| Source | https://read.readwise.io/read/01kvxr0fk1hbbm5jfcfw35faj0 |
|---|---|
| Readwise URL | https://read.readwise.io/read/01kvxr0fk1hbbm5jfcfw35faj0 |
| Readwise ID | 01kvxr0fk1hbbm5jfcfw35faj0 |
| Date | 2026-06-10 |
| Author | Your Average Tech Bro |
| Category | video |
| Site | YouTube |
I built and scaled an app to five granda month of revenue only using AIgenerated social media content, and inthis video, I’m going to do a completeend-to-end guide on how exactly I dothat entire process to market my apps.I’ll show you all the sauce, all themodels, the exact workflow that we use.I’m truly going to do a tell-all videohere. So, hopefully, this is the onlyvideo you’re going to need to figure outhow to make AI-generated social mediamarketing content that actually gets alot of views to market your app onsocial media. But before we do that,you already know I gotta show you theproof that I’m not lying to you. So,let’s get into that. So, first thingsfirst, this is the tool that I marketedwith AI-generated social media marketingcontent. It is called Your Beat, and itis an AI content script writer. It’s
essentially a tool that lets people gothrough and search through our viralcontent database, and then they can thenRemix and write scripts better with AIor very opinionated AI to help you writebetter scripts, optimize your videos toget a lot of views on the internet. Andwe also have a feature that lets you spyon different accounts. So, let’ssay you want to spy on competitors andget alerted whenever one of their videosis going viral and doing well. And thisis the account that we use to market theapp to over 7,000 followers, and everysingle piece of content posted on thispage is AI. This person is not real,does not exist, and we’ll show you whatthe videos are like. This is what 99.99%of the content on this page is. It’swhat we call a hook and demo format.
Essentially, we have an AI-generatedperson here with some type of juicy hookin the beginning, right here. Then we puta demo of the product, showcase it inaction to kind of sell the user on thatproduct. And last but not least, thelast most important piece of proof,let’s go to the revenue. So, right now,You can see that our total gross volumeof the last four weeks is over $7,000,and our MRR is currently hovering aroundmid-6,000s. But, I’ll be honest, rightnow we’re no longer exclusively doing AIUGC, but that’s still a huge chunk ofour marketing budget. But, pretty muchup until the five grand a monthmark, we were exclusively doing only AIUGC, AI social media content. And now
let us get into the actual sauce, theactual guide of how to do this all. So,the first step of any type of socialmedia marketing is whetherit’s AI or human or anything. You needto warm up your accounts. These aremulti-trillion-dollar companies thathave huge teams dedicated to security,and fraud, and bot detection, and spamdetection. And your goal right now iswhenever you make a new account, youjust need to prove to the platform thatyou’re a real person, you’re not a bot,you’re not a spammer. And the best wayto do that is by warming up yourAccount. And what does that really mean?
That really means just scrolling and onthe account and using it like a normalperson because you are a real person,you’re not a bot. Scroll, watch somevideos, like, comment, share, all thatgood stuff. Just prove to them thatyou’re a normal person.
We typically, atleast internally, myself and myco-founder, we typically warm up ouraccounts for maybe 3, 4, 5 days until westart beginning to post content. Butit’s an imperfect science. We know somepeople that do really well without anytype of warm-up. We know other peoplethat do well only when they do warm up.Doesn’t matter. Just for us, I think forlike our little tinfoil hat conspiracytheory, we feel a little bit better andmore confident when we do do someaccount warm-up ourselves. So then,after you do the warm-up, now it’s timeto actually go off and start making your
AI-generated social media marketingcontent. And before I show you theActual end-to-end workflow of how wemake our content, let me just give yousome tips on what type of content doeswell with AI-generated content. In ouropinion, videos where the person is nottalking are the ones where AI can reallyexcel. Because, at least right now, at thetime of filming, which is in May of2026, AI and talking videos, like itthey’re a little bit easier to tellwhether that thing is AI or not. And atleast right now, there’s a visceralanti-AI sentiment on the internet.
People are hate-watching AI-generatedcontent. But interestingly, I thinkpeople are okay with watchingAI-generated content when it’s soclearly AI. I mean, right now, at thetime of filming, the currenttrending account, an account where fruitare talking to each other in like asitcom parody show, and that is 100%purely AI-generated. But people stillwatch it because they’re, they know 1billion percent this person is notTrying to lie to them and fool them thatit’s not AI when it clearly is. Theyjust own up to it. It seems like rightnow most people hate that AI where it’sso clear that it’s fake and they’retrying to come off as real, which iskind of what we’re doing, but we can doit in a much better way. Because, onceagain, we did make a talking head videohere, which I’ll play back for you rightnow.
If you’re still scripting your videosfrom scratch, you are actually fallingbehind.>> It’s not bad, but definitely it’s not asgood as these videos where it’slegitimately almost impossible to tellthat this is AI generated, right? Justcuz they’re not talking, and I thinkit’s the talking portion that is reallytough to pull off with AI, at leastright now. I think in due time we’ll getthere, but right now that’s that’s kindof the downfall with AI generatedcontent. This is a format we like toCall the hook and demo format where wehook it, hook the user with somethingjuicy text, hook in the beginning, thenwe plug in a demo of whatever product orapp that you are building to solvewhatever problem that the hook poses.
And just for some other proof, this isanother separate account that I’mrunning for actually, uh, someone thatwanted to pay us to run AI social mediamarketing for them. We’re running it forthem, and we made the same types ofvideos, hook and demo videos, and youcan see that they are gaining a prettygood amount of views. 6K views, 16Kviews. And here are some other, likecontent that we just flagged and savedas some good content inspiration ofcontent that has no talking involvedthat can probably be recreated with AIin an effective way. Like this righthere. This is literally a still photo.There’s no video. It’s a still photowith text on the screen. This is kind ofwhat we call text on screen content.
It’s a little bit less converting becauseit’s harder to plug a product inside ofit, but this is so simple and it hasgotten a ton of views. Like, look, it hasgotten, uh, I guess you don’t really know,but 167,000 likes, probably multiplemillions of views. Another one, it’sjust a simple photo with some textoverlay on top of it. This one is anactual video, I believe. It’s just notplaying right now, but it’s a video ofjust a dude laying in the rain with textoverlaid on top of it. So, this isbasically another really good formatthat you can probably recreate with AI.So, in general, AI-generated content,our personal strategy is I just don’tthink that talking content is quitethere yet. It only really, really worksright now when it’s like no talkinginvolved, just the image, just the videowith some type of text on the screen.
So, now let me show you exactly how wegenerate content ourselves. So, now thatwe went over how to warm up yourAccount, setting it all up, and talkedabout some of the formats that AI reallyexcels at. Now, I’m going to kind of walkyou through the actual sauce, all themodels, the entire workflow of how weactually make high-performing AIgenerated social media content to marketour tools on the internet. So, we’regoing to walk you through the exact toolthat we use, which we actually areoffering as a product that people dowant to use. You can find it at
socialq.ai. I’ll include a link righthere, as well as a link in thedescription of this video. So, we dosell this as a product, as a tool, aswell as a little bit of consultingservice with our social media marketingteam, myself and my co-founder to helpyou, you know, really do well and getexpert advice on how to make goodcontent to market your app. But, youliterally do not have to use our toolbecause you can just vibe code thisentire tool yourself if you really want.
To, because we’re going to show youstep-by-step all the sauce, all themodels, everything that we do to makethis type of content yourself. Allright, so first off, for us, we create abrand new run. So, our approach tomaking AI social media marketing contentis rather than trying to create a brandnew video from scratch. Our whole ideais remix content that already works.
Find existing videos on the internet,hook and demo videos, text on screenvideos on the internet that has gotten alot of views, and then we’ll just remixit ourselves. So, for example, we canuse this post as an example. I needsomeone to explain why I’m finding thisafter 3 years of making content. So,this is an example post of a post that’sdone really well that we can then justcopy and use ourselves. But, actuallywithin our tool social queue, we
actually have a viral content databasefeature where we’re literally sourcingall of this content ourselves and theHelp of our team to find other reallyhigh-performing viral marketing contentthat follows this hook and demo format right now. So, you can see right here, Ijust typed show me viral hook and demovideo, and you can see it shows it right here. Here are a ton of videos of otherhook and demo videos, just endlessendless, endless amounts, as well asdedicated accounts that do hook and demo really well as well. You get the gist.
So, yes, this is an advantage of usingSocial Queue. We find all these high-performing hook and demo videos foryou. But, for yourself, once again, justfind something that works. So, we’lljust use this video as an example,right? So, the way that we work withinour tool, Social Queue, you come in, youenter in a URL, and then when you enterin the URL, we scrape it, and then wesource all the information here. So, nowthat we have this URL, right? Let’s addour source video. I’m going to pastethat URL, and we’re going to begin.
Scraping it and analyzing it. So, wedownload it, run a whole bunch ofanalysis on it, extract the hook fromit, and then we return the entire hookclip right here as a reference. And thisis the hook text. I need someone toexplain why I’m finding this after 3years of making content. That soundsgood. All right, so now this step isdone. This is just to get the originalvideo to remix and get the really goodperforming hook, and then we can remixit for another product later on down theline. So, we have this video, let’s moveon to the next step. And the next stepis to generate a UGC avatar, an image ofan avatar. Essentially, what we’re goingto do is we are going to take the very
first frame of this video. So, thisimage of this person right here, thisAI-generated person we created. Andthen, when we generate our UGC avatar,we’re essentially going to upload areference image. In this case, we’lljust use this photo of myself. I’ll addThis as a reference image, and thenwe’re going to face swap myself orwhoever photo, whatever photo you uploadhere, we’re going to face swap thisphoto to recreate this first frame ofthe video, to recreate this hook, thevery first frame of this hook. We’re notgoing to generate the video yet, we’regoing to generate just a still image.
I’ll explain to you later on. So, we’regoing to kick this off, we’ll generatethree different variations that we canpick from. And the model that we’reusing underneath the hood here is NanoBanana offered by Google. We found theNano Banana to be pretty effective sofar. We have explored other models likeOpen AI’s image generation two, but theone downside of the image gen model twofor Open AI is you cannot provide areference photo of somebody that has anactual human face involved. I don’tthink they allow you to generate newimages of somebody when you provide theface to them. They don’t allow otherHuman faces as reference photos, whereasNano Banana does allow you to do that.
So, that’s why we’re using Nano Banana,because they are an image modelprovider, an image generation model thatlets us provide a reference image likeour human face right there.
What an old picture of me. I don’t evenlook like that. My hair is totallydifferent. But, from that referenceimage, we can then also pass in thereference image of the original faceright here and then tell it to do almostlike a little face swap. So, as you cansee right here, it is generating acouple of different variations. We havethis one, and then we also have thisone, and then this one. So, these areall not all too different from eachother. Let’s just move forward with thisone that I selected. So, then right now,after we generated the image, what wedid was we generated the first frame ofthe image. Now, the reason why we’redoing that is because in our videoGeneration step, right now we’re justgenerating the hook video. In our hookvideo generation step, the model we’reusing is Kling’s latest video generationmodel. I think it’s Kling video genmodel. What’s the latest version? It isI think it’s Kling 3. Yes, Kling 3.0.
With Kling 3.0, you can actually pass ina static image as the first frame ofthat video. Then, when you give it aprompt of the video, it’ll then takethat first frame and then create therest of your video using that as thefirst frame. So, as you can see, wealready generated a video generationprompt, and then we did that during thisfirst step where when we dropped in theentire video, we ran a bunch of processon it to get the timestamps for what thehook of that video is. So, not the demopart, just focusing on the hook. And aspart of that, we extracted the actualhook text, and then additionally, wealso have a prompt to like almostgenerate a prompt to recreate the hook.
Video that we can feed into anothervideo generation model, and that’s wherethis prompt comes in. A close-upauthentic UGC smartphone video of aperson lying in bed, head on a pillow,warm dim bedroom lighting, soft shadows,blah blah blah blah. Cameras heldnaturally with slight hand-heldmovement, no zooms. Person looksdirectly at the camera with a warmgenuine smile, creating an intimate cozyatmosphere. So, that is the prompt togenerate to recreate this hook of thevideo. Then, when we press generate hookvideo, we’re also going to then pass inthe selected first frame image that wegenerated with Nano Banana, and thenwe’re going to pass it into our clean[snorts] 3.0 model, and it will generatethat hook video, and it’s going to take
a little bit because video gen is reallyslow and taxing, so we’re going to waitfor this to finish generating. Okay, soit seems like the video finishedgenerating. Let’s play it back and takeA look at it. All right, so all right. Imean, it doesn’t really look like me. Itdidn’t get the bottom half of my facecorrect, but at the end of the day, ahook video is a hook video. I think thathonestly, I would have no complaintsusing this video. Maybe I would modifythe prompt to just say keep the handcovering the mouth the entire time sothat the bottom half of the face doesn’tshow, but you know, this is just a proofof concept. And for now, I think thevideo is good enough to use as a hook.
So, then we go off and hit the nextbutton. So, now we have the hook video.Now, we have to stitch it all together witha hook text and add a demo video toshow right afterwards. So, in thisscenario, just for the sake of this, Ialready have this video right here. Thisisn’t really like a proper product demo,right? This is just a video of me likesaying hello, all that stuff. Just asample one. So, we’ll just use that as ademo video for now, but in reality, weWould then attach an actual product demoright here. So, nothing fancy here. Andthen, we’ll take this original hook andwe want to modify it for a differentproduct. So, let me get that productdescription. So, now I’m going toprovide some remix instructions for this
fictional dating app called Wingmates,that like helps you analyze your datingprofile and gives you critiques on yourpage, all that stuff. Then, we’re goingto remix this original hook from theoriginal source video to match theproduct description of Wingmates. Andalso, for all of these text generationremixing stuff, we’re using ClaudeSonnet 4 6 underneath the hood. Andthen, we’re passing in our video righthere. Our demo video, in this case, it’snot a real demo video. It’s just asample demo video of myself, like wavingto the camera. And then, we are, for themusic, we’re just going to use the samemusic that was in the original video.
And then, now we have our entire video.
Here. You can preview it by pressingthis play button, but in this case weare just going to export the finalvideo. And then exporting the finalvideo, it will just stitch together thehook with the demo with the text overlayon screen with whatever music is goingon in the background. And for that, weare using FFmpeg, which is kind of thego-to standard for any type of videoediting video operations doneprogrammatically. So now you can see thewhole video is exported out here, and wecan preview it.
And then here’s the fake demo video. Soyou can see that it’s all done end toend. And then from there, we would justpost it onto social media. So that’sreally the entire process of how we arepersonally doing AI-generated socialmedia marketing content. We are reallyfocusing in on just a hook and demoformat, as well as a text on screenformat as well. But on our road map, wehave so many other formats that we wantTo add into this product as well, likefor example, carousel posts do superwell on the internet, and we wantto do some more carousels as well. Andthat’s just something we’re adding bit
by bit as time goes on. So just as aquick summary, the whole flow end to endof how this all works, we generate thefirst frame of the video using NanoBanana because you can pass in areference face as an image to do like aface swap operation. Then from thatgenerated first frame of the video, wethen generate the actual hook video,Clean 3.0 right now at the time offilming, May 2026. And that’s how wegenerate the hook video, which is thepart that’s AI generated. And then interms of the actual demo part, the demovideo portion is the part that you still
have to manually film yourself. And thenonce you have all that, we stitch ittogether into one final video. And onceagain, we don’t like to make videos fromscratch. We just like to see what’sWorking out there on the internet andthen remix that video ourselves. Sothat’s how what we do it, and you cantotally copy this entire workflow ofNano Banana plus Sonic for 6 for anytext remixing and then Nano Banana forimage generating and then Clean 3.0 forthe video generation portion itself tomake your AI social media marketingcontent. But also, shameless plug, ifyou want to check it out and you want to
just use a tool that will do this allfor you, you also get access to myselfand my co-founder as like social mediamarketing consultants. We’re there foryou all the time answering any questionsyou have through messages, through callsto help you make good social mediamarketing content, as well as gettingaccess to our library of viral hooks anddemo videos that other apps andbusinesses have used to market theirtool onto social media. You can use ourproduct at socialq.ai. Sign up there. Butonce again, no pressure, just vibe code.
Your entire solution to recreate thisyourself if you just want to do ityourself and you’re built likethat, you know? Anyway, that is anentire end-to-end tutorial of how we arecurrently doing AI social mediamarketing on the internet right now. I’msure things will change, strategies willchange, new formats will come up in thefuture. In that case, we’ll probablyjust make more content in the future totalk about it then. And then we’ll doanother recap of this video then. Butthat is all I got for today’s video. Ihope you enjoyed. Leave any questions orcomments in the comment section downbelow. I’ll do my best to answer as manyof them as I possibly can. But that’s itfor today. Thanks so much for watchingand I’ll see you in the next one.