to convert this into like an a key and then a value key and then a value key and then a value and the idea here is that data may be easier for you to parse procedurally if you get the exact same invoice every single time or exact same format every single time um you could easily just say oh okay every time we run this I want I always want row 20 column 2 or something then you don't really have to use AI to do the formatting for you and then um there's CSV as well CSV will be pretty similar it's just going to turn all of this stuff into comma separated value which you can then upload to Google Sheets directly um but just keep in mind that like you know this may not necessarily be more parsable basically we've just added a bunch of random spaces wherever their proprietary process sees fit which is pretty interesting um but you know you might be able to feed this into a Google sheet and then see where that goes why don't I try down here so if I try and separate this then I get this and then I think if I go to data oh split text in the columns here then now I sort of have it all in like column format granted this is still not super meaningful just keep in mind that you know PDF um parsers and processors they all process things using their own Super proprietary method and uh whether or not that method actually translates to like good csvs like if it's too good to be true probably is it's not going to like turn this into a perfect Google sheet every time and yeah there are a variety of other formats down here as well I also mentioned that um right now we're importing the file from the URL but a lot of the time you're going to be getting data directly from like an email or something right so how exactly do you do that how do you get data from an email um if you know instead of it being like this instead of it being a URL it's instead like like a file that you have to download or something well I'll show you how um the first step is we're going to get this URL from the Google Sheets output right so what I'm going to do next is I'm going to go http and then I will get a file from a given URL I'm going to feed in the URL of the invoice here I'm going to unlink this so I'm only testing these two then I'm going to click run ones and what I ended up getting is I ended up getting a bunch of data at this buffer code page binary format which may not mean anything to you but all this really is is this is just a translation of this from English into binary basically like a machine language uh that machine parsers will use to rebuild a p PF on the back end and now anytime you have a download a file module as a previous step in a flow and then a subsequent step requires like a file to be uploaded you can just um automatically map it it'll automatically find the downloaded file module for you in that format if you go to map here you'll see what it's really doing it's feeding in the binary data here as the data field and the output file name here is the uh is the file name but this will work exactly the same as what we were doing previously so I'm just going to go to PDF to text no layout fast then sorry I have to do this twice because sometimes I'm a little too quick for make and then essentially what we've done is we' fed the exact same data in it's just instead of pulling it from a um file URL uh we're actually doing the downloading ourselves and maybe this is good because maybe then you can do something else with this I don't know maybe you can store this data somewhere or maybe you can download a file onto your server right they variety of other flows that you can use but for the purposes of this demonstration I'm just going to go back to importing a file from the URL and selecting this source and then I'm going to download this as the blueprint for you guys so that anybody that watches this that wants to use this as sort of like their inspiration uh can do so pretty easily so let's just export this blueprint awesome and we have it now cool well I hope you guys enjoyed that video had a lot of fun putting it together this is how to essentially take a PDF um any sort of PDF whether it's an invoice or a resume or I don't know even like a white paper and then turn it into structured data using a p F parser and make.com oh and artificial intelligence don't forget the like Galaxy siiz brains that are humming on servers 24/7 that make all this possible if you guys have any questions uh feel free to leave them down below as a comment otherwise like subscribe do all that fun stuff and I'll catch you on the next video thanks