# The N8N Instagram Parasite System (10K Followers In 15 Days)

## Метаданные

- **Канал:** Nick Saraev
- **YouTube:** https://www.youtube.com/watch?v=9zBtU1mwOR4
- **Дата:** 11.02.2025
- **Длительность:** 1:19:48
- **Просмотры:** 84,435
- **Источник:** https://ekstraktznaniy.ru/video/12629

## Описание

Join Maker School & get automation customer #1 + all my templates ⤵️
https://www.skool.com/makerschool/about?ref=e525fc95e7c346999dcec8e0e870e55d

Want to work with my team, automate your business, & scale? ⤵️
https://cal.com/team/leftclick/discovery?source=youtube

Watch me build my $300K/mo business live with daily videos + strategy ⤵️
https://www.youtube.com/@nicksaraevdaily

Get the template ⤵️
https://leftclicker.gumroad.com/l/mcpkm

Summary ⤵️
I grew my Instagram profile from 0 to 10K followers in 15 days. How? I used this N8N scraping system to source tools from people in adjacent niches, perform automated research on those tools, and then generate improved, customized versions of scripts that I fed into a teleprompter. This is a step-by-step guide on how I did it.

My software, tools, & deals (some give me kickbacks—thank you!)
🚀 Instantly: https://link.nicksaraev.com/instantly-short
📧 Anymailfinder: https://link.nicksaraev.com/amf-short
🤖 Apify: https://console.apify.com/sign-

## Транскрипт

### Introduction to Instagram Scraper []

hey Nick here today I'm going to show you how to build an endtoend Instagram parasite scraper in n8n that subscribes to channels that you're interested in watches them for new reels takes those reels and downloads them exports the transcript takes that information pumps it into chat GPT and perplexity to tell you new angles and twists you can make on that content before finally rewriting a script that you can use to record your own videos in a tenth of the time I just use this exact approach to scale my Instagram channel from 0 to 10,000 th000 followers in 15 days and I just wanted to give you guys the sauce show you the behind the scenes and let you build it for yourself so if that sounds like something you're interested in stay

### Building the System [0:38]

tuned let's get into the video so just before I I want to be clear I haven't actually built the system yet it is a system I was planning on building and I figured I'd just record a video while I did it and narrate my thoughts out loud um but because I haven't built it yet I'm going to keep in the circuitous sort of detours that I take while doing the building I'm going to leave in any dead ends that I come across because to me it's very important for people that watch videos like this to see the live build process not just some glamorized super pristine sexy version of it but like how somebody that works in nen actually goes through step by-step debugging problems and that sort of stuff because otherwise it's like if I were to give you a picture of I don't know the bur Khalifa or something and say hey pretty building right okay great now you go and build it it's like you know picture only tells you so much right looking at things from the outside in only gives you so much of the

### Context of My Instagram Channel [1:26]

whole story so anyway let's back up a little bit let's talk about content text and then I'm actually going to go through the building um CU you know I know you guys just want to see that beautiful anaden canvas uh but essentially I have an Instagram channel it's called Nick Sur launched it about a week and a half ago no two weeks ago maybe something like that maybe a little bit longer and essentially we just do AI news so we're subscribed to tons of AI newsletters uh you know AI research papers AI research labs and the second that this new accessible technology comes out you know I record a video using a little teleprompter that I just set up actually which is very handy um although you know given how much I Yap I'm definitely not going to be using it for these sorts of videos and then you know I I take these scripts I record them and then voila you know we're basically one of the first people on the scene with this and uh people love it right like check out some of the reals they've you know this one just got 43,000 today uh this is 160k right this one's almost 400k so this is obviously like a very profitable area for me to be in and you know the bottleneck for me is like producing this content getting access to this news so one thing that me and the team were thinking about is okay well if we're subscribe to all these newsletters why don't we go one step further and why don't we get other creators in our Niche maybe smaller maybe bigger that are talking about related topics they might not necessarily be me an AI Channel but it might be like an Art and Design Channel talking about new Art and Design things for instance or I don't know some programming Channel talking about programming things and essentially why don't we watch those channels see the new outputs that come like U like track their reels scrape them on a daily basis and then if these things have new technologies that are related to artificial intelligence well great we're going to take them we're going to add them to a database then we're actually going to use AI to rewrite the AI portion of the script of their reals and then we're going to add a couple of additional layers to it so that it gets even better and you know more geared

### Quality Content Repurposing [3:16]

towards an AI audience this is what I'd call like quality content repurposing quality um I don't know if you want to call it parasite or whatever there's probably a ton of different terms for it but essentially this just solves a big bottleneck in my business so that's the idea behind this okay we're going to be taking like other people's Instagram posts basically so um scrape Instagram posts and I'm just going to get a giant list of like the best people in their respective Fields maybe they'll be like art I don't know design I do like programming I'll probably do um like business you know as we know these purpose built AI models for all of these niches like these Niche specific ones are coming out every day I mean it's so hard to keep track of them so I'm going to like get people that are number one in all of these fields and then I'm going to scrape their IG reels and I might do YouTube and newsletters and stuff like that as well and then once I've scraped these IG reels I'm just going to check if it exists in a database this is what I'm thinking anyway now when I say database don't get freaked out a database is just a Google sheet you know it's a Google sheet it's an Excel uh file right there's nothing fancy about a database is just the way the programmers try and make themselves feel good but it's just like a table okay with some sort of heading and then some sort of value and then usually some sort of like index or something right so we're all used to this we've all been using this we've grown up using this sort of stuff so basically I'm just going to check to see hey does this real already exist in my data base if so I don't need to do anything and then we kind of have like two options right the first option is uh yes it exists so if it does exist we do nothing why because there's just no use running all this infrastructure like twice but if it doesn't exist if it's new basically then we go through um the next step and I don't really know what the next step's going to be yet but it's probably going to be like transcribing the real realistically um I don't know how we're going to do it there are bunch of tools that like enable us to grab files and then do transcriptions and stuff but I'm going to play around with it live and show you guys what the build process looks like when you don't know what tool you're going to use and then once I've transcribed The Reel I'm going to um you know probably use AI to do some filtering just make sure like it's something that I'm actually interested in right I'm probably going to be doing some top level stuff like determine if it has the term AI in the transcript or little stuff like that before I do this and then finally I'm going to run through chat GPT basically that's not how you spell chat GPT Gat GPT uh run through chat GPT and we're going to identify angles or I don't know added value let's just leave it at that added value ways to make the content better and you know like I have a feeling that my content is going to be better anyway um simply inherently due to like the quality of the presentation and probably the editing because you know I just started working with a really good editor and all that stuff so I have a feeling that literally if I weren to do anything my content would still be better um this might just be me slapping my dick on the table being like I'm awesome and hopefully it doesn't come across that way but I don't just want to do that I actually want it to be even higher quality I wanted to like I want to do some research with perplexity or uh something before this step and then once I'm done that I'm going to rewrite as a new

### Scraping Instagram Reels [6:33]

script and then I can just like upload it to a teleprompter or something whatever I mean that's just me and my flow so um this is what the flow is going to look like logically let's actually go and turn it into an n8n thing so how the hell are we going to scrape Instagram reals first and foremost now if you've never done any sort of Instagram scraping before the number one tool that people use for this purpose is called apify if you're new to my channel might be the first time you've heard of if you're not you're probably sick uh bored to death of it by now I uh I unfortunately can no longer say that I'm not affiliated with appify because I've talked about how great their service is so much they've just given me and all of my audience members a 30% discount for uh three months so if you guys want to sign up to fi feel free but anyway basically what it is it's like um it's like a marketplace where you can get any scraper that you want for basically any service so you

### Using Appify for Scraping [7:23]

know you can scrape websites Instagram Google Maps contact details Tik Tok Facebook cheerio web scraper Puppeteer Apollo LinkedIn right we obviously want Instagram but in to make a long story short I don't just want Instagram I want Instagram reals so I'm going to go Instagram reals over here okay now we got a ton of different options we could download them using this service we could uh scrape this in profiles easily I don't know I typically gravitate to the one that's made by appify so if it has a little appify logo here you know odds are they maintain it it's probably pretty good so what I'm going to do first is I'm just going to check to see if I could do what I want to do inside of um apify then assuming that I can do it inside of appify then I'm going to think okay great now that we've like accomplished this Mission uh you know how do we bring it into n and actually like proceed with it how do we actually take it into our no code tool because right now it's on the appify website right so first things first it gives you a bunch of prompts and information here what is your Instagram username um I've already typed in Nick sarf maximum reels per profile I set it to five just now I don't really know if like five is enough um like logically just thinking out loud do people really post more than five Instagram reels a day probably not I'm sure the algorithm would punish you or something so yeah let's just not do that um what I'm going to do instead I'm just going to run it so right now it's actually starting a crawler in the bottom left hand corner of my screen you see memory just went up to 1 Gigabyte that's like um you know this is like a paid service in my case uh I don't know it's $2 per thousand reels I scrape so probably spending like 10 cents a day or something but uh we're actually like renting out a server somewhere in order to do this and great you know we just did this in 14 seconds we scraped five entries that's making good time you know if I follow 10 people it'll probably take like I don't know like a minute and a half or something uh and we have all the data so I have one of my Instagram reals here how to get infinite content ideas use a YouTube comment scraper um how I made a million dollars with automation take my community programs and okay I'm not just going to make this an ad um yeah so I have all the data that I need here what do I do with it well uh I mean we need to find a way to stick it into nadn right so the good news is appify has a bunch of very good documentation so I'm going to go appify synchronous actor run I think that's what it's called and we're going to want to do this uh basically what this allows us to do is it allows us to run an actor an appify server called an actor when you trigger it but we can run one of their actors just using naden doing an API call so that's what I'm going to do um I'm just going to check to see if I still have this pasted no I don't so I need to go grab my API key the cool thing about apify is you just make your own API key for whatever purpose then you can just like delete or retroactively um uh like take it out and it makes it really easy to get up and running with their API so what I'm going to do is I'm going to paste in my Bearer token um and then it's going to ask me in their API docs on the right hand side they have this sort of like request Builder which is really handy uh you know what exactly am I going to feed in as an actor ID well usually what happens if I just go back here usually the actor ID is just whatever's up um in the URL so I don't know for sure but we're just going to give it a go I'm going to paste that in and looks like we need some Fields honestly um I don't know exactly what we're going to provide as a field okay so obviously we need actor ID fine we need body is just an object yeah crap I don't really know what I'm going to feed in there not really sure um I think I need to grab the input right so what we're going to have to do is go here go Jason okay here we go and then this is going to be the input so I'm just going to copy this over and I'm going to go back to their API documentation and I'm just going to feed in this and is you know as you see if I send this API request let's see if there's an error or something as you know we just probably started something right I think so this may be running I'm just going to go back here and check yeah yeah okay so this actually worked okay perfect so um I just verified that it worked using appi's own little like API Explorer thing where you put in your API token you put in the actor ID um input Json and then voila so now that I have that what I'm actually going to do is I'm just going to copy all of the curl just using the little copy button go back to n8n and I'm going to go HTTP request and I'm going to import curl and just paste in all of this code okay so as we see here what we've done is we've just reconstructed the curl request curl being um just a library that people typically use to send request just in their terminal we've taken all that data and we've just copy and paste it and built our own request here the only thing that we're going to need is like we we're hardcoding an actor ID and that's fine but the only thing we're going to need is we're going to need our API token so I'm just going to go back to the account I'm going to copy my API token come back here and then paste it beautiful and we should now be able to run everything inside of NAD um let me just wait to see if that other

### Setting Up N8N [12:24]

actor is now completed good because I don't want to double up on it I'm going to go back and then give this a quick test so crossing our fingers apis are always fun to work with but if there are any bugs then they'll actually be kind of exciting because then we get to work with them together like friends anyway uh let's go back to appify before somebody calls me a loser steals my lunch money cool it looks like it worked awesome succeeded awesome and it looks like uh looks like it succeeded but we didn't get any anything which is kind of weird so I don't actually like this I would prefer to get some data back right so let me take a quick Peak here um just want to make 100% sure that there's nothing weird going on because like we should get some sort of response right data I don't really know why we're not looks like we had a 2011 400 or 408 okay actually that's kind of annoying um I guess we don't actually get the do we not get the data oh you know what I'm using the wrong endpoint okay I take that back don't do it like me actually do this was actually going to return all of a data for you for whatever reason I guess they have two endpoints one called run actor synchronously the other run actor synchronously and get data set items how the heck was I supposed to know okay so I'm just going to copy this over and um paste this in and then grab the ID and then we're going to redo it okay let's see if that works or see if I need to de feed in some stuff that I haven't already done I think that everything is the same here so if I go back yeah okay cool we're running it now so let's see if we can actually get the data inside of apify let this be a lesson to always check the API request or the endpoint that you are sending to um sometimes certain endpoints are better than other endpoints in this case this will combine a two-step process and turn it into a one-step process for me instead of just firing off the actor and having it run and then having to pull it later to see if it's done um I'm going to be able to do it two in one basically going to be running it and then there's some built-in trigger flow or whatever where after you're done running it it'll send the results back to me that's the idea anyway okay so it's five of six pages

### Creating the Instagram Database [14:40]

now beautiful we should be getting close awesome and it worked beautiful we got all the data inside of NN fantastic that's cool look at that okay so what I'm going to do is I'm just going to pin this so this is now pinned inside of nadn that just means I don't have to request the data in order to do something with it um for now we're just going to do the when clicking test workflow uh for organizational purposes I'm just going to rename this to uh just run actor synchronously just so I kind of know what's going on from a glance and then if you think about it what do we do next well we just have to build our database and we have to like cross reference against it so um I have a little Google sheet over here that I was just putting together before I was like oh man I should probably record a video on this I'm doing this anyway it's called Instagram real database right pretty straightforward pretty simple U what I'm going to do is I'm just going to get all the header fields that I want put them in here and then once I have them um we can actually go through and do something with this so what header Fields do I want input URL do I need that no I don't really need that let go Jason ID I definitely need an ID so this is how we're going to cross reference and check to see whether or not like the data that we are sending is already in the database type video do I need that really if I'm scraping reals I probably don't right short code I don't know what short code is might be some other ID so I'm GNA put that in there caption for sure I'm gonna want the caption hashtags um yeah sure why don't we just like get the whole let's just combine all of the hashtags and then um we'll put like a comma and then um that way like instead of it being an array we're just going to have one mentions I'm not going to do mentions URL that's good comments count that's good first comment I like that uh latest comments this looks really big I don't really want this so no get the hell out of here height and width do I really care about this no display URL probably I think that's like the thumbnail leave that in there video URL for sure this is probably where going what we're going to use to generate the transcript alt likes count definitely uh video view count so 57,000 oo video play count wait what's the difference between video views and video plays H that's weird um time stamp you know what we should do now that I'm thinking about it we should probably organize these based off time stamp yeah like how re recently they came in would be good what's weird is this isn't actually a time stamp like that's not what you call a time stamp is usually a um like a Unix timestamp and that's usually this so they're apparently just putting in a date and calling it a amp okay whatever um owner full name so owner username so I'll probably grab like username sure that's fine um product type Clips video duration that seems like it might be interesting video duration 36. 7 seconds cool is sponsored false yeah I'm probably just going to filter out everything well if I'm going to filter everything that's a sponsored then I don't really need it do I music info do I really need that no okay great so honestly we have everything that we need for our database now just going to that is our database this beautiful lovely Google sheet um what I'm going to do is I was just like to like have the header be a different color so I'm going to make this nice pastel blue and the text will be dark blue perfect and then I'm going to change the font to enter because I'm just a giant slot for enter okay um perfect we now have everything that we need in order to like add data to it oh why don't I just change the name from sheet one to reals or something um that'll be easier so if you think about it um okay we should also actually add a transcript and then on top of that uh maybe we do scraped trans transcript on top of that we're going to

### Adding Data to the Database [18:08]

do like a new transcript yeah that's going to be like mine new transcript and then we can stick that all together beautiful so okay now we have basically everything we need right fantastic um we can move on to the next step which is verifying whether or not this data exists in the database so logically I kind of just like in order to test this I need to dump a few records into the database that's what we're going to do first then once we've dumped the records into the database I'm going to delete a couple and then I'm going to see if we can like incorporate the logic that um sorts them out so the first thing I'm going to do is I go sheets just going to go add rows so a pendro I guess um I've already done this connection if you haven't you basically have to sign in with Google if you have a cloud end account like I do or you have to go through like connecting uh I don't think you have to do a service account but you have to go through like adding ooth which is you have to go into your um Google Cloud console account I think it's back here probably and then you have to sign in of course because my organization wants to make sure it's really me um you know it's not just a robot pretending to be me like so many people on YouTube apparently think and what you want to do is uh just go to like your homepage here on the left hand side go to apis and services go down to Sheets right over here from Google Sheets uh there'll be a credentials Tab and this credentials tab will um allow you to create an ooth 2. 0 client so you actually to create credentials go through the rig Ro of creating an ooth client ID it's just going to ask you some simple questions just call like an in end project or something and then um you're going to have a client ID client date you'll be able to generate client secrets and stuff like that you can copy and paste the new ones anyway and then you can put all that information in there but yeah for our purposes we don't need any of that cuz I've already done the connection multiple times and I'm on cloud so I just need to click that one button now what I want to do is the resource I'm going to be targeting is a sheet within a document operation of pendo what I want is I want that Instagram real database and it looks like the newest file just pops up at the top which is great the sheet that I want is I just want reals okay great so now I have the ability to uh map each column manually so I can map each call I guess what I can do also though like if you think about it is I've already gone through all of the work um naming it the same thing here right so can I do this hashtags and array you know what I screw it I might as well try it the thing is though I don't like that there are all five items here so what I'm going to do is I'm just going to add a limiter the limit is just going to take it so that I can only feed in one item at a time and this is going to do a couple things number one with n8n unfortunately the Google Sheets um query is the Google Sheets query is typically array limited pretty fast so they just have really crappy like rate limiting or they're making a lot of requests or whatever the hell's going on I don't know but uh we're just going to fix that so we're going to make it a lot slower of a request or an engine I should say and we're only going to do one at a time during testing just so that I don't like ramp up my API usage okay let's do test step and let's see what happens just ran test step it's executing okay great seems like it added so let's go back here and did we add everything I think we added most things yeah hashtags is an array so there's that but that's okay I'll just leave it as an array it'll be easier to do work with it later uh you know what it's just it's adding too much data I don't really want all this extra data man I do not want the extra data man all right you know I'm just going to map it manually screw it not worth I don't want an ugly database I want it beautiful and young and fresh okay so uh we're gonna we're just going to map this so ID is just going to be here right so I'm just going to grag it drop it voila you can just map directly from the Json I'm a big fan of Json obviously uh okay we need the time stamp so where the hell is that time stamp let's go back here timestamp come to me time stamp short code uh caption we got to do the hashtags as well the cool thing is since I'm doing the hashtags I can now like join the Hashtags with a comma and then a space right so that'll kind of get it in this format do I need to no I don't really think I need to honestly uh might be a little bit prettier but does that really matter no okay the URL just going to pop that in there comments count right that's over here first comment I don't know I just like the first comment because you know that there's going to be like a man Nick really fell off for his comment soon as these are wants to do uh video URL likes count where is likes count video view count you know the biggest quality of life upgrade to n and end would be if they didn't make their uh drag and drop thing hide the subsequent field like it does here because it just hides the subsequent field like they should pop down this like an extra inch or something so I don't cover my own the thing I'm trying to do next logically right but what are you going to do um just get a bunch of big brain Geniuses together they don't really care about stuff like this okay so scrape transcript new transcript yeah that's it uh and then I'm just going to delete all the rest of these so they don't dump in uh no and then just get rid of all these and then yeah okay we should be good did I delete everything yeah I did okay let's test this again so now I'm adding them in we should just blast every row here except for scraped and new transcript now because that I'm going to widen this a little bit this looks good I don't like the wrapping thing so I'm actually just going to turn off wrapping um yeah no wrapping and then I'm just going to make all these way smaller I'm just going to grab this and then go like that make mebe make it a little bit bigger that way um when I add a new row it's just all going to be on one line instead of like a 100 I just don't really like the 100 line thing hashtags obvious ID okay time stamp so we're going to sort probably by timestamp and then like we're probably going to use this as the primary key URL of the post is here comment count fine okay display URL uh this is really long I don't like that let's make that way shorter it's important to me to be able to see everything while I'm doing testing so this one step that might you might think that I'm just being weird and unnecessarily I don't know what you want to call it OCD about this or like detail oriented or missing the big picture but it makes a big picture makes a big difference when you actually go through and you do the testing uh okay so this is pretty good obviously we're just going to make this a little bit smaller so I can see the scrape transcript coming

### Verifying Database Entries [24:50]

when it does so now we basically have the database right so if you think about it I've just added one entry to the database so what I want now is I want logic I'm going to pass two through the limiter and then I want logic to see hey is this already present if this is already in the database then you know I don't want to add it but if it's not then I do want to add it so how do you do that logic and edit end you know it's interesting a lot of people um you know I kind of come from like a make. com background right it's a similar no code tool but um a lot of people don't really know how to do this logic in N which is interesting because to me it's just like the foundational logic behind everything so I'm just going to change Max items to two so now we can pass an extra entry through so we could like check to see whether or not um more than one is in and if you think about it I just want to add it in right what I want to do is something that looks up in my database to see whether or not that entry exists I'm going to go get rows and Sheet I'm going to connect to my other credential and the database which was Instagram real and then sheet again was just reals and now I need to add a filter so this is a part that I think a lot of people Miss but essentially what you can do is you can add a filter that searches to see if one of your columns is equal to something so if you think about it what I want to check to see if ID is equal to the ID of the entry that's coming in right if it is equal then I'm going to find an entry basically then if it's not equal then I'm not going to find the entry so that's pretty cool right uh I have a bunch of options here data location on sheet output formatting return only first matching okay so I'm just going to leave it there and then basically what I want is if I don't find anything I'm going to have it output the data regardless this is very important if you miss this step it's our logic isn't going to work then what I want is I kind of want like an if and then right and then I want to say like I have to execute the previous node but logically what I want is you know if uh if this ends up being empty then I'll proceed down the true route and then if it doesn't end up being empty if there is an entry I'll just stop right here and like there's a do nothing operation which is just kind of Handy to logically see what you're doing so in my case I'm doing nothing and I'm just going to like leave the do nothing no down here just so it's blatantly obvious that like okay well we're just stopping um if we don't find anything so this is where we're at right now um we are going to have to add a little bit more complexity to this when I run this at scale um simply because as I mentioned earlier um naden a lot of the time struggles with like doing these bulk um API requests to Google Sheets and like you can technically increase your um your quota limit and stuff like that uh I just I personally haven't done it and I think the vast majority of people just don't ever do it so um just to make our lives a little bit easier we are going to be passing this through what's called like um a loop the loop is basically going to slowly do it one at a time instead of you know try and do 100 entries at once like idn likes to do but yeah anyway we're at this Google sheet step let's see what happens um we're feeding in two entries logically right the first is already in the database the second is not in the database so let's see what happens I'm going to feed in two and we're going to see um what the output is on both of these runs so test step it's finding and it only ran uh once or rather it ran twice once for each input item but it only found one entry here okay so we're returning that one entry and uh we're only going to be proceeding with that one entry I guess so we found one item basically we fed in two items we got one so um yeah so now that we have this one item what we're going to be doing is checking sorry it found the item that was in my database that's what I meant to say so now that we found the item that's in the database uh there are a variety of ways that we could do this like what I could do is I could take ID and I could say you know I believe I could say exists or is empty or something actually let me just go Jason if Jason exists then I don't want to proceed with the flow logically right let's just run this through so Jason does exist which is true so what we want is we want doesn't exist Jason doesn't exist which is false then it'll end right here you can just give this a little test step a run even though it does nothing awesome okay great so now what we need to do is we need to somehow sort out the true Branch Okay then if it's true then we need to append the new record to the database so just thinking I'm not seeing I'm not actually under fully understanding why it's not running um this should run uh twice should have this return to row one should be found and then the other should not be found cuz this is running twice so let me see okay you know what I'm going to do instead some pretty interesting merge logic is going to solve this problem and also just make it easier for us um we're going to grab the input and then we're going to create an output the output is going to be the elements that are new so let's go here I'm going to go merge and what I want is I want combine fields to match I want IDs and then I want to keep non-m matches so what I'm have to do is I'm going to feed in a merge and if you're unfamiliar with how this works then don't sweat show you how this works in a moment let's just do this so it's kind of visible basically what I'm going to do is I'm going to feed in this Google Sheets output and I'm going to feed in the original input and it's going to remove the element in the original input that was found inside of the Google sheet and that way the output of the merge node is only going to be the data that I want to add so let's run this let's see we have one item the item is this ID 356 38 let me just check my Google sheet that is a different ID beautiful it's comment speech not comment roses right we feed in comment roses we get comment speech beautiful um now as an output I don't believe I actually need this I lied and we can just go directly down the Google Sheets and we can actually grab the data from the merge node instead which should just be the same honestly and now we should be good and I think I've avoided that if then logic which is nice because if then logic can just be kind of annoying so let's give this a try now what should happen start to finish so we should do get the five and then we should only add one yeah there you go so we only added one right so as a sanity check right what I want to do now is I want to run this one more time with both of the elements then in my database as well as the same elements here because I just want to double check that this is going to work um expected behavior is we're going to get two items to come in from the limit right we're going to read the Google sheet and we're going to get zero items coming out those zero items and these two items uh sorry we're going to get two items coming out the two items and the two items are both going to subtract and we're going to be at zero that's my expected behavior let's see what happens perfect that's what happened we got two items here nothing carried forward so our system is now sort of like self annealing basically a fancy word for saying it's just like regulates itself you know it's uh elegant at least more elegant than what I was trying to do before all right so now that we have that uh what we need to do is we need to add the new entries as we've already done once we've added the new entries you kind of think of it like okay great so we have all of the front part of the flow done what we need to do now is we need to actually go through and find a way to download this thing and then turn that into a transcript uh once we have the transcript then we can do all that fun AI stuff so where we are right now is we're right over here we just figured out how to do this stuff so if I just get nice and green sorry we did this let's go here we did this we did this now basically we just have to figure out how we're going to transcribe the

### Transcribing Videos [32:49]

the real so um I think there variety of ways what I'm going to do first is I'm just going to see if I can get the data like I probably should have done this initially but I'm going to see if I can get the video and like play it hey there's a complete oh my God I'm loud okay thanks man uh okay great so I mean we can obviously get the video that's cool so could I just feed I could probably just feed this in a whisper couldn't I think I could uh whisper might need an MP3 whisper for those of you that don't know is just open AI um open ai's model which allows you to basically just like transcribe any text could we do MP4 could we do whisper MP4 let me check M we might be able to demux the audio the mp4 file okay interesting well let's just give this a try first so I'm going to go open AI first because I'm familiar with open a and I know how it works that's kind of what's going here um transcribe a recording uh limits the size of the audio file to 25 megabytes let me see how big the audio file or this file hey there's a completely new a all right relax let's download this see just how chunky it is 2. 1 megabytes so the MP4 is 2. 1 megab I'm probably going to be fine so let me grab my YouTube credential the resource is going to be audio I'm going to transcribe recording the input field name is data so it needs to be a binary file okay so I basically just need to download that file turn it into a binary file um so well I guess we have to download right can I just convert adjacent data to Binary data is that going to work I don't actually know I don't remember how to download a file and aen let me take a quick peek HTTP request to download file okay yeah we're just going to give it a download response format is going to be Tex so first of all we're just going to download this so what I basically need to feed in here if you think about it uh we don't have data accessible from our previous node during this test unfortunately it's really annoying so we actually need to execute them all again why don't I go back here and just delete one of these okay let's just execute all of the previous nodes all the way up until here apparently we're not getting anything which is annoying I'm not really sure why we are feeding in two uh is it the pinned output here H I think realistically what we should do is I should test this again step by step and then pin all the respon uh pin all the results they'll make it easier for me that way I just don't have to do all these dang API requests to um Google Sheets which I don't really like doing okay we're going to pin that output Let's uh run this we should get one now cool we got one pin Google Sheets now we can test this we're going to get one output here because we just ran it thank you okay great so now that we have this we can actually do the HP request to that data right so the URL is going to be video URL and get authentication no um can I just do like a file format here or I do need do I need to pass a header that says download a file I don't know let's just test this stab and see what happens we're calling the page okay we got data looks like it was a video to mp4 file so now I'm going to pin this as well uh I guess I can't do that with binary data H that's annoying contains a binary file to be processed uh well this says it's an mp4 I don't actually know if this is technically binary let's see what happens if I run this who knows we might get lucky oh okay great yeah no we don't actually need to do anything fantastic that's way easier than I thought it would be we just downloaded the file and we passed it through the open AI whisper node and it looks like it doesn't only work off audio it also works off MP4 probably has a built-in API end point to get the file name get the file extension and then if it's a video or MP4 it automatically does like the demox which is awesome okay great so yeah um we have the script right here which is nice what I'm going to do is let's just uh well we can't pin that but I think we should be able to pin this right cool let's pin this so now we have the transcription what we basically want to do is you know if we go back to our little drawing here um this is now done right so this is now good what we

### Filtering Content for Relevance [37:19]

basically want to do is we want to filter this so I'm going to write a really quick AI prompt that AI prompt job is going to be to go through grab the contents of the real and then I mean you know in my case it's obviously AI so but I'm going to make sure like hey is this an AI you know development is this something about AI or technology or automation maybe I'll go a little bit broader I'll say like hey you know is this a tool a technology or Ai and then if the answer to that any one of these questions is yes then proceed with the flow if not we'll just drop it off and stop okay uh yeah that probably makes the most sense and then um after that what we're going to do is we're going to go through and we're going to grab the tool and technology and then maybe we could figure out how to use the tool and Technology to do something cool maybe we get like a step by step on how to use this for some automation agency purpose or something I don't know I haven't really thought it out that far but maybe we'll just kind of do it organically as I'm putting things together all right so we've come a long way um one small step for Nick but one giant step for YouTube kind here I'm just going to save this really quickly and now I'm just going to take this over to an open Ai call again we're going to do message a model which is where I just talk to artificial intelligence um candidly I'll use the YouTube Feb for credential if you don't know how to connect with this you do have to go through some rig roll of like getting your API key which you can do so just by clicking more info down there what I'm going to want to use as a model is I'll probably use GPT 40 just because I think it's the smartest one for the lowest cost right now and I'm always going to add um at minimum like three prompts I'm going to add a system prompt then user prompt and another user prompt but I usually add more because I want to give it an example um but let me run you through this assistant prompt is just how it identifies you are a helpful intelligent I'm going to say admin assistant because I wanted to make some administrative decisions then under user um we're actually going to write our prompt now you know it's hard for me to like dictate what I'm thinking when I'm writing these prompts but generally speaking like some overarching rules of thumb um to write as simply as possible don't use unnecessary verbiage if you don't have to keep in mind that the models are smarter with fewer tokens and simpler instructions give uh as few conflicting pieces of data as possible uh generally just like pretend that it's a staff member and just talk to it and give it instructions um that you would give a staff member so like you know the way that I like to talk with staff members just like hey what is it that you're doing why is it important um what are some examples of like a good job done and a bad job done so uh let me just run through this and kind of see so your task so we are an AI and automation maybe I'll say I Ai and automation Instagram Channel I'm looking through News archives to find mentions of tools that I can repurpose for content purpose into content for my channel that seems pretty reasonable that's who I am let's now um give it some instructions and maybe some context so your task is to take as input a transcription of a news archive and then determine if it if the transcription is about a tool a technology or AI if so you'll um let me think you'll identify the tools that's a good thing to do write a list of stepbystep instructions to use the tool identify the tools then write a list of step-by-step instructions use the tool and then output everything uh let's not do that what do I want out of this I wanted to identify the tools used because I'm going to use it probably to do some searches later I also wanted to write list of step-by-step instructions to use the tool easily ideally for free or for low cost and then once I do this I'm going to do the generation later and one suggestion on how to make the content more palatable on exactly how to make this content more palatable to an AI automation audience let's do that sounds pretty interesting return your output in Json using this format now I'm going to give it a format the very first um element I want is I want tools and I just want like a do I want an array of tools no I don't I mean an array would be the best logically but I don't think I'm going to ah maybe I will list of tools let's do list of tools or resources okay and then why don't we do step by step we can't do that with Jason so step by step instructions detailed instructions on how to use the tools that looks good and then why don't we go suggestion Fair comprehensive let's go comprehensive in-depth so suggestions on how to make this content more interesting and palatable to my audience AI automation that sounds pretty cool okay let's give that a go um this looks pretty good to me we have basically everything we need um now if you did want to make this a little bit better you know obviously what you'd have to do is you'd have to give it examples so I may do this with a couple of reels um if I do I'm not going to do it on like the video I'm just going to really quickly whip it up and then show you guys after um if not you know we're going to see how it goes so I've written my first US my system prompt now my first user prompt I'm going to write my second user prompt and you think about it logically like I'm just going to feed this in so I'm going to say transcript and just make sure the words are the exact same because I want this to be as simple as possible for this model to understand um sorry I guess first we should determine um verdict true or false then I'm going to say if verdict is false and we're just going to do it in a string format which may be a little bit confusing to you if you're used to it in Boolean um but we're just going to say verdict is false leave the rest of the fields empty just for Simplicity okay great so transcript we're going to put the text right over here nice and then we're outputting it as Jason awesome any fields here anything I'm going to do no I don't think so let's test it let's run it now obviously our verdict should be true because I'm feeding in my own data which is all about you know some AI text speech generator right cool the verdict is true if I turn this into schema um I have ai oh well that's not what I want H I guess I'm not actually saying the name of the freaking thing right that's interesting I'm not even saying the name of the tool it's just a text to speech generator uh man that's wild um can I like just edit this and can I just go back to my real and see what tool I was using because I think it was Speech MO is it speech Mo yeah okay when don't I go it's called uh Speech Ma and it allows you okay great okay um that looks pretty good so now it's going to say it's called Speech Mouse so now when I actually run this we should extract the tool it's fine I didn't actually mention the tool name in my real maybe at some level I was like hm I wonder if anybody else is going to scrape me okay great now we have the step by step visit the speech mod website enter your P paste or your text in the Brad text box foreign voices cool this looks pretty good voice comparison test using speech mod versus other proper tool that's a great idea yeah very cool okay I think this is great um awesome so right over here I'm just going to rename this to transcribe make it really simple over here I'm going to say filter content and generate suggestions uh it's a little long let's say filter and generate suggestions that looks cool um anything else I have to do here just to make my life a little bit easier so run actor synchronously limit um let's just do search for entries going to say drop duplicates over here we're going to do add entries um download video then we're going to transcribe filter and generate suggestions awesome um I'm now going to pin this output thank goodness for the ability to pin out puts man blow my mind um how so many no code other no code tools don't have this and now once I've done filtering and generating the suggestions I'm going to use one of my script templates or maybe I should H maybe you know what I should probably do I should check out perplexity is perplexity in here no maybe what I'm going to do is do an a perplexity HTTP call look up the service and then I'm going to have it scrape the website of the service and then tell me something really interesting about the service

### Using Perplexity for Insights [48:00]

that is not in the thing and then I can I'm going to add that to the script genius that's cool and then I get to show up perplexity too what a win Okay cool so check this out right so perplexity is right over here I'm going to do um what do you want to know uh like what is one interesting thing about speech Ma what are three interesting peculiar things about speech Ma the AI text to speech tool I mean this is pretty cool right three interesting features of speech mod Okay cool so uh you know what I'm going to do now is I'm going to go to perplexity I think I have it over here right no I don't I should be signed into perplexity honestly not really sure why I'm not so let me log in again perplexity is a tool that allows you to um not only do a search and talk with AI but can actually get data from Google basically I mean like it's kind of like Google um I'm just going to copy the link and move it back here just in case there's an issue with the authentication sign in link is oh no I'm signed in man I don't need that okay great um where do I get no I don't want to download this I want to get my API stuff do I have any money on this account I do I have $263 so I'm going to copy over this API key and I just need to figure out how to actually do an API request to perplexity right so let's go to Docs let's quick start I have a payment so that's fine can I just run this be really cool if I could run this okay make your API call perplexity model sonar Pro System be precise and good size how many stars there in a g I mean yeah that looks pretty good so I'm going to go over here um make another HTTP call this is going to be my perplexity search perplexity then I'm going to go over here double click this import curl paste in this new request then I'm going to have to feed in the API key that I just copied right so let's go back here copy this API key go back paste it in how many stars are there in our galaxy right now I'm going to test this sample API request um it looks like it's an invalid model maybe because there's a space there I'm going to get rid of that they should probably fix that in their API docs they haven't already Milky Way galaxy Okay cool so I have the content um so now what I'm going to do is what did I just ask perplexity a second ago I asked it tell me three interesting things peculiar things about speech ma the AI text to speech tool free resarch vast Library vers okay I mean like you know reasonable um obviously it's going to be different every time but let me just run this query now through AI so I'll say tell me three interesting things about speech ma let's give this a test let's see if it can do it first of all very cool okay sweet so now we have a perplexity output and the cool thing is this is going to combine with the generate suggestions now for us to rewrite our thing so let's pin the outputs there um by the way you can pin just by uh mousing over and then pressing P I've been doing it sort of the long way all around just cuz I want to show you guys but that's that easy um and then now let's go and feed this into another Ai call so I'm just going to copy and paste this then drag this puppy in instead of filter and generate suggestions now it's going to be rewrite script or maybe like write real um WR new script let's do that we're going to feed in the generated suggestions with the perplexity data and we're going to put stick all that stuff just in one like prompt and then we're going to train it and give it two or examples and we're ultimately going to have it um write me a script which then I can add down scrape transcript and then new transcript and we're going to take new transcript and then that's what I'm going to pump through my little um yeah teleprompter how cool so uh you're helpful intelligent let's do writing assistant now let's not do script writing I don't want it to get confused but you're a helpful intelligent assistant now I'm going to say I run an AI and automation Instagram Channel I'm looking to through news archives so I've already done that

### Writing a New Script [52:31]

my editors would like to make a new video they've got Source data a stepbystep guide they found Source they found um an interesting tool or technology compiled a step-by-step guide on how to use it done some searching on the internet about what's interesting about it to find um some interesting things about it and also um written suggestions for how to make the content better okay ready script about it ready let's say rough draft on it about it and we also had our edit our chief editor write suggestions concrete suggestions for how to make the content better your task is to take us input all of these things and then write a new highquality script for us to say tonight for us to feature tonight let's do that I mean you know none of this really matters so maybe I'll remove that I want to confuse it tonight are you a news channel okay great return your output okay return your output in Json using this format um what I'm going to do so I'm just going to call this script your script goes here and then if I just go back what's the length of the thing that I just made I go over here and I just go to word counter and then I just paste this in it's about 100 words right so what I'm going to do is I'll just say script goes here let's go around 100 words then for rules I'll say be concise uh use a casual Spartan tone of voice No Frills say follow the provided examples as close CL ly as possible follow provide examples closely as possible in tone of voice or um and subject matter and then I'll say end the script with a call Action like want the want thing just comment keyword and I'll send it straight to your DMs perfect man it'd be so cool if we could fully automate Instagram in so cool all right so listen why don't we just start by providing it all of the information and then just seeing how it performs on like a on one run and then after that we'll um we'll go through and we'll um actually like give it some training examples of stuff so if you think about it uh we're going to need perplexity output I'm going to feed in this and then I actually just want this to look pretty so I'm going to go to Json formatter and then I'm going to pop this in so I'm just going to do that and then in addition to perplexity output I also want or uh let's do rough draft script actually first maximize the quality of this rough draft script then perplexity output then okay maybe we'll actually go tool name rough draft script perplexity output step by step guide then suggestions for improvement so now I'm just going to map these variables so suggestion is going to go here step by step guide tool names uh I think I should just go tools. jooin so I guess if there's only one element we're just going to feed in thing but what if there's more you know why don't I just go join yeah I'm just going to use join that's probably the simplest way to do it uh rough draft script is going to be the output of the transcription node right Okay cool so now if you think about it we have everything we need to actually generate Json and then scrape it so I'm just going to see how it works on like this completely Bare Bones uh sample input and then we're going to give it a go um I'm look let's look it looks like I'm having an issue with item why is that transcribe video item Jason text I think the reason is um I can't figure out the matching item it's because open air return incorrect matching information okay well I'm just going to go first because it's just one here right it's always just going to be the same one no strange uh we should get that data me see let me delete this we'll go again and then what I want is I want um transcribe video then I want item first right or is it dot first um oh boy all DOT first do json. text should be good I'm going now to test this we should be feeding in the rough draft script correctly I hope so I don't like the I really don't like the script here guys check out speea it's kind of boring so we're definitely going to have to train it we're like show it our tone of voice the good news is um we have a database that uh you know is going to be pretty simple for me to use so why don't I give that a quick look I don't actually remember exactly where that is scripts voicemail scripts no hm this right over here okay yeah this is it okay um these look reasonable right let's do some my new ones so now what I'm going to do is I'm actually going to give it some examples so what I have to do is I have to come up with all this for like some example tools so uh let's add a couple of messages cuz I'm going to do two I'm going to do one example once I'm done with the example I'm going to give it an assistant so this will be the output here me copy all this I don't like why the formatting is like this it's very annoying we should be good here oh you know what no that's not what I want I just want one object oh I hope I didn't just delete a bunch of stuff I think I did yeah okay let's do this uh user um why am I having a stroke right now tool names down here okay so no we want this last one to be the expression with the actual data we want this to be an assistant prompt which will contain a single item called Script and then this will be an example so I'm actually just going to go over here and I'm going to strip all this have to write my own example um so that should be reasonable um not going to be too big of a deal but yeah anyway Let's uh let's feed it this as an example and let's just replace all new lines with actual new lines new Lin above with characters just have ai do it really quickly thank you o wonderful single new lines really I don't know about that man might need some doubles there okay we're going to copy it um I'm then going to generate this I'll leave it as fixed um for now oh we doubled up on the quotes let's just remove that looks like we use some weird quote character too so that looks pretty good um cool so I basically now need an example of 11 Labs so I a shitty script that we turn into a good script or like a subpar so I'm going to go to this expression editor and then tool name we'll call this speech cuz you know we didn't actually do it for this right so I need to just come up with this which is kind of annoying but um anyway tool names it'll just be speech ma rough draft script uh I guess what I could do so I could say rewrite this but make it worse that's actually pretty funny get long in Circus and Bland remove the 60 languages and the no restrictions part so that way um you know it's hopefully going to pattern match the fact that I want it to add more details um and you know maybe The Source content is GNA be really long and annoying okay so oh sorry that's not what I want I wanted um rough draft script to be that right so I got to go back here and I just got to copy all of this in rough draft script is going to be this it didn't really do that it has a lot of voices and the best part is you don't need to sign up for an account okay the process is simple go to the site and type whatever you want converted there are lots of voices cool so that's going to be a rough draft script um perplexity output will say what it actually said which is here are three interesting things about speech ma uh wait yes stepbystep guide will then just be this step bystep guide then we're going to have to you know feed it like a different data basically no sorry just going to want to hard code this in suggestions for improvement uh okay feed that in Okay cool so we now have if you think about it that was a lot uh we now have an example of an input that we're going to provide then we have an example of an output which is a tight script which you know I actually like we created before so I'm just feeding in something like I've actually said before and then we're going to feed in the actual data and then we're going to try and have it do something similar make sense I sure hope so the unfortunate thing is because I'm pinning all this data we're actually going to have to go and we're going to have to um redo it uh so why don't I just change the limit so that the max item is one but then we're getting the last items then I can just work my way through this whole thing so let's test the step it's okay we can overwrite the pin data cool so this is a different one humanizer prompt very cool test the step no put data returned because there are no entries in the sheet that have that we can out test the step now is going to add to our Google sheet if you think about it so I'm just going to go check out the Google sheet as it gets added very cool this one looks pretty short at all things considered we're then going to download the Json video URL cool um we're then going to go to transcribe video test this puppy it's going to pump that through whisper is then going to generate a transcript just like I did before beautiful o very cool uh it doesn't look like it got chat GPT though then we're going to filter and generate the suggestions that's okay you can unpin and test this is going to take a little bit longer because we're feeding you know fair amount of text and it's not just like what we were doing before chat PT doesn't know what chat GPT is it is not self-aware yet then um we will test this step on perplexity I'm I love doing this one by one because if you do all of this one by one you eliminate the possibility for many errors you basically hone in on where the error is if there does end up being an error okay great and now we have uh why is this still oh I think I hard cuted this in didn't I oh uh yeah let's do oh that's actually going to kind of suck man chat gbt humanizer prompt ah man this is such a shitty crappy example Sor I'm trying not to swear why don't I just like um pin this output and pretended something else okay um chat GPT chat uh no no chat GPT and what I'm going to say is um fool humanizer okay um the tool is just going to be called humanizer then we need to provide some context actually when I filter and generate suggestions we need to do one more thing search term a short search term we use to look up search prompt a short search prompt we'll use to look up the the service write it like um tool name the type of tool okay I mean that's not perfect but now when we test this we're going to overwrite the output but it's also going to have a tool a search prompt parameter which I should be able just to slot into perplexity chaty that's so funny that sucks man it's not supposed to be chaty PT anyway humanizer AI um Cutting Edge series of prompts that improves let's see um AI output generation okay cool now I'm going to save and pin this awesome so I shouldn't be able to search perplexity using if you think about it this so I can go expression tell me three interesting or peculiar things about search prompt which is going to say humanize a cutting at series of prompts that improves that improve AI output generation cool we're now going to search perplexity they can't find anything it'll just say I'm sorry I can't find anything so we've sort of like built that in right meta prompting techniques very cool awesome and then now we're going to write a new script on this with all of the stuff we had before including the example oh man I just hate that it's chat PT that sucks it really undermines my example here uh I should change it from chat GPT to chat GPT chat GPT chat GPT cool there you go now when I search this uh I don't actually think I need to do I now I'm going to search this oh geez did I feed it in as an input like way back here filter and generate suggestions tools with chat gbt this look pretty good from what I remember anyway if we pin this now and we write the new script is it really going like where is it going to get the chat CH PT from if I go to filter and generate suggestions okay tool names rough draft script perplexity output stepbystep guide I think it's probably stepbystep guide right no I don't think so anyway I don't know where the chat she PT is coming from it's saying that I'm feing in chaty PT so instead of me yeah like I'm not going to worry too

### Updating the Database [1:12:09]

much about it looks like it's um hey tired of hey are you tired of chat GPT sounding like a robot you can transform it into a conversational champ with this humanizer prompt it's just 16 lines add it before any prompt and watch your attacks go from dull to Dynamic here's a quick example before using it and then after huge difference right this prom can at storytelling sprinkle in some humor and make your content truly shine want to try it out just comment chat GPT and I'll send it straight to your DMs the CH chpt is obviously the error here um I don't like I really don't like the way that it's laid out so we need to add some new lines so I'm just going to have rule um add new lines as back sln between sentences let's take this out and the script called action no FRS bre straightforward and serious maybe not serious be straightforward and don't use um what would you call convers ational champ just don't use poetic language let's say then over here I wanted to reduce the output Randomness to 0. 7 let's see if chat chpt tired of chat GPT sounding like a robot try this 16 line humanizer prompt just paste it before any prompt and watch your content transform into something more natural and engaging that's pretty good yeah that sounds a lot more reasonable cool all right so the last thing that we need if you think about it so we just have to update the record so how am I going easy for all of the entries that come in right which in our case is just going to be one item this time um I'm just going to go add entries and then instead of a pendro we're going to want to update the row the column that we're going to match is going to be ID and then if you think about it we got to go back and we have to find the data that we just added so added is going to be under add entries right so I'm going to go here go ID awesome so now for all the other information we don't actually want to update it so let's get rid of all this stuff the only thing we want to update is um scraped transcript and then new transcript which was over here scraped transcript is going to be um right over here uh nope sorry right over here then new transcript which I just deleted because I'm a silly man is going to be all the way up here and it's going to be this script right here okay all right let's give this a test beautiful if I go back here voila we now have the scraped and the new transcript beautiful Okay cool so now if you think about it logically all I have to do is I just have to go and I have to find a bunch of people in various domains and niches add them to the input at the very beginning of my flow which is all the way over here um feed them in over here under username then the results limit I'm probably just going to make it five per person and then in terms of how I'm going to do this the last thing is I'm just going to change a trigger and I'll just do um on a schedule so I'm just going to do once per day I'm going to trigger it at just do 6 a. m. feed that in as my input then obviously I need to like unpin everything and so on and so forth so that looks pretty good to me yeah I'm going to save this for you guys so if you guys want to try this out on your own doing the exact same thing with me for instance feel free um but this is more or less uh exactly the system that I wanted to build we ran into a couple stumbling blocks but I think we figured it out and yeah I also hope that you guys appreciated seeing what some of these detours and dead ends look like what some of my thought processes behind building nadn work flows look like um and so on and so forth so just to make sure this works we're just going to unpin right you know you don't want to be using pin data to actually run it um and then what I'm going to do is I'm actually just going to I'm going to run this whole thing including getting um actors getting actor information rather go two Max items go back here and then I'm just going to delete all of this so it can actually run a new and then once this is done I will download set blue print oh let's just change this so it says update entries I will download said blueprint and I will give it to all of you so first step is we need to obviously run the actor synchronously that's what's going on here we are pulling apify as we were before it's now starting a crawler as we see last time it took us about 20 seconds or so to get the information so it'll probably be taking someone similar post scraper finished so looks like we got the data it's now adding the entries to the sheet so we just added two for each we're downloading them for each we're also transcribing the video we are now filtering and generating suggestions now we're going to do two API calls to perplexity then we're going to do another two API calls to open AI where we're going to rewrite the scripts and then finally the last thing we're going to do is we're going to update the entries in our database if I go back here you'll see that we've now added two new rows we have both of the scraped transcripts and we have the new transcript up the top check out this new aex speech generator it's just as good as 11 last but it's free blah and we also have this down here um now make. com it's saying make. com for me just because I said hey I used make. com is awesome so obviously when you feed in your own data source in your own specifications you know because it's not going to just be an automation guy talking about automation stuff it's gonna be this new painting tool is enabled us to make amazing art or something or this new video AI tool is what we're using as part of our editing workflow right um we can combine all of that funnel it into my AI automation agency Channel um and then I'll have basically like non-stop content which I'm pretty excited about so hope you guys enjoyed and saw what the real build process behind systems like this look like um I'm not selling this particular system although I want you guys to know that you totally could as you see it took me less than an hour and a half to set this up start to finish um I think I you know I'm going to do like one cut in this video it's going to be like 30 seconds while I figure out a prompt or something so realistically it you know it legitimately uh we put it together in under an hour and 20 minutes uh this is something you could absolutely sell for a fair amount of money to any quickly

### Conclusion and Future Prospects [1:19:08]

growing Channel I don't know exactly how much and I don't really know what sort of maintenance fee but I would certainly uh have been willing assuming that I'm not as technically competent as I am to probably pay three four $5,000 just to have this problem solved for me never have to worry about it again um there might be some people out there that are willing to pay more honestly so wherever you are using this for yourself using this for a client considering building it and productizing and selling it as a service uh I hope that you have everything you need now to make it work please leave comments down below with suggestions for future videos I Source a lot of these ideas from you guys at this point and I really like making anend content that you know cils to your guys' interest and if you could do me a big solid like subscribe do all that fun stuff that bumps up to the top of the algo I'll catch you on the next video see you