i just built this AI system that 100% automates viral shorts like these while you sleep as you can see it's a pretty simple five-step system and we didn't have to write any code step one is we're grabbing the main story from our Google sheet and we're feeding it into an AI agent to create those prompts then step two is we're going to use Flux to create those images and they turn out super high quality step three is we're going to feed those images into Cling to turn them into videos and these also turn out super high quality then we're going to use 11 Labs to create one unique sound effect for each video step five we're going to use Creatmate to render all the video files and audio files together in one video and then we're going to be using Botato to auto post to Instagram Tik Tok and YouTube as always I'm giving away all the resources that you need for free so later in the video I'll talk about how you can access all those resources and how you can actually plug them in and what you need to set up on your end so you can have the system running in less than 10 minutes so let's do a live run and then we'll walk through looking at every single node in detail so you guys can see exactly how I built this system okay so before we test this workflow we have to go look at the Google sheet template that you guys will be able to access of course and this is what it looks like we basically have a category and then we're going to have four brands within that category and this is what the workflow is pulling from to start the whole process and then it's going to come back and throw that final URL back in the sheet so before we hit test workflow and process fast food giants we're going to take a look at these examples real quick so these three outputs came from our NN workflow let's take a look at the sneaker [Music] brands next up we have gaming brands [Music] that was funny and the Nintendo one i even saw like a bunch of mini like Mario and Luigi's down here finally we'll look at this tech giants one [Music] [Music] so as you can see they were super cinematic they had little people walking around you could immediately tell which brand it was as soon as you saw the video so we're going to go back into the workflow i'm going to hit test workflow and this should be pulling in the first item in the sheet that's marked as to-do so this one should be processing fast food chains and if we click into this node right here we can see this one did pull in fast food chains so this is pretty much what starts the process the schedule trigger before this one you could set to run three times a day four times a day once a day whatever times you want but let's just say right now we're running this once a day it would hit this Google sheet and it would basically look for any row that has the status column equaling to-do just so we know we're not processing rows that have already been created and then it's only going to pull back one item so that way you can fill the sheet up with 100 items but it's only going to process one at a time and as you can see right here that is fast food chains with these four fast food brands from there what we do is we're basically just setting brands into one array and this is just so we can split them back out later in the next node over here where we're splitting them out into four individual items and the reason we want to get them you know a little bit of data manipulation to get them into four unique items is because we want to process four images four videos and four audio files so running four items through the rest of the workflow is how we can accomplish that once we have those as four individual items we're going to go into this prompt generator agent which is going to do a few things so first thing I want to call out is that in the user message we're giving it the category and a brand and there's going to be four different ones so if you see if I click into this we basically just have the actual category so I went from here and I pulled the category in and then I said brand colon and then I pulled in the brand from here and like I said this is going to process four different runs so we have fast food chains brand McDonald's In-N-Out Taco Bell and Chick-fil-A so that's the user message but now we have the system message which is telling the AI agent how to behave and what its instructions are so I'm going to open up the system prompt not going to read the whole thing because you can download this template and you'll have access to it but we're just going to talk about highle what's going on so for the overview of the agent I wrote "You are a multimodal creative AI that transforms famous brands into tiny hyperrealistic visually satisfying digital resources you'll be given a well-known brand and your job is to output a highly detailed image prompt a cinematic video prompt and an immersive audio prompt and those are kind of the three elements that we need from this thing in order to pass those on the rest of the flow when we're generating the images generating the videos and generating that audio and the way we're able to actually get it to output these things and their individual fields over here is because we turn on this option called require specific output format and what that does is it adds one more option under the AI agent for us to attach a structured output parser which if we follow the line over here basically this says okay for each item you're going to output the brand's name an image prompt a video prompt and an audio prompt and that's exactly what we get back over here for our four different brands as you can see McDonald's In-N-Out Taco Bell and Chick-fil-A and then the final thing to call out is that we're using Open Router for the chat model and we're accessing OpenAI's GPT 4.1 chat model all right so we have all of the prompts that we need now we're going to start to create those elements so the first thing we're doing is making a request to FAL AI which is going to basically be similar to PI API if you've watched other videos I've done similar to Open Router honestly where we're basically just going to access one service that lets us through that service access a ton of different text to image models or a ton of different imagetovideo models and we're using FAL for both our images and for our videos so let me explain what that means so here is FAL AI as you can see and if I click on explore you can see that there's just a ton of different models to choose from so we have image to image we have image to video we have image to 3D we have text to image text to video all this kind of stuff and so what we did is we went to text to image and we're using this one right here which is Flux Pro V1.1 Ultra we're using this for our text to image generation so when you click into here what you can do is in this playground environment you can basically like enter in a prompts like this and you can hit run and this is how you can test out what prompts are going to give you what sort of result in foul with Flux Pro and as you can see down here it says that your request will cost 6 cents per image and this is the image that it got based on this prompt so what I would do is just test out a few of the image prompts that the AI agent spits out to us in NN and if I like it then I'll move forward and as you can see if I click on requests you can see all of these different images that we just have generated through this account here's a sneak peek at the Chick-fil-A and the In-N-Out and the Taco Bell stuff but we'll see the full video later anyways what we want to do is click on API and this is going to show us how we can connect to this model through FAL in NAD so what we have to do to call the API is we have to copy this URL we have to put in our credential we have to put in the content type put in a prompt all this kind of stuff so the first thing I do real quick is go to foul and go to API keys and then you're just going to have to create a new key and copy that and that's what we need to put in when we're in NAN so if we go back to the Flux API documentation we can see right here that in order to put in our API key it's going to be a header parameter and it's going to be the name is authorization and the value is going to be key space file API key so what we want to do right away is save this as a header off so in authorization click on generic and generic click on header and then in header off you're going to create a new one and then you're going to do the name is authorization and the value is going to be key space and then paste in your file API key you'll save that save it as foul because we need to use this credential in multiple other nodes in this workflow but once you have it saved once you're set the next thing you need to do is add a header which is going to be content type with application/json and then the most important is we're sending over two fields in the body request the first one's called prompt and this is just us dragging over the image prompt right here and that's actually going to send four requests to foul to make images as you can see we have the first one for um McDonald's the second one for In-N-Out the third one for Taco Bell and the fourth one for Chick-fil-A and then the last thing we have to do is make the aspect ratio 9 by6 because obviously we want to make shorts so what this is doing is it's hitting foul server it's hitting the flux model and it's basically saying okay we got your prompt and we got your aspect ratio we're going to work on those images right now so that's why we see the status is in Q we have information about the actual request but we don't have the request back yet so what we do is you know we played around in the playground we can see it took about 10 seconds so I just set a wait node for 12 seconds and then we move on to get the status of that image and we do this by accessing this next URL which basically lets us put in that request ID from that first one and check in on the status so as you can see we drag in the request ID from that previous node where we generated the images so right here put that in there and now we see that all of them are complete so we're good to go and because they're complete we do this little check just to make sure that all four are complete before we move on if even one of these was not complete yet it would basically just come over here wait three more seconds and then continuously check in until all four are complete but they're all complete so it moves on to this stage where we're actually going to download those files and we once again just have to use that request ID from earlier and now that we've done that we have four items back and if I click into this one this should be our image for McDonald's if I scroll down a little bit this should be our image for In-N-Out there we go and the other ones would also be the other images but don't want to spoil it yet so that's step two we have all of the images created for us and one thing to note is in this node where we're checking the status we have to once again put in our file credential and same thing for this one where we're downloading the image we have to put in our file credential so that's why it's super important to just go ahead and save that right away okay but now we have those four images and now what we want to do is we want to access FA's cling model in order to turn those images into videos and when we're turning those images into videos we have to send over to Clling both an image and a video prompt so I'm going to open up this node and we're going to dive into what's going on here so once again we're using foul but this time we're using a video model called cling so I'm going to click into cling right here and same thing we have like a playground where we can talk about what's what we want and we can give it an image and then we can hit run and it will generate one over here but what we want to do is go to the API and this again is going to tell us how we can use this in NAD so this one's very similar to the previous one we have an endpoint we need to hit we're going to use that same file credential we're going to input a prompt an image URL but then of course we also want to put in our aspect ratio so I'm going to head back over to Nitn i put in the endpoint i put the method as post i connect to the file credential we already have we put content type application/json and now in the prompt we're sending over three things or sorry not prompt in the body request we're sending over three things the prompt is going to be that video prompt that that first AI agent made for us so all you would have to do is scroll over here to the prompt generator and then drag in the video prompt and once again this is going to be doing four separate runs each for the different brand and then what we want to do is drag in the image URL for it to turn that image into a video so right here the node was get images and it gives us a URL i would just drag that in right there and once again this is processing four different images as you can see and finally we're telling it okay this aspect ratio has to be 9x6 it's going to be vertical it's going to be a short and once again it takes all those elements and says okay cool we're working on that now these are all in Q they're all processing they're not ready yet so I've seen that those videos take anywhere from 5 to 6 minutes sometimes a little less so right here we wait for 5 minutes and then what we do is once again we check the video status so we're doing the exact same thing from earlier where we check the video status we'd be able to find this endpoint to use if we came back in here to the documentation for cling and we just scroll down a little bit right here I scrolled past it right here is fetch request status so we're using this endpoint and we're going to put the request ID in there and then to get the result we're using this endpoint and putting the request ID in there once again it's good to follow along with this but if you download this workflow for free you'll be able to just everything will be set up for you already all you have to do is plug in your own file credential but as you can see they all four have come back now as complete and we're doing the exact same check in this thing basically to see if even one of these is not complete we're going to loop back we're going to wait and this is a method called polling but anyways um these all four were done in this case so we're good we move on we download all those videos by using that request ID once again and our file credential of course and now we have four items four videos let's take a look at this last one this one should be Chick-fil-A I believe so it looks like a chicken sandwich and there's people walking in the door super cool let's just take a look at one more down here which should be for Taco Bell and just looks like a giant crunchy taco with people walking in awesome all right so we have four images we turn those images into videos and now we need to get a sound effect for each of those videos so what we're doing now is we're making an HTTP request to 11 Labs so if I go to 11 Labs and I come down here and I go to my API key first of all that's how you can create a new key and copy that in but then what I'm going to do is click on API documentation and what I want to do is come down here to sound effects and create sound effect so this is going to give us this curl command once again we could copy and paste that into N we could copy in the endpoint but this is already going to be set up for you so all you have to do is create your 11 Labs credential and according to the documentation this is going to be a header off so same thing that we did for foul but when you create this one instead of saying authorization we're just going to do xi- API- key and then you don't have to put keyspace API key you can just paste in your API key that's the way that 11 Labs accepts credentials cool so now that's saved and what we're going to do is send over a body request with just two things so the first one is text and this is basically just going to be the sound prompt that the original agent gave us all the way down here so I would just drag in this audio prompt right into that parameter right there and then finally we're just telling 11 Labs make each audio source 5 seconds long because all the videos are 5 seconds long all right so as you can see we have these four different audio files let's just give one a [Music] listen cool apparently that suits McDonald's but we'll see when we have it all rendered together okay so now this part's pretty important we have that audio file but what we have to do is make it publicly accessible so that create a mate over here when we want to render it all together can actually access it so we upload it to drive which basically just looks like you know we sign in with Google and if you're having trouble with Google credentials I'll tag a video up here where I walked through how to do that we're uploading a file and we're inputting a field that's binary called data so as you can see over here there's four different items and they're all called data so that's what's being pulled in basically just setting the file name of this um MP3 file so I called it fast food chains.mpp3 and the reason I got to fast food chains is I went down to the sheet that triggered this whole workflow and I just pulled in the category so we're getting four audio files in our drive called fast food chains.mpp3 and just choose the actual folder you want that to go to and so what we get back over here from Google Drive is basically like an ID for that file as well as some other little information but all we really care about is that ID so there's four different IDs because there's four different audio files and then we move into this next node to actually make it public because now it's in Google Drive but it's not public yet so this next operation is to share a file we're going to share it by ID and that's exactly where we just dragged in this ID that we got from that previous node and we're basically adding a permission that anyone can be a reader of this file so now it's like completely public cool so we have four videos we have four audio files that are public now we can basically pull everything together so a little bit more data manipulation here we're just setting all of the four video files and we're setting all of the four audio files and if I switch to JSON you can see this is exactly what it looks like and we just want to get it like this so we can aggregate everything back together into one clean item so the next step that's exactly what I do i use an aggregate node to just pull all of the incoming items into one field so now we have video one and audio one video two and audio 2 video 3 and audio 3 and video four and audio 4 because that's exactly what we need to do for our next node which is create a mate so let me hop over to Creatmate real quick to show you guys what that looks like this is the template that we're using which you guys will be able to access don't worry it's called Famous Brands but all it is is it's a template that's going to be 20 seconds long and there's going to be five audio files or sorry four 5-second long audio files and four 5-second long video files so all we're doing is we're going to send a request to create a mate and say "Hey I want to access this template and here are the four video files and here are the four audio files." So in order to do that I would click on use template i would click on API integration and then you can see this is what would pop up and I can click on copy to clipboard and import this curl into HTTP request in N and we'd be set up already with like our API key and everything but when you have the template in your own creatate you'll just copy this template ID right here so you'll copy that come into N and open up this node and you'll just put in your template ID right here instead of mine and then you just need to swap out your API key right here which you can get to in creative mate by clicking up here project settings and then you can click right here to reveal and copy your API key and then you just replace it right here bearer space your API key but once that's set up it's really simple all we're doing is we're sending over the four videos and the four audios and because we aggregated them all together right here it was super easy i basically just went video one video two video three video four and then same thing with those audio files and what this does is pretty much the exact same thing that that foul was doing when we made a request it says "Okay cool this is the template you want here are all the elements um I'm going to work on that right now." So here is my status it's planned it's processing and if we go into create a mate and go to our API log we can see that this one took 61.59 seconds so we go ahead and wait for 60 seconds looks like it just finished up in time but keep in mind we have this little polling method just in case it doesn't go through that it would come back and try again but looks like it worked so it went ahead and downloaded that video based on the URL that it provided us and now we have the full video right here in our edit end so let's take a quick look [Music] wow that one turned out really cool i liked the In and Out there were cars moving by and there were like palm trees you could tell it was like West Coast anyways that is the final video and now we're moving on to this final step where we're actually just auto uploading everything and updating that Google sheet so the first thing we're doing is we're hitting that Google sheet and we're updating a row so we're linking back to that same sheet that we were looking at earlier and when we're doing an update row operation in Nad we basically have to tell Nadn which row to update and we do that by mapping on a certain column so the column we're mapping on the column we're matching on here is called category which as you can see is like the first one over here so we're matching on category and all I had to do was drag in the category that triggered this whole process all the way down here so now it sees whenever so now it basically says "Okay I'm going to update the row called fast food chains." And for the actual updates I'm making I'm just going to change the status to done and I'm going to upload that final URL so if I come back in here we can see fast food chains has been marked as done and we now have this final URL which is the video okay cool so now that we have that all figured out we want to automatically upload to Instagram Tik Tok and YouTube and the way we want to do that is through a platform called Blot which I'll leave a link for in the description you can also get 30% off if you use code Nate30 for 6 months you get 30% off so anyways at a high level how this works is we have a URL that points to our video but in order to upload it through Blotato to Instagram Tik Tok and YouTube we have to upload that URL to Blot and it's going to give us a potato URL same video just a new URL and then we can publish it so that's what we're doing in this first HTTP request is we're accessing Blot's servers we have ourselves credentialed of course with an API key so same thing we're doing a header off blot- API- key and we have to get our API key from Botato so if you go into Blot and you come into your settings you can basically see okay here's my account here are the nine different things I can log into and auto post to and then down here we have API access copy API key so back in Nitn you'd basically just you know paste your API key right in there and now you have it saved to potato and it's always just good to save your credentials i didn't do it in the create a mate step but it's always good to save them so when you need them later you can use them easily and you'll have to use these three credentials later for Instagram Tik Tok and YouTube but anyways what we're doing is sending over a JSON body and it's just one thing it's a field called URL and we're just going to drag in that URL that we just got from Google Sheets and from Creatmate right here and now as you can see it gives us back a URL which is the exact same video it's just now a blot database URL now all we have to do is link to these three nodes which are pretty much the same thing just a few different configurations so let's take a quick look at the Instagram one first so we're making a post request to Blot to actually post something we have our Blot header off in there once again and now for the body what we're doing is we're using this JSON which is saying okay I want to post to Instagram the text or the title of this post is going to be um the category reimagined so it's going to come through as fast food chains reimagined what is the actual media URL this is where we need that Bot URL in order for it to actually post it and then finally we need our Instagram account ID which all you have to do is go to Bot you'll sign into Instagram and then right here you can see my Instagram account you'll click copy account ID come back and edit in and then you'll just paste that account ID right there and now you're good to post to Instagram tik Tok is pretty much the same thing just a few differences everything up here is the same the only difference is the actual JSON body this one has a few different like settings you can tweak with so is AI generated i put that as true but what we're doing here is text so same exact thing fast food chains reimagined platform is Tik Tok and then the media URL once again and then for our account ID you'd log in with Tik Tok you would come here click copy account ID and then you would just paste that right in there and you'd be good to go finally with YouTube like I said all of this is the same but then the JSON body is a little bit different youtube asks for a title and text and I'm not exactly sure which one really is which so I put the exact same thing in both which is fast food chains reimagined i have this going to unlisted should notify subscribers false and um same thing with the media URL and then once again of course with the account ID you'll log in with YouTube copy the account ID right there and then you will just paste that right there and you'll be good to go and then one thing with potato that I wanted to mention is when you come into your account you can go over here to publish posts and then you can see exactly which ones have been posted and you can click on the links right here and it will take you to that post on your Instagram or your Tik Tok or Pinterest or LinkedIn whatever you want to post to but one thing I wanted to mention is if you go to failed posts you can see some that have failed and I've gotten a few errors with Tik Tok which basically have said like the creator has made too many posts via OpenAI so I was doing too much when I was testing and then also I had gotten multiple errors with YouTube and I think this is just because it says you know I've exceeded my quota so if you're seeing some errors or if you're not seeing posts pop up you can go to your failed post section and you can see why those aren't working okay so that's basically this entire workflow as you can see it's not too complicated and I really want you guys to download this workflow test it out play with the prompts and customize it for you know creating your own unique type of shorts but what we're going to do now is talk about how can you access all the resources what resources actually are involved and how can you set this up for yourself in under 10 minutes so first thing you need to do is go to my free school community the link for that will be down in the description and you're just going to click on YouTube resources or you can search for the title of this video right up here in the search bar and when you find this post so say this is the post associated with this video you'll see a JSON file right here and you're going to download that and once you download that JSON file you can open up a new workflow in Naden come up here and click on import from file and paste in that JSON and then this exact workflow will just pop up right there and there'll also be a little setup guide right here which will have all the links you need and say like here's what you need to plug in get your API key from here plug it in there and it'll be pretty simple but I'm also going to just kind of walk through that real quick in this video so I'm just going to go kind of chronologically through the workflow and tell you what you need to set up so the first thing would be the schedule trigger how often do you want this to run the second thing is the Google sheet so you need to get this template which I will have also in that school post you know you'll have the template you'll have a create a mate file all this kind of stuff this is the template you'll have and you'll need to come in here and make a copy of it so a file make a copy and then it will be in your own Google suite and then you'll just have to come in here and make sure that you actually choose the right sheet and link it to your own Google environment otherwise it's going to be trying to access a sheet that you don't have access to and stuff like that so set up that you're also going to have to set this sheet up right over here by the way so pretty much the exact same thing but then moving along the next thing you have to do is connect some sort of chat model so we're using Open Router and I'm using GPT4.1 if you don't want to use Open Router you don't have to but you'll have to connect a different one and if you're doing Open Router all you have to do is go to Open Router connect your API key in here set up your Open Router account and then you'll be good to go okay next thing you're going to have to do is set up your file credential for all six of these nodes and if you do it once it's so much easier for the rest of them we kind of talked about this earlier but you're going to go to file you're going to grab your API key you're going to come into here and you're going to add a new header off and it's going to be authorization the value is going to be key space your file API key so you'll paste that all in there and then you'll have this file credential saved as a header off which you can then link in this node this node this node this node and this node then you're going to have to go get your 11 Labs API key so you'll go to 11 Labs like I said all the links will be over here in the setup guide but you'll go to 11 Labs and get your API key and then you're going to make it a new header off XI- API- key and then you'll paste in that 11 Labs API key right there okay then we get to create a mate down here so the first thing that you need to do is when you're in that school post right here you'll see like a PDF and it will look like this so this is the source code that you need for Creatmate so let's say you come in here and you copy this source code you go to create a mate and you go to templates and let's just say that you you know open up a new blank template and there's nothing in here and you'll come up here and click on the source editor you'll then just basically paste in that source code that I gave you and you'll have this template where you have a 20 second long video with four audio files and four video files and then the easiest way to set this up for you would be to click on use template API integration and copy your template ID right here go back into edit in and then when you open up this node that's already there just replace the template ID with the one that you got from your own creatate and then all you need to do is replace my API key right here with your API key so you'll go back into creatate you'll come out here and go to project settings and you can click copy or you can click to reveal your API key and then copy it and then you'll just replace it you know right in there cool and then the last thing you're going to have to do is blot so you'll click on the link in the description or over here to go to Botato you can sign up and get 30% off using code Nate30 for six months and then you'll go get your API key from Botato you'll save that as a header off where you'll do potato-i- put in your API key then you're pretty much good to go the last part would just be in Botato if you go to your settings choose which ones you want to post to and then once you log in you'll have to link all of your own account IDs you know in those individual requests like I kind of talked about earlier but that's going to be basically it that should all take you less than 10 minutes and then you can start filling in that Google sheet template with with categories and bam you're going to be auto posting to different socials on autopilot the final thing I wanted to talk about here was the actual cost breakdown of running the system so there's a couple different elements here the first one would be the chat model so we're using OpenAI through Open Router GBT4.1 and it's super cheap i mean um in the grand scheme of things it's less than a cent per run so that's not really even an issue now the video generation in this template is definitely a little more on the pricier side compared to some other workflows that we've done um but I think that that's the you know trade-off you make this is really really high quality stuff so first of all image generation we're using Flux Pro through FAL this is going to be 6 cents per image so four images per run so 24 cents per run you could always go down to different image generators like a lower flux dev would be like one and a half cents per image you could try that or you could scale up if you want anyways then for video generation same thing you could scale down or scale up i really like the way that these outputs were were coming out so with Clingv 1.6 Pro it's going to be 46ish cents for a 5second video so it's a little less than $2 per run per short just for the video generation aspect and then we have some weirder ones because they're more of a monthly not a pay as you go model so we've got sound effects from 11 Labs which if you do a $5 a month starter plan from 11 Labs that really should be more than enough i've never really gone over that then we have Creatmate which will be you know you have a free trial so you can play around for a bit but then on top of that it'll be 41 bucks a month which will give you 2,000 credits so real quick if I go into Creatimate and look at our runs we can see each of these videos is taking around 12.5 credits so this math isn't perfect but that should get you around 200 runs for the month and then of course an optional one is Botato which is going to be 30 bucks unless you do code Nate30 you can get 30% off um and this is going to help you automatically post to those three socials but also there's six other socials that you can automatically post to cost per run roughly is going to be a little over two bucks if you're taking into account the image and the video generation that should be cling with an with a K um and then we have these other optional sort of monthly subscriptions which you know you kind of do need the sound effects and the creatate but these are what the costs look like for that all right so that's basically the whole breakdown hopefully this is cool to see that it's not even that complicated the only really complications were doing this polling with multiple items running through and maybe setting up the Creative Mate template but luckily I have this template for you so all of that is good to go all you have to do is plug and play and play with some prompts so if you enjoyed this video and you learned something new or you enjoy this style of teaching and this style of video definitely give this video a like it helps me out a ton and also definitely consider checking out the paid community the link for that's also down in the description we've got a little over 1.8,000 members in here who are all building every day with Naden and learning AI automations it's really cool to be in a community where everyone is as obsessed with this kind of stuff as you are so it's really helpful to be in there we've also got two main courses one's called Agent Zero the foundations of like LLMs and vector databases and stuff and then the second one is called 10 hours and 10 seconds where you learn how to identify design and build time-saving automations on top of that we also do five live calls per week where I'm in there and we do guest speakers we do Q&As's we do coffee chats and tech support as well so you can always get questions answered and you can also make sure you're connecting with other people in the space but anyways that's going to be it for this video really appreciate you guys making it to the end as always and I'll see you in the next one thanks everyone